PDA

View Full Version : Mac Pro



TZ
08-08-2006, 05:57 AM
Thought I would throw some stuff into this thread on the new Mac Pro.
www.apple.com/macpro/ (http://www.apple.com/macpro/)

Dual-Core Intel Xeon Processor
Ushering in a new era of outstanding performance and completing the Mac transition to Intel processors at the same time the new Mac Pro introduces the 64-bit Dual-Core Intel Xeon Woodcrest processor to the Mac lineup. Running at up to 3GHz, its a state-of-the-art processor that from day one makes the Mac Pro one of the fastest desktop computers on the planet.

Quad Core. Up to 3GHz.
Every Mac Pro in the lineup features two of the newest Dual-Core Intel Xeon processors. Two dual-cores. One powerful quad workstation. And you get to decide how fast it flies: 2GHz, 2.66GHz, or 3GHz. And at 3GHz, the Mac Pro runs up to 2x faster than the Power Mac G5 Quad.(1)

OEM Intel Xeon 5160 Woodcrest 1333MHz FSB 2 x 2MB L2 Cache Socket 771 Processor - $917 (http://www.newegg.com/Product/Product.asp?Item=N82E16819117100)

The Mac Pro ships with Mac OS X 10.4.7
The value of the computer model machine identifier string is MacPro1,1.
The Xeon processors are socketed, using dual 771 pin LGA sockets.
This means that the processors may be upgradeable
Airport antennae are located on the bottom of the machine, covered by plastic.

The Mac Pro is the first Intel-based Mac to have an Intel-native version of Mac OS X 10.4.x Server available as a custom option.

And, because it is socket cpu, and uses a bus multiplier, the future could be "interesting." Woodcrest Xeon cpu with the Clovertown Xeon cpu expected to be available next quarter.
Mac Pro Developer Note (http://developer.apple.com/documentation/HardwareDrivers/Conceptual/Mac_Pro_0608/index.html) (preliminary)
Mac Pro Schematics (http://developer.apple.com/documentation/HardwareDrivers/Conceptual/Mac_Pro_0608/Articles/M43_0906_arch.html)
Mac Pro Technical Specifications (http://support.apple.com/specs/macpro/Mac_Pro.html)

Mac Pro Do It Yourself Manuals
http://www.apple.com/support/manuals/

Mac Pro: How to reset the SMC (http://docs.info.apple.com/article.html?artnum=304123)

ricks
08-08-2006, 07:45 AM
RAM Is Good The new Mac Pro supports up to 16GB of RAM (there are now 4GB DIMMs available for up to 32GB memory) using two memory riser cards. Each card provides four DIMM slots for 667MHz fully buffered ECC RAM.

TZ
08-08-2006, 02:45 PM
We were unable to access Apple Hardware Test, until we pored through the manual and found that it must be invoked via the "D" key and not by holding down the Option key at startup, as before. (Yet another change in the secret key combinations for critical functions, helping to keep professional Apple technicians in business. At least it's documented!)

TZ
08-08-2006, 02:48 PM
The hard drive system worked fine, and we're thrilled to have four internal bays.

Four drive "sleds" are included, each with the necessary screws for attaching a SATA drive. The 250GB Seagate 7200.9 drive included with the computer had SATA-only power connections, while a 80GB Maxtor SATA drive from our Power Mac G5/1.8 had both SATA and traditional power connections, but it, too, fit into the supplied sockets when the sled was pushed into place.
Serial ATA Drive Interface

The Mac Pro comes standard with one 7200 rpm, 3 Gbps Serial ATA (SATA) disk drive and three additional 3 Gbps SATA slots for adding hard disk drives. The SATA drives interface through an AHCI 1.1 controller that supports advanced SATA-II features Native Command Queuing (NCQ) and PHY power management. NCQ increases performance on random workloads by allowing the drive to re-order commands to reduce seek time and increase transactional efficiency.

In addition, the Mac Pro has two unpopulated 3 Gbps SATA buses for expansion. Looking at System Profiler, all 6 SATA busses seem to be active :)

For more information on SATA, see Serial ATA International Organization (http://www.serialata.org/).
For information on the AHCI controller, see AHCI Specifications (http://www.intel.com/technology/serialata/ahci.htm).
_________________________
Ultra ATA Interface

In the Mac Pro, the South Bridge controller provides two Ultra ATA/100 interfaces for optical drives. The Mac Pro comes standard with one SuperDrive. The drive can read and write DVD media and CD media, as shown in Table 1 (http://developer.apple.com/documentation/HardwareDrivers/Conceptual/Mac_Pro_0608/Articles/M43_0906_arch.html#//apple_ref/doc/uid/TP40004479-SW4).
________________________
FireWire Controller

The Mac Pro has two IEEE-1394a FireWire 400 ports, which support transfer rates of 100, 200, and 400 Mbps and two IEEE-1394b FireWire 800 ports, which support transfer rates of 100, 200, 400, and 800 Mbps. For more information, see FireWire Developer Note (http://developer.apple.com/documentation/HardwareDrivers/Conceptual/HWTech_FireWire/index.html#//apple_ref/doc/uid/TP40003028).
________________________
Block Diagram Note: For additional information and a link to specific documentation regarding PCI Express and the interface to the AirPort Express module, refer to PCI Express Architecture (http://developer.apple.com/documentation/HardwareDrivers/Conceptual/Mac_Pro_0608/Articles/M43_0906_arch.html#//apple_ref/doc/uid/TP40004479-TPXREF110).
___________________________
Mac Pro: DIY (http://www.apple.com/support/manuals/macpro_diy/) - Do It Yourself Upgrades

Thanks to Barefeats (http://www.barefeats.com/hard91.html) for the heads up:

[QUOTE] the new 750GB Western Digital Caviar SE16 drive is the surprising new Speed King among 7200rm SATA drives for both random and sustained transfers (based on QB4) Western Digital 750GB Caviar is selling for about half as much as the Hitachi 7K1000



Hitachi: The First 1TB HD The Desktar 7K1000 should be in stores for Q1 2007. Both format, SATA II and PATA 133, will be available. The SATA version will feature 32MB of cache, vs. 8MB of cache for the PATA version. The estimated public price should be 399$. This drive will also be available with fewer platters. You will be able to install 4TB of storage space in a MacPro, without modifying either the case or the internal organization!

Transintl The Pro Sled - Mac Pro hard disk drive sled (https://www.transintl.com/store/category.cfm?Category=2705)

The Pro Sled gives you the capability of unlimited swappable drives for your Mac Pro.
Drives can be assigned and labeled for different clients, projects or backup.
They will plug directly into the Mac Pro main logic board.
Each Pro Sled kit includes drive screws and sells for $29 (one sled), $49 (two sleds) and $89 (four sleds).

Transintl Pro Caddy for Mac Pros (add 5th HD in spare optical bay) -
Pro Caddy for Apple Mac Pro (http://www.transintl.com/store/category.cfm?Category=2704) is a unique internal optical bay hard disk drive mounting solution.

Using the existing optical bay housing the Pro Caddy enables the installation of one additional, bootable hard drive. Once installed, the Pro Caddy frees up all four of the removable hard disk drives in the Mac Pro's factory bays.

SEAGATE PUZZLES Firmware version 3.AEE or later solves the slow sustained large block write speed issue for a single 7200.10 inside the Mac Pro.

The remaining performance issue is slow small random read speeds for one, two, three or four drives. No matter how many drives you configure in RAID 0 sets, the average random read speed for combined block sizes from 64K to 1024K is less than 30MB/s (based on QuickBench 3 testing). Until Seagate fixes this, we can't recommend the 7200.10 series as the ideal boot drive for the Mac Pro (or Power Mac).

Update: the latest Mac Pros seem to have resolved the issue with slow read speed on 7200.10.

Newer Tech offering SATA Extender cables for Mac Pro (2 unused onboard SATA ports) $24 bootable.
http://www.newertech.com/products/esata_cable.php
http://www.xlr8yourmac.com (http://www.xlr8yourmac.com/) helps explain how to, and what to expect.

Latest comparisons for best boot drive:
http://www.barefeats.com/quad08.html
Mac Pro: 10 drive comparison:
http://www.barefeats.com/hard88.html

TZ
08-08-2006, 06:39 PM
Graphics/Video Subsystem

The standard configuration graphics subsystem on the Mac Pro is the Nvidia GeForce 7300 GT card with 256 MB GDDR2 SDRAM, connected to the North Bridge IC by a x16 link (16 lane), 2.5 GHz PCI Express bus. The configuration supports two DVI ports for external video monitors and supports video mirroring mode and extended desktop display mode.

The ATI Radeon X1900 XT and Nvidia Quadro FX 4500 are available as configure-to-order options.
Video Developer Note (http://developer.apple.com/documentation/Hardware/Conceptual/HWTech_Video/index.html#//apple_ref/doc/uid/TP40003504).

PCI Express bus that supports the graphics subsystem and the PCI Express power constraints, refer to PCI Developer Note (http://developer.apple.com/documentation/Hardware/Conceptual/HWtech_PCI/index.html#//apple_ref/doc/uid/TP40003027).

PCI EXPRESS BUS CONCERNS Intel 5000X chipset (http://www.intel.com/products/chipsets/5000x/index.htm) for Mac Pro's PCI Express bus dynamically allocates a maximum of 26 lanes between the four slots.

Lane options:
16 + 1 + 4 + 4
16 + 1 + 1 + 8
8 + 8 + 1 + 8
8 + 8 + 4 + 4

THIS developer doc (http://developer.apple.com/documentation/Hardware/Conceptual/HWtech_PCI/Articles/pci_implementation.html#//apple_ref/doc/uid/TP40003937-DontLinkElementID_3)

By comparison, the Quad-Core G5 uses an Apple proprietary PCI Express controller that provides a total of 32 lanes (16 + 4 + 8 + 4).
It's also a shame Apple didn't throw in at least one 133MHz PCI-X slot in the Mac Pro since the 5000X chipset supports it without stealing bandwidth from the PCIe slots.
Barefeats Quick Takes (http://www.barefeats.com/quick.html)
__________________________

Video cards used in Mac Pro's must be EFI compliant cards.
BIOS compliant video cards will not work in EFI equipped Mac Pro's.

Apple Video (http://developer.apple.com/documentation/Hardware/Conceptual/HWTech_Video/Articles/Video_implementation.html#//apple_ref/doc/uid/TP40003994-SW1)

Nvidia says GeForce 8800 cards are PCIe compliant

GeForce 8800-series graphics cards and some motherboards. The report suggested that the compatibility problems arose because the cards weren't fully PCI Express compliant. Indeed, the PCI Special Interest Group's compliant components list (http://www.pcisig.com/developers/compliance_program/integrators_list/pcie/) lacked (and still lacks) mention of any GeForce 8800 or G80 products.

Radeon X1900 series R580 GPU incorporated 384 million transistors into its design. The GPU features 512-bit memory interface with support for GDDR3 and GDDR4.

X2900 XTX retail card comes as a two-slot, 9.5" design with a vapor chamber cooler. Vapor chambers are already found on high-end CPU coolers, so it would be no surprise to see such cooling on a high-end GPU either. The OEM version of the card is a 12" layout and features a quiet fan cooler. 1GB of GDDR4 memory is the reference configuration for Radeon X2900 XTX. Memory on the reference X2900 XTX cards was supplied by Samsung.

The GDDR3 version of the Radeon X2900 XT, features 512MB of GDDR3 and lower clock frequencies than the X2900 XTX. The X2900 XT is also one of the first Radeons to feature heatpipes on the reference design.

Native HDMI, will appear on all three versions of Radeon X2900. One 6-pin and one 8-pin (2x4) VGA power connectors are featured on Radeon X2900. Both connectors are also backwards compatible with 6-pin power supply cables.

Will Apple looks to new NVIDIA graphics for Final Cut
While aiming at its usual audience of 3D modelers and other graphics pros
with its new Quadro FX video cards, NVIDIA has let slip that Apple is eyeing the technology closely as a key to driving its video editing tools.[/b]

The Santa Clara-based GPU maker announced Quadro FX 4600 and 5600 (http://www.nvidia.com/object/quadro_fx_5600_4600.html) cards, calling them a perfect fit for video editing thanks to a much-improved unified architecture that does much of the heavier work for effects processing and video decoding.

Nvidia's G92 GeForce 8700 GTS due out November 12
8-layer PCB
512MB of GDDR3 memory
256-bit memory bus, Memory clock speed of 900MHz to 1GHz
— that's an effective 1.8GHz to 2GHz
PCI Express 2.0 connectivity
HDMI and DisplayPort display outputs
65nm
Nvidia's PureVideo video processing engine . . . .
Aperture and iPhoto 6 also use the pixel shaders on newer video hardware for non-destructive still image editing.

However, Apple's centerpiece Final Cut Pro editor has so far depended exclusively on CPUs to draw live footage, rapidly scaling down the quality of live previews as an editing crew shifts to editing HD video or multiple streams.

Offloading some or all of this task to a particularly flexible video card like the new Quadro would free Final Cut Pro to render multiple HD streams in real-time without sacrificing accuracy -- especially for the fledgling 2K and 4K video resolutions that frequently demand specialized hardware.

TechReport: Radeon HD X2900 XT (http://techreport.com/reviews/2007q2/radeon-hd-2900xt/index.x?pg=1)

Radeon HD 2950 XTX 3Q '07 - Radeon 2950 XTX will feature a new R650 graphics processor, which will be based on 65nm process technology. The card will also be outfitted with 1GB of GDDR4 memory, and it should arrive in August or September. The 80nm R600 die - will be in a very limited amount of chips, and pulling all resources to go 65 nano

1GB Radeon HD 2900 XT "Diamond Multimedia Announces the World's First 1GB Single GPU Consumer Graphics Card." will be the same size as the existing model— no 12"-long behemoth. Price-wise, it will cost something slightly north of $500.
_____________________
AMD talks DX10 drivers:
... we had to have a new OpenGL driver for HD 2000. And finally, the DX9 driver was mostly inspired from the R5xx DX driver, and so little influence came over from Xenos again. However, from a SW porting standpoint, we can truly make a difference. The HD2000 family’s architecture leverages so much from Xenos, that porting Xenos games to the R600 should be relatively easy. The games should have similar performance characteristics. There’s a very good synergy there.

How would you describe the maturity of R6 drivers at this point and how "target rich" the optimization opportunities still are going forward?

I think we have a very dedicated and excellent set of engineers working on our drivers.

However, given a new operating system, a new DX, a new OpenGL driver and a whole new architecture, we got the perfect storm!

I think that the drivers are very stable and give users a good experience. But when it comes to performance, there’s still a lot to optimize for. And it will take time to exploit the full performance potential of the hardware. They have done what was required, which is give the best $399 part, but there’s so much more to do… So I would say that a lot of progress has been made, but things are still “green”, especially on the DX10 front.

Finally:

I can’t help but be a little disappointed that we did not have enough time to get more optimizations into our drivers in time for the launch. I still cringe when I see poor performance (especially if it’s compared, to, say, our previous generation products)...

We also had some last minute performance possibilities, which is always incompatible with stability... We see in the latest driver, some 2x to 3x improvement in adaptive AA performance, for example, which is great but came later than I would have liked.
Beyond3D Interviews AMD (http://www.beyond3d.com/content/interviews/39/4)HD Radeon X2900 (http://www.techpowerup.com/img/07-05-02/HIS01.png)
Specifications (http://www.techpowerup.com/img/07-05-02/HIS03.png)
______________________
Mac Pro with Radeon X1900 XT
4 lane, 8 lane, and 16 lane PCIe slots (http://www.barefeats.com/quad14.html)
Mac Pro 3D Game "Slug Fest" (http://www.barefeats.com/quad10.html)

Revision 2 Radeon X1900 XTs
I'm no longer seeing issues with the new card, but that isn't the only change.

My Northbridge temperatures which formerly ran at a CONSTANT 140-160 farenheit with SMC on default (and about 120 with my fans all turning at 1600rpm) are now running 110-113 degrees farenheit with a 900 rpm PCI fan and only 130 farenheit at default. I'll keep an eye on this as the week goes on, but something definitely changed with this video card compared to the old one.Mac Pro/X1900 XT owner notes on WoW FPS Cap preventing artifacts/freezing -

buying a new replacement ATI X1900 XT The ship time for a new ATI x1900xt (Kit/separate card) is listed as 7-10 weeks might signal a possible graphics card refresh on the horizon....

Checked the X1900 XT heatsink for Dust clogging it (http://www.xlr8yourmac.com/Graphics/radeon_X1900XT/X1900XT_dust_buildup.html#storytop)

One of the hints found on the Blizzard website (this hint only applies while playing WoW) has helped immensely. I have found that if I limit the fps in WoW that I no longer see any of the artifacts or freezing that I was experiencing before. All one needs to do is: in the chat field type:
/console maxfps XX (XX being the desired fps frame limit)I currently have mine set at 40 fps and am seeing zero problems.

With that said, I hope that there is GPU refresh on the horizon (Perhaps the nVidia 8800)

Nicolas
08-09-2006, 02:55 AM
Pictures from the Mac Pro innerts (CPUs, etc)
http://www.powermax.com/articles_reviews/article.php?id=32

Reviews from the web:
ComputerWorld: Hands on The new Mac Pro is 'one screamer' (http://www.computerworld.com/action/article.do?command=viewArticleBasic&articleId=9002545)
ComputerWorld: Why IT loves the Mac Pro / Xserve (http://www.computerworld.com/action/article.do?command=viewArticleBasic&articleId=9002267)
MacIntouch: Preliminary Review & Tests (http://www.macintouch.com/reviews/macpro/prelim.html)
AnadTech reviews Mac Pro (http://www.anandtech.com/mac/showdoc.aspx?i=2811&p=5)
AnandTech's "Apple Mac Pro - PowerMac Successor" (http://www.anandtech.com/mac/showdoc.aspx?i=2816)
Ars Review (http://arstechnica.com/reviews/hardware/macpro.ars/)

Benchmarks:
Mac Pro vs G5 Photoshop test (http://www.chaddahlquist.net/2006/08/11/19/)
Barefeats: Quad Intel vs Quad PPC (http://www.barefeats.com/quad06.html)
MacIntouch Follow-up (http://www.macintouch.com/reviews/macpro/followup.html)
Mac Pro Rumors Archive (http://macpro.macrumors.com/)

Hands on with the Mac Pro: Putting it to work
Audio, video, and virtualization tasks put to the test

* The Mac Pro makes short work of ripping CDs into iTunes, and converting existing tracks between formats.
* HD video playback doesn’t pose much of a challenge for the Mac Pro. Playing back multiple HD clips at once didn’t cause any issues, even when I was recording that playback at the same time.
* The four CPU cores in the Mac Pro greatly speed the process of converting video from one format to another.
* More RAM is always good, particularly if you’re running applications in Rosetta.
* Rosetta’s performance on a fast Mac is quite impressive. (Rosetta is a great solution in the short term.)
* The Mac Pro runs very quietly, even with a high-powered video card installed.
MacWorld: Part II (http://www.macworld.com/2006/11/features/handsonmp2/index.php?pf=1)

for this Geekbench comparison, IÕm including most of the recent AMD
and Intel desktop processors (along with some not-so-recent desktop
processors) along with a wide variety of recent Mac models, ranging
from the PowerBook G4 (http://www.apple.com/macbook/) to the Mac Pro (http://www.apple.com/macpro/).

Average results from all the computers in the Geekbench Result Browser (http://www.geekpatrol.ca/browse/2006/) database. For Mac systems, results were collected from Geekbench for Mac OS X (32-bit) running on systems with standard processors (i.e., no over-clocked processors or processor upgrades) and with at least 512MB of RAM.

100 is the score a Power Mac G5 @ 1.6GHz would receive.
As always, higher scores are better.

This Entry
http://www.geekpatrol.ca/wp-content/themes/fauna/images/feed.png (http://www.geekpatrol.ca/2006/12/geekbench-comparison-december-2006/feed/)

"Geekbench Comparison (December 2006)" (http://www.geekpatrol.ca/2006/12/geekbench-comparison-december-2006/) was written on Dec 06 2006 by John Poole (http://www.geekpatrol.ca/author/John/) and filed under Articles (http://www.geekpatrol.ca/category/articles/), Benchmarks (http://www.geekpatrol.ca/category/benchmarks/), Geekbench (http://www.geekpatrol.ca/category/geekbench/).

http://www.geekpatrol.ca/2006/12/geekbench-comparison-december-2006/

AMUG (Arizona Macintosh Club) have put online a very comprehensive article about the MacPro 2.66 GHz.

firmware flashing procedure of the Sony DVD drive. (http://www.hardmac.com/articles/62/) Once its limitations overriden, it is a NEC again and becomes world champion for DVD ripping (three times faster).

[/URL][URL]http://www.amug.org/amug-web/html/amug/reviews/articles/intel/macpro/ (http://www.amug.org/amug-web/html/amug/reviews/articles/intel/macpro/)

Nicely down full set of photos of internals and components:
http://mac.powerbras.nl/

TZ
08-13-2006, 12:37 PM
A Basic Guide for Migrating to Intel-Macs

If you are migrating a PowerPC system (G3, G4, or G5) to an Intel-Mac be careful what you migrate. Keep in mind that some items that may get transferred will not work on Intel machines and may end up causing your computer's operating system to malfunction.

Rosetta supports "software that runs on the PowerPC G3 or G4 processor that are built for Mac OS X". This excludes the items that are not universal binaries or simply will not work in Rosetta:


Classic Environment, and subsequently any Mac OS 9 or earlier applications
Screensavers written for the PowerPC
System Preference add-ons
All Unsanity Haxies
Browser and other plug-ins
Contextual Menu Items
Applications which specifically require the PowerPC G5
Kernel extensions
Java applications with JNI (PowerPC) libraries
What Can Be Translated by Rosetta (http://developer.apple.com/documentation/MacOSX/Conceptual/universal_binary/universal_binary_exec_a/chapter_7_section_2.html#//apple_ref/doc/uid/TP40002217-CH210-BCICICJH).

In addition to the above you could also have problems with migrated cache files and/or cache files containing code that is incompatible.

If you migrate a user folder that contains any of these items, you may find that your Intel-Mac is malfunctioning. It would be wise to take care when migrating your systems from a PowerPC platform to an Intel-Mac platform to assure that you do not migrate these incompatible items.

If you have problems with applications not working, then completely uninstall said application and reinstall it from scratch. Take great care with Java applications and Java-based Peer-to-Peer applications. Many Java apps will not work on Intel-Macs as they are currently compiled. As of this time Limewire, Cabos, and Acquisition are available as universal binaries.

Do not install browser plug-ins such as Flash or Shockwave from downloaded installers. These versions are not universal binaries and will not work on Intel-Macs. The version of OS X installed on your Intel-Mac comes with special compatible versions of Flash and Shockwave plug-ins for use with your browser.

The same problem will exist for any hardware drivers such as mouse software unless the drivers have been compiled as universal binaries. For third-party mice the only current choice is USB Overdrive. Contact the developer or manufacturer of your third-party mouse software to find out when a universal binary version will be available.

Also be careful with some backup utilities and third-party disk repair utilities. Disk Warrior (does not work), TechTool Pro (does not work), SuperDuper (newest release works), and Drive Genius (untested) may not work properly on Intel-Macs. The same caution may apply to the many "maintenance" utilities that have not yet been converted to universal binaries.

MacFixit's Rosetta Compatibility Index (http://www.macfixit.com/article.php?story=20060126094146180).

Additional links:

Intel In Macs (http://www.apple.com/intel/)
Apple Guide to Universal Applications (http://guide.apple.com/universal/)
MacInTouch List of Compatible Universal Binaries (http://www.macintouch.com/imacintel/ubinaries.html)
MacInTouch List of Rosetta Compatible Applications (http://www.macintouch.com/imacintel/rosettacompat.html)
MacUpdate List of Intel-Compatible Software (http://www.macupdate.com/macintel.php)
http://www.versiontracker.com/macintel/
(http://www.macupdate.com/macintel.php)

As a test, installed a base OS and used the Apple Migration assistant to move over everything from a preconfigured PPC machine. Worked wonderfully. Users, Apps, prefs, everything - almost. A very few things did not work:

- Unsanity haxies: reinstalled current UB versions - work fine, had to reconfigure prefs.
- Flip4Mac: reinstalled current UB version - work fine...older version worked fine with Rosetta.

Other things that were noted in first tests, all using the most current versions available:

- MS RDC client: works fine via Rosetta.
- Monolingual: works fine.
- NetRestore: works fine.
- Tinkertool: works fine.
- Batchmod: works fine.
- Sophos AV: works fine.
- SharePoints: works fine.
- Onyx: works fine.
- Cisco VPN cleint: works fine.
- WhatSize: works fine.
- LoginWindow Manger: Works fine.
- Casper (imaging/updating/admin/remote acecess/tracking tool): works fine.

Tested on an early MacBook Pro, a MacBook, and a Mini.....all stock, base configurations. Still testing, printing and the like.

MacStumbler does not seem to work.


Intel-based Mac: Some migrated applications may need to be updated (http://docs.info.apple.com/article.html?artnum=304302)
After using Migration Assistant or Setup Assistant to migrate applications from a PowerPC-based Mac to an Intel-based Mac, some of the migrated applications may not launch or function correctly. [...]
When Migration Assistant or Setup Assistant is used to migrate applications between two computers, it will bring over any applications it finds on the source computer that are not already installed on the destination computer. In some cases, a PowerPC-based Mac may have applications that will not run correctly through Rosetta on the Intel-based Mac.

TZ
08-21-2006, 09:56 AM
Run MacOS 9 and the gClassic Enviromenth on an Intel Mac
http://www.uneasysilence.com/archive/2006/08/7352/

Who said you canft use MacOS9 because you have an Intel Mac?

It is completely possible with a little bit of tinkering, and a really cool universal application called Sheep Shaver, which came to us via tip from Kazaki. Sheep Shaver is a full speed eClassicf emulator for Windows, Linux, and Intel based Macs, that runs older MacOSfs at shockingly full speed!

TZ
08-23-2006, 07:35 AM
In any case, after setting up the computer with my accustomed work and personal software, I experienced a kernel panic each time I tried to shut down. I traced the problem to the latest Cisco VPN Client (universal binary, version 4.9.00.0050). Once I uninstalled the Cisco VPN Client, I no longer got the kernel panic at shutdown.
I'd like to know if any other MacInTouch readers have seen this problem.

TZ
11-05-2006, 06:37 AM
Power Usage of Mac Pro It requires 410 watts to start up, idles at 300 watts, and peaks at 430 watts when running Doom 3, the most demanding power wise of the ten apps we tried.

The Mac Pro had 16GB of RAM, four internal hard drives, and a Radeon X1900 XT. We're confident the Mac Pro's 980 watt power supply can handle just about everything you can do to upgrade or expand a Mac Pro.

Mac Pro: Power consumption and thermal output (BTU) information
Update: http://docs.info.apple.com/article.html?artnum=304378

Mac Pro (Late 2006) Quad Core 2.66 Ghz Mac Pro
Idle and max consumption:
171 W 445 W
854 BTU/hr 1,519 BTU/hr

Notes

*Power consumption data (Watts) is measured from the wall power source and includes all power supply and system losses.
Additional correction is not needed.

*"CPU Max" is defined as running a compute-intensive test application that maximizes processor usage and therefore power consumption.

*These numbers reflect a 23° C (73.4° F) ambient running environment. Increased ambient temperatures will require faster fan speeds which will increase power consumption.
At 35° C (95° F), 50 W should be added to reflect increased power consumption.

*1000W PSU do not deliver more than 500-600W effectively

the main problem is not really power required in load, but rather the power burnt for nothing when being idle or in sleep mode. Power management system for desktop should be strongly improved in order to reach similar level than the one found in notebooks.

TZ
11-05-2006, 04:37 PM
http://developer.apple.com/technotes/tn2006/images/tn2166_subtitle.gif (http://developer.apple.com/technotes/tn2006/tn2166.html)

Apple has switched to a new disk partitioning scheme known as the GUID partition table, or GPT. This new scheme offers a number of advantages over the previous scheme, but it also presents some new challenges. This technote describes GPT in general, and gives some specific details about how Apple uses GPT.

You should read this technote if you're writing software that interacts with Apple's implementation of GPT. Specifically, this technote is vital if you're writing a disk utility for Mac OS X, or trying to run some alternative operating system, like Linux, on a Macintosh computer.

The Road to GPT (http://developer.apple.com/technotes/tn2006/tn2166.html)

Apple's previous disk partitioning scheme, known as Apple partition map (APM), was introduced with the Macintosh II in 1987. The scheme was very well designed, and it has survived, with very few changes, for almost twenty years.

However, in recent years APM's limitations have been looming on the horizon. Specifically, APM is restricted to 32-bits worth of blocks. With a standard block size of 512 bytes, this translates to a maximum disk size of 2 TB. With the rate that hard disks are growing, it's easy to imagine a typical desktop computer shipping with more than 2 TB of storage in the next few years.

Apple did consider extending APM to support larger disks. However, as such a change would break all existing partitioning tools, it was just as convenient to switch to an entirely new partitioning scheme. After some serious thought, Apple decided to adopt GPT.

GUID partition table (GPT) partitioning scheme was introduced by Intel as part of an effort to introduce more modern firmware to generic PC hardware. Traditional PC hardware uses BIOS firmware, which uses a partitioning scheme known as master boot record (MBR). MBR has lots of severe limitations, and is not appropriate for a modern computer. Intel's modern firmware, known as the Extensible Firmware Interface (EFI), includes a new partitioning scheme, GPT.

Apple's Support for GPT Any Macintosh computer running Mac OS X 10.4 and later can mount GPT-partitioned disks. Intel-based Macintosh computers can boot from GPT. By default, the internal hard disk is formatted as GPT.

IMPORTANT: While Intel-based Macintosh computers can boot from GPT and APM, Apple only supports booting Mac OS X on these machines from GPT. Apple's GUI tools, like the Installer, will prevent you installing Mac OS X for an Intel-based Mac on non-GPT disks.

On all Intel-based Macintosh computers (and, starting with Mac OS X 10.4.6, PowerPC-based computers as well), Disk Utility has full support for GPT. The <code>diskutil</code> command line tool also includes GPT support. See its man page (http://x-man-page://8/diskutil) for details.

f you're using a PowerPC-based computer running Mac OS X 10.4.6, you'll be disappointed to discover that the <code>diskutil</code> man page has not been updated for GPT. Rest assured that the actual <code>diskutil</code> code is GPT-aware. Use the tool's built-in help to learn how to create a GPT disk.

ESP Explained

The EFI system partition (ESP) is a special partition from which EFI can load EFI (boot-time) device drivers. The EFI firmware in Macintosh computers fully supports the ESP, although Apple does not currently use it for anything. We create the ESP on big disks to make things easier if we ever need to use ESP-based drivers. We strongly recommend that you do the same.

The Last Block

There are a number of third party disk drives that, due to buggy firmware, report an error if you access the last block on the disk. This was not a serious problem for APM. It's easy to create an APM that avoids the last block by creating a small <code>Apple_Free</code> partition at the end of the disk (this is what Apple's implementation of APM does).

For GPT, however, this problem is much more serious. The GPT specification (http://developer.apple.com/technotes/tn2006/tn2166.html#SECFURTHERREADING) requires that the last block of the disk contain the backup partition table header; you can't just choose to avoid it!

Apple handles this problem by ignoring any error when reading the backup partition table header in the last block of a GPT disk. If such an error occurs, the disk has a valid primary partition table and an invalid backup. According to the rules laid out by the GPT specification, such a disk is still considered to be a valid GPT disk.

On the other hand, when partitioning a disk as GPT, if writing to the last block results in an error, the partitioning operation will fail. Such a disk can't be partitioned as a GPT disk using Apple software.

GUID Based Partitioning Scheme (http://www.kernelthread.com/publications/firmware/) (half-way down page)

GUID Partition Table (GPT) and Master Boot Record (MBR)<table border="0" cellpadding="0" cellspacing="0"><tbody><tr><td style="padding-top: 8px;" align="left" height="19" valign="top" width="16">http://www.microsoft.com/whdc/images/fileType/icon_text.gif</td><td style="font-family: Verdana,Arial,Helvetica,sans-serif; font-size: 70%; line-height: 140%; padding-top: 5px;" align="left" valign="top">Custom Bootstrap Actions (http://www.microsoft.com/whdc/system/platform/firmware/OEMBoot_Vista.mspx#)http://www.microsoft.com/whdc/images/shared/common/space.gif</td></tr><tr><td style="padding-top: 8px;" align="left" height="19" valign="top" width="16">http://www.microsoft.com/whdc/images/fileType/icon_text.gif</td><td style="font-family: Verdana,Arial,Helvetica,sans-serif; font-size: 70%; line-height: 140%; padding-top: 5px;" align="left" valign="top">MBR System Disk Conversion for 64-bit Windows (http://www.microsoft.com/whdc/device/storage/GPT-64bit.mspx)http://www.microsoft.com/whdc/images/shared/common/space.gif</td></tr><tr><td style="padding-top: 8px;" align="left" height="19" valign="top" width="16">http://www.microsoft.com/whdc/images/fileType/icon_text.gif</td><td style="font-family: Verdana,Arial,Helvetica,sans-serif; font-size: 70%; line-height: 140%; padding-top: 5px;" align="left" valign="top">Modifications to Microsoft Boot Components: Update (http://www.microsoft.com/whdc/system/platform/firmware/Boot_Modifications.mspx) http://www.microsoft.com/whdc/images/fileType/new.gif</td></tr><tr><td style="padding-top: 8px;" align="left" height="19" valign="top" width="16">http://www.microsoft.com/whdc/images/fileType/icon_text.gif</td><td style="font-family: Verdana,Arial,Helvetica,sans-serif; font-size: 70%; line-height: 140%; padding-top: 5px;" align="left" valign="top">Using GPT Drives (http://www.microsoft.com/whdc/device/storage/GPT-on-x64.mspx)http://www.microsoft.com/whdc/images/shared/common/space.gif</td></tr><tr><td style="padding-top: 8px;" align="left" height="19" valign="top" width="16">http://www.microsoft.com/whdc/images/fileType/icon_text.gif</td><td style="font-family: Verdana,Arial,Helvetica,sans-serif; font-size: 70%; line-height: 140%; padding-top: 5px;" align="left" valign="top">Windows and GPT FAQ (http://www.microsoft.com/whdc/device/storage/GPT_FAQ.mspx)http://www.microsoft.com/whdc/images/shared/common/space.gif</td></tr></tbody></table>

TZ
11-19-2006, 11:39 AM
Here's the full quote from Intel Tech:
-----------------------------------------------------------------------------

I understand that you are seeking information regarding the operating temperature of the Intel(R) Xeon(R) processors 5000.

Quad-Core Intel Xeon Processor 5300 Series with system bus speed of 1066 and 1333 MHz
http://support.intel.com/support/processors/xeon5k/

The maximum operating temperature of the Intel(R) Xeon(R) processor 5150 is 65 degrees Celsius.
The maximum operating temperature of the Intel(R) Xeon(R) processor 5160 is 56.5 degrees Celsius.

As long as the processor is operating under this temperature it is operating within specifications. We recommend setting any thermal alarms about three (3) degrees below the maximum recommended temperature for the processor. We do not have a normal operating temperature for the processor as this temperature will vary depending on the chassis and other hardware installed on the system as well as the actual load the software is placing on the processor.

You can double check this above mentioned temperatures at the following websites:

Recommended temperature for the Intel(R) Xeon(R) processor 5150:
http://processorfinder.intel.com/details.aspx?sSpec=SL9RU

Recommended temperature for the Intel(R) Xeon(R) processor 5160:
http://processorfinder.intel.com/Details.aspx?ProcFam=0&sSpec=SL9RT&OrdCode=

Currently, we do not have the minimum and maximum values for the cores of the processor, nor for the CPU A and CPU B Heatsink.

Please do not hesitate to contact us again if you need further assistance.

Sincerely,

Adolfo S.
Intel(R) Technical Support

Intel(R) Processor Support Web Site:
http://support.intel.com/support/processors/index.htm
Intel is a registered trademark of Intel Corporation or its subsidiaries in the United States and other countries.
_____________________
http://support.intel.com/products/chipsets/5000x/index.htm
http://support.intel.com/support/processors/xeon5k/

From the author of Hardware and Temperature Monitor:
- When speaking about temperature specifications it is important to note that all limits can be applied to specific sensor locations and specific measurement methods only. So it does not make sense to speak about a "maximum operating temperature for processor X", you'll have to specify the "maximum operating temperature for the sensor location A when monitoring processor X under condition B".

- Intel Xeon 5100 processors use a variety of internal and external temperature sensors, and several different monitoring technologies:

1) The processor case temperature is monitored by an internal temperature diode and the so-called "Thermal Control Circuit" (TCC).

There are two different mechanisms called "Thermal Monitor" (TM) and "Thermal Monitor 2" (TM2) that allow the CPU to check and control its own temperature. This technology is considered to be *outdated* for the Xeon 5100
although it is still built in to maintain compatibility with older system designs. Temperature Monitor 4.0 does NOT display this sensor on the Mac Pro.

2) Each core of each CPU is monitored by several internal temperature sensors that are placed at critical points on the chip-die. These sensors are called "Digital Thermal Sensors" (DTS). Their values are automatically
averaged and smoothed, reflecting high-precision readings for the situation at the cores themselves. The DT Sensors are displayed as "CPU Core" sensors by Temperature Monitor. They represent the *recommended* technology to monitor current Xeon 5100 processors.

3) There are external sensors, measuring the temperature at the processor heat sinks under specifications defined by the manufacturer, in this case Apple. These sensors are reflecting the processor package and airflow temperatures. They are being displayed as "CPU Heat Sink" sensors by Temperature Monitor.

- The maximum operating temperature mentioned by the Intel Tech representative is based on a special measurement method defined by the "Dual-Core Intel Xeon Processors 5100 Series Thermal/Mechanical Design Guidelines"
( http://www.intel.com/design/xeon/guides/313357.htm ).
The readings basically reflect a "TCC-like" approach, *not* the DTS technology. DTS values are expected
to be much higher than the values measured at the outer CPU package.

- It is correct that Intel did not and will not officially publish absolute temperature limits for the "on core" DTS sensors of the Xeon 5100. However, based on a series of tests run with Apple Mac Pro systems, it can be assumed that the maximum allowed DTS temperature for which Intel and Apple designed the Mac Pro is 85&#176;C. (To put this in relation, it should be noted that the maximum DTS temperature for Intel Core processors under normal operating conditions is 100&#176;C and the so-called "catastrophic" limit is at 125&#176;C. For more information, see the official data sheets at
http://www.intel.com/design/mobile/datashts/309221.htm .
Of course the Intel Core series should not directly be compared to the 5100 series.)

- Temperature Monitor displays the maximum sensor limits in the third column of the window "Sensor Overview". The application makes sure to only display limits for sensor locations where the specifications are known. The values either come from official documentation of Apple, or internal tables built into Mac OS X, or tables built into the firmware of the system, depending on what Macintosh model and what sensor location is being monitored.

- Even if somebody should assume that the maximum limit of 85&#176;C for the Xeon 5100 DTS core sensors were incorrect, the basic statement of Temperature Monitor that the "displayed reading for core X is Y degrees below the displayed limit" would still be correct. That's because the DTS technology computes the current temperature as *difference* to the upper limit, not as absolute temperature. The core will be overheating only if the current value approaches the limit. It doesn't really matter what the exact limit is for monitoring purposes at that particular sensor.

- All up-to-date Intel processors simply cannot overheat. As mentioned, they use several different technologies to monitor themselves and will take immediate action automatically if the temperatures reach a critical condition. The CPUs will "throttle down" either by reducing their clock frequencies, or by repeatedly switching to periods of inactivity for some microseconds. These protection mechanisms work independently of operating system and mainboard.

- The Mac Pro uses four different fans. Each fan is controlled *individually*, using the temperature information in different thermal zones. Some third-party applications to manipulate the minimum fan speeds may give you the wrong impression that fan speeds are controlled by CPU temperature only.

- Each FB-DIMM contains its own controller, called "Advanced Memory Buffer (AMB)".
Each AMB also contains its own temperature sensor. The maximum temperature of an FB-DIMM, measured by the AMB, lies in the interval between 95 and 125&#176;C (it will vary from manufacturer to manufacturer), the typical value is 110&#176;C. So all memory-related readings mentioned in the thread should be considered low and well within specs.

I hope this information is helpful to you. If I can be of further assistance,
don't hesitate to contact me.

Best regards,
Marcel Bresink

TZ
12-05-2006, 09:29 AM
EFI and Windows Vista (http://www.microsoft.com/whdc/system/platform/firmware/efibrief.mspx)
Updated: April 20, 2006

BIOS as PC Firmware
EFI as PC Firmware
Evaluating BIOS vs. UEFI Support
EFI, UEFI Support, and Windows Vista

This paper briefly describes Microsoft plans for supporting the Extended Firmware Interface (EFI) in Microsoft Windows operating systems and provides a brief background for related technical issues.

Unified Extensible Firmware Interface (UEFI)

Q: What is UEFI?
A: UEFI (Unified Extensible Firmware Interface) will be a specification detailing an interface that helps hand off control of the system for the pre-boot environment (i.e.: after the system is powered on, but before the operating system starts) to an operating system, such as Windows* or Linux*. UEFI will provide a clean interface between operating systems and platform firmware at boot time, and will support an architecture-independent mechanism for initializing add-in cards.

Q: What is the relationship between EFI and UEFI?
A: The UEFI specification will be based on the EFI 1.10 specification published by Intel with corrections and changes managed by the Unified EFI Forum. Intel still holds the copyright on the EFI 1.10 specification, but has contributed it to the Forum so that the Forum can evolve it. There will not be any future versions of the EFI specification, but customers who license it can still use it under the terms of their license from Intel. The license to the Unified EFI Specification will come from the Forum, not from Intel.

http://www.uefi.org/about/
http://en.wikipedia.org/wiki/UEFI

Vista also supports booting from GUID.

TZ
12-30-2006, 10:59 AM
Intel says good-bye to FB-DIMMs in some 2-socket configs

<!-- /page title --><!-- intro text -->FB-DIMM has been the sole memory option for Xeon machines now for quite some time. FB-DIMMs are necessary for high levels of performance in a shared bus architecture, but they also contain some interconnect logic, which makes them run hot, consume power, and be generally undesirable for all but the highest-end applications.
<!-- article ad --> <!-- BEGIN-NEW NetShelter Ad Tag for Geek.com 300x250,336x280 --> <script language="JavaScript" type="text/javascript"> if (!window.netshel_ord) { netshel_ord=Math.random()*10000000000000000; } if (!window.netshel_tile) { netshel_tile=1; } document.write('<script language="JavaScript" src="http://ad.doubleclick.net/adj/ns.geek/chipgeek;sz=300x250,336x280;tile='+netshel_tile +';ord=' + netshel_ord + '?" type="text/javascript"></scr' + 'ipt>'); netshel_tile++; </script> <!-- END AD TAG -->
<!-- /article ad --> In the near future, some dual-socket Xeon motherboards will begin offering DDR-II as a memory option. The dual-sockets (up to 8 cores) will only be supported due to the limited bandwidth and memory sizing.

FB-DIMMs can scale to 16, 24, or more DIMM sockets. DDR-II is limited to 8 DIMM sockets per channel. And whereas AMD’s architecture can have 8 sticking off every CPU due to its on-die memory controller, Intel’s Xeon solutions must still go through the north bridge. That means a shared, common point of access to the memory beyond.

FB-DIMM memory consumes a significantly greater amount of power compared to DDR-II. In moving from 2 GB in a system to 8 GB, DDR-II would typically see an increase of about 11 watts. The same memory leap with FB-DIMM would see a 26 watt increase. In projecting forward the power numbers, 16 GB, 32 GB, and 64 GB configurations, overall system power draw is much higher thanks to FB-DIMM’s hungry hungry hippo attitude toward power consumption.

Read more about the FB-DIMM to DDR-II migration at The Inquirer (http://www.theinquirer.net/?article=42183). And read more about the power consumption curves at Neal Nelson’s website (http://www.worlds-fastest.com/wfz991.html) .

Samsung (http://www.samsung.com/global/business/semiconductor/newsView.do?news_id=837)Samsung announced that it will launch mass production of 800MHz DDR2 chips based on 60nm engraving process by the end of the year.
Such chips feature 512MB capacity, and will strongly reduce the cost of 4GB DDR2 RAM modules

http://files.macbidouille.com/news/200709/samsung.jpg

This thinner engraving process is also an advantage for Samsung as it will allow the company to save on production cost. Indeed, one will be able to generate 40% more chips from the same silicon wafer. This increase of the yield will come right on time to allow DDR2 manufacturers to remain profitable during a period where Vista-driven memory over production has not been fully digested by the market.

TZ
01-01-2007, 12:17 PM
Extensible Firmware Interface (EFI) Boot ROM

The Extensible Firmware Interface (EFI) boot ROM consists of 2 MB of on-board flash EEPROM. It includes the hardware-specific code and tables needed to start up the computer, load an operating system, and provide common hardware access services.The EFI boot ROM connects to the South Bridge IC via the Firmware hub (FWH) bus.

Unified Extensible Firmware Interface (UEFI)

Q: What is UEFI?
A: UEFI (Unified Extensible Firmware Interface) will be a specification detailing an interface that helps hand off control of the system for the pre-boot environment (i.e.: after the system is powered on, but before the operating system starts) to an operating system, such as Windows* or Linux*. UEFI will provide a clean interface between operating systems and platform firmware at boot time, and will support an architecture-independent mechanism for initializing add-in cards.

Q: What is the relationship between EFI and UEFI?
A: The UEFI specification will be based on the EFI 1.10 specification published by Intel with corrections and changes managed by the Unified EFI Forum. Intel still holds the copyright on the EFI 1.10 specification, but has contributed it to the Forum so that the Forum can evolve it. There will not be any future versions of the EFI specification, but customers who license it can still use it under the terms of their license from Intel. The license to the Unified EFI Specification will come from the Forum, not from Intel.

http://www.uefi.org/about/
http://en.wikipedia.org/wiki/UEFI
_________________________

EFI firmware update Boot ROM Version MP11.005C.B04
Mac Pro EFI Firmware Update 1.1 (http://www.apple.com/support/downloads/macproefifirmwareupdate11.html)
http://docs.info.apple.com/article.html?artnum=303469-en

Reader reports of problems - the EFI update often doesn't work until all external devices are unplugged. In my case, I reverted back to stock keyboard and mouse, and it ran fine after that.

Drivers, pre-boot applications and OS loaders for EFI are written in C,
not assembler. The EFI specification includes clean APIs for text and
graphical screen output, keyboard and mouse input, and access to file
systems and block devices. EFI allows settings to be stored in NVRAM in
the form of named variables, i.e. key-value pairs.

OSx86 Project: Extensible Firmware Interface (http://wiki.osx86project.org/wiki/index.php/EFI)

According to an unofficial Apple/Intel FAQ (http://appleintelfaq.com/), the graphics drivers seem to have a direct association with EFI modules. This can be seen in the System Profiler listing (http://appleintelfaq.com/imac/system_profiler.html) that they provide as "EFI Driver Version: 01.00.063". Some users have commented that an EFI module may be responsible for ROM BIOS initialization on ATI graphics cards.

Graphics/Displays:

ATI Radeon X1600:
Chipset Model: ATY,RadeonX1600
Type: Display
Bus: PCIe
VRAM (Total): 128 MB
Vendor: ATI (0x1002)
Device ID: 0x71c5
Revision ID: 0x0000
EFI Driver Version: 01.00.063

See Also:
More Power to Firmware (Kernelthread) (http://www.kernelthread.com/publications/firmware/)

The Unified Extensible Firmware Interface (UEFI) (http://www.uefi.org/about/) specification defines a new model for the interface between operating systems and platform firmware. The interface consists of data tables that contain platform-related information, plus boot and runtime service calls that are available to the operating system and its loader. Together, these provide a standard environment for booting an operating system and running pre-boot applications.

http://www.uefi.org/about/

In the rare case that your computer displays a black screen after starting up, you will need to restore your computer's firmware back to factory condition before you can apply the firmware update. Follow the directions at here to restore your firmware.

About firmware updates for Intel-based Macs http://docs.info.apple.com/article.html?artnum=303364

Mac OS X: Firmware Updates for Intel-based Macs
http://docs.info.apple.com/article.html?artnum=303880

Firmware Restore CD
http://www.apple.com/support/downloads/firmwarerestorationcd13.html

Unified Extensible Firmware Interface - EFI<table border="0" cellpadding="0" cellspacing="0"><tbody><tr><td style="padding-top: 8px;" align="left" height="19" valign="top" width="16">http://www.microsoft.com/whdc/images/fileType/icon_text.gif</td><td style="font-family: Verdana,Arial,Helvetica,sans-serif; font-size: 70%; line-height: 140%; padding-top: 5px;" align="left" valign="top">EFI and Windows Vista (http://www.microsoft.com/whdc/system/platform/firmware/efibrief.mspx)http://www.microsoft.com/whdc/images/shared/common/space.gif</td></tr><tr><td style="padding-top: 8px;" align="left" height="19" valign="top" width="16">http://www.microsoft.com/whdc/images/fileType/icon_text.gif</td><td style="font-family: Verdana,Arial,Helvetica,sans-serif; font-size: 70%; line-height: 140%; padding-top: 5px;" align="left" valign="top">Imaging Guidelines for Windows Server Longhorn on UEFI Systems (http://www.microsoft.com/whdc/system/platform/firmware/uefiguide.mspx) http://www.microsoft.com/whdc/images/fileType/new.gif</td></tr><tr><td style="padding-top: 8px;" align="left" height="19" valign="top" width="16">http://www.microsoft.com/whdc/images/fileType/icon_text.gif</td><td style="font-family: Verdana,Arial,Helvetica,sans-serif; font-size: 70%; line-height: 140%; padding-top: 5px;" align="left" valign="top">Microsoft Extensible Firmware Initiative FAT32 File System Specification, V. 1.03 (http://www.microsoft.com/whdc/system/platform/firmware/fatgen.mspx)http://www.microsoft.com/whdc/images/shared/common/space.gif</td></tr><tr><td style="padding-top: 8px;" align="left" height="19" valign="top" width="16">http://www.microsoft.com/whdc/images/fileType/icon_text.gif</td><td style="font-family: Verdana,Arial,Helvetica,sans-serif; font-size: 70%; line-height: 140%; padding-top: 5px;" align="left" valign="top">Microsoft Portable Executable and Common Object File Format Specification (http://www.microsoft.com/whdc/system/platform/firmware/PECOFF.mspx)http://www.microsoft.com/whdc/images/shared/common/space.gif</td></tr><tr><td style="padding-top: 8px;" align="left" height="19" valign="top" width="16">http://www.microsoft.com/whdc/images/fileType/icon_text.gif</td><td style="font-family: Verdana,Arial,Helvetica,sans-serif; font-size: 70%; line-height: 140%; padding-top: 5px;" align="left" valign="top">UEFI Support and Requirements: Microsoft Windows Server "Longhorn" (http://www.microsoft.com/whdc/system/platform/firmware/uefireg.mspx)http://www.microsoft.com/whdc/images/shared/common/space.gif</td></tr><tr><td style="padding-top: 8px;" align="left" height="19" valign="top" width="16">http://www.microsoft.com/whdc/images/fileType/icon_powerpoint.gif</td><td style="font-family: Verdana,Arial,Helvetica,sans-serif; font-size: 70%; line-height: 140%; padding-top: 5px;" align="left" valign="top">Unified EFI Update [WinHEC 2005; 493 KB] (http://download.microsoft.com/download/9/8/f/98f3fe47-dfc3-4e74-92a3-088782200fe7/TWAR05018_WinHEC05.ppt)http://www.microsoft.com/whdc/images/shared/common/space.gif</td></tr><tr><td style="padding-top: 8px;" align="left" height="19" valign="top" width="16">http://www.microsoft.com/whdc/images/ext_ani.gif (http://www.microsoft.com/info/cpyright.htm)</td><td style="font-family: Verdana,Arial,Helvetica,sans-serif; font-size: 70%; line-height: 140%; padding-top: 5px;" align="left" valign="top">Extensible Firmware Interface Specification (http://developer.intel.com/technology/efi/download.htm)http://www.microsoft.com/whdc/images/shared/common/space.gif</td></tr><tr><td style="padding-top: 8px;" align="left" height="19" valign="top" width="16">http://www.microsoft.com/whdc/images/ext_ani.gif (http://www.microsoft.com/info/cpyright.htm)</td><td style="font-family: Verdana,Arial,Helvetica,sans-serif; font-size: 70%; line-height: 140%; padding-top: 5px;" align="left" valign="top">Implementing the Unified Extensible Firmware Interface with Intel's Framework (http://www.intel.com/intelpress/sum_efi.htm)</td></tr></tbody></table>

TZ
01-01-2007, 12:53 PM
About the SMC Firmware Updates (http://docs.info.apple.com/article.html?artnum=303725)
Power and thermal management, the Sleep LED, and battery.
Controlling the fans, supporting Sudden Motion Sensor, ALS, and the power switch.There are two parts to this process: Installing the firmware update
application, and using the firmware update application to install the
SMC Firmware Update.
Download the SMC Firmware Update 1.0, which is available at Apple Downloads (http://www.apple.com/support/downloads/).
Quit open applications, which are marked in the Dock with a small triangle.
Double click the SMC Update.pkg file.
Click Continue after reading the licensing agreement, then click Agree if you accept the terms.
Click Install to install the update application.
After the software is successfully installed, click Close to start the
updater application. (If you inadvertently skipped this part, you can
find the updater in /Applications/Utilities.)
Read the onscreen firmware update installation instructions, then click Restart to proceed.As your computer starts, the SMC will be updated. Your computer's fans will run at full speed during the update but will return to normal once the update completes. The computer will restart again once the firmware
update is complete. Important: Do not interrupt the update process.

After your computer restarts a second time, you should see a message
stating that the update was successful. Click OK to continue.

Note: If you see the dialog that you saw in step 7 instead of a message that the update was successful, try installing the update again (use the updater in /Applications/Utilities/ ).

Note: This update requires a Macintosh-formatted disk or
partition. If you are starting up using a RAID (Redundant Array of
Independent Disks) subsystem, you must connect a Macintosh-formatted
storage device to apply this update.

Some USB and FireWire devices may prevent firmware updates from
installing correctly. If you are having trouble installing an update
try disconnecting non-essential devices and use only an Apple keyboard,
mouse, and display.

How to reset the System Management Controller (SMC) (http://docs.info.apple.com/article.html?artnum=304123)

The SMC controls several functions, including:
Telling the computer when to turn on, turn off, sleep, wake, idle, and so forth.
Handling system resets from various commands.
Controlling the fans.Note that resetting the SMC does not reset the PRAM.

Resetting the SMC will not resolve issues in which your computer is unresponsive in these situations, restarting your computer will generally suffice. If your computer isn't responding, perform these steps one at a time, in this order, until the issue has been resolved:
Force Quit (Option-Command-Escape).
Restart (Control-Command-Power).
Force Shut Down (press the power button for 10 seconds).Resetting the SMC can resolve some computer issues such as not starting up, not displaying video, sleep issues, fan noise issues, and so forth. If your computer still exhibits these types of issues after you've restarted the computer, try resetting the SMC. To reset the SMC on a Mac Pro:
From the Apple menu, choose Shut Down (or if the computer is not responding, hold the power button until it turns off).
Unplug all cables from the computer, including the power cord and any display cables.
Wait at least fifteen seconds.
Plug the power cord back in, making sure the power button is not being pressed at the time. Then reconnect your keyboard and mouse to the computer.
Press the power button to start up your computer.Another way to reset the SMC on a Mac Pro computer is:
From the Apple menu, choose Shut Down (or if the computer is not responding, hold the power button until it turns off).
Open and remove the metallic outer door.

Press the SMC_RST switch that is located slightly below and to the
right of the row of diagnostic LEDs. Use a non-metallic, non-conductive
object, such as a wooden pencil.
Replace the metallic outer door.
Press the power button to start up your computer.Mac Pro SMC Firmware Update 1.1 (http://www.apple.com/support/downloads/macprosmcfirmwareupdate11.html)

The SMC Firmware Update 1.1 adjusts fan behavior in Mac Pro.
08/08/2007 544KB

After this update has completed successfully, your SMC Version will be:
1.7f10 (Mac Pro)
1.5f3 (Mac Pro 8-core)
http://www.apple.com/support/downloads/

Mac Pro SMC Firmware Update 1.1 (#3): Problems with startup, fix; Display issues, fix; more (http://www.macfixit.com/article.php?story=20070815084737785) Problems with startup, fix Some users have reported problems starting up after applying Mac Pro SMC Firmware Update 1.1. This issue can sometimes be resolved by disconnecting all external USB devices then re-attempting startup.

MacFixIt reader Randall Calvert:
"Upon restarting, I got the white screen, the Apple logo, and the little 'clock' icon -- and then the Gray Screen of Death, 'You need to restart your computer.' This happened several times. On two of about eight tries, I got completely restarted, but then the Gray screen again. Finally I came across the instructions about possible conflicts with 'Some USB and FireWire devices.' Upon unplugging all these, I was able to restart. First time I plugged USB hub (with all devices) back in, Gray Screen again. Second time, success (so far...)."
Display issues, fix It appears that some users are experiencing issues with graphic distortion after applying the Mac Pro SMC Firmware Update 1.1. One reader writes:
"I installed the firmware update and everything was fine. the next day a portion of what was on my main monitor was actually stretched across both monitors and I could not see or reach my dock or my menu. The image of what I could see of my desk top was over zoomed and the pixels were displaying very large. I thought I had accidentally zoomed in but I had not. "
This issue can sometimes be resolved by performing a PRAM/NVRAM reset as follows: shut down your Mac, then start it back up while immediately holding the following keys: Command, Option, P and R. Hold the keys down until the computer restarts and you hear the startup sound for the third time.

Fan activity We previously reported that the Mac Pro SMC Firmware Update 1.1 -- as indicated by release notes -- modifies fan behavior for the machine. Some users noted that fans rev at very high speed for about 1 minute whenever the computer is brought out of sleep. Users are continuing to report this behavior -- which appears to be normal. It may be important to not interrupt the fans while they are throttling, as MacFixIt reader Sebastian did with problematic results:
"The update was no problem. But after my mac pro went for the first time in sleep-mode and wake up again, I recognized some strange fan behavior. The fans of the Mac Pro (2,66 GHz, 3 GB RAM, Radeon X1900XT, 500 GB Samsung HD and 160 GB Western Digital HD), not of the Radeon, spin faster and faster for 50 seconds or a bit longer depending how warm the heatsink was at wakeup. Another problem came up when I shut down the Mac Pro immediately after wake up from sleep before the the fans starting to spin faster. The Mac Pro froze, and did not shut down. I reset the Parameter RAM and push the SMC button, but the problem did not vanish. Maybe someone else got the same strange behavior after the SMC-Upgrade."

TZ
01-04-2007, 01:35 PM
Eight Core Mac Pro Benchmarks (http://www.geekpatrol.ca/2006/12/eight-core-mac-pro-benchmarks/)
Anandtech: Quad Core Intel Xeon 53xx Clovertown (http://www.anandtech.com/IT/showdoc.aspx?i=2897)
How to upgrade the processors in the Mac Pro (http://www.anandtech.com/mac/showdoc.aspx?i=2832&p=6)
Mac Pro (http://www.apple.com/macpro/)
quad-core Xeons (http://processorfinder.intel.com/details.aspx?sSpec=SL9YL)
Geekbench (http://www.geekpatrol.ca/geekbench/)
GeekPatrol - 8-core MacPro Benchmarks (http://www.geekpatrol.ca/2006/12/eight-core-mac-pro-benchmarks/)
GeekPatrol Benchmark Redux (http://www.geekpatrol.ca/2006/02/geekbench-comparison-redux/)

Comments:

As the number of processors increase, the amount of contention for common memory objects increases. Internal OS-level locking (needed to prevent corruption of these objects) can create bottlenecks that all threads have to wait for. This is why even massively multithreaded apps won't see a 2x speedup when you double the number of cores. It also can cause single-threaded apps to run slower than they would on a system with fewer cores, even at the same clock speed.

This creates a very strong rate of diminishing returns. Operating systems that can support massive numbers of processors (or cores), like some of the systems produced by IBM, SGI and Sun, are not using generic SMP tech. They need specialized code in the kernel (fine-grained locking, duplicate copies of some objects, alternate memory management schemes, etc.) to make it perform well.

I'm sure we will eventually see an 8-core Mac, but Apple is going to need to make some changes to Mac OS X (and possibly some of their apps) if they want to make efficient use of all those cores.

Quad Core Test Results summary: All test results have shown that even-though Mac OS X natively supports multi-core CPUs better than Windows, the benchmark applications won't prove so. From our point of view applications must be written with libraries supporting multi-core processors. Of course apart from what all synthetic benchmarks showed, our general impression was that Mac OS X was faster with Core2Duo processors, we hope the upcoming Mac OS X 10.5 to have better support for multi-core systems. cdrinfo (http://www.cdrinfo.com/Sections/Reviews/Specific.aspx?ArticleId=19194&PageId=6)

Upgrading CPUs of a Mac Pro to Quad Core Xeons
Dual Core Xeon CPUs of a Mac Pro 3.0GHz by Quad Core Xeon X5355 CPUs clocked at 2.66GHz.
As shown below, OSX perfectly handles doubled number of cores:
http://files.macbidouille.com/news/200702/mp8.jpg
http://www.hardmac.com/articles/70/

Intel Quad Xeon MP Platform

Advantages:
Better RAS than other platforms
Good integer performance thanks to huge L3 cache

Cons:
Expensive MP CPUs, especially compared to Xeon E5345, and very expensive platform (motherboard, memory boards.)
Pretty bad FP/rendering performance
Very high latency memory subsystem, L3 cache. (bad HPC performance)
Bad Performance/Watt, compared to Xeon E53xx and Opteron

Intel Dual Xeon Platform / Clovertown

Advantages:
Quad socket performance...
...For very low dual socket price in CRM, SAP, Financial analyses and Java server
Excellent rendering performance at high resolutions (>=720p)
In some cases, a simple upgrade for Xeon 51xx.

Cons:
Mediocre scaling in many applications
Slightly higher power consumption but little or no performance gain compared to Xeon 5160 in flow modeling, 3D rendering (lower resolutions), structural simulation, MySQL and TPC.

To the financial analysts, CRM, ERP and Java server people, the new quad core Xeon E53xx is close to irresistible. We observed at least a 40% performance increase compared to probably the best dual core CPU of today: the Xeon 5160.

You get better animation and 3D manipulation performance (mostly single threaded) and better rendering performance at resolutions lower than High Definition with the Xeon 5160. 3D render servers are better off with the Quad Xeon E53xx but only if they have to render at 720p or full HD (1080p) resolutions.

When Intel does move to 45nm proc's it will also be more efficient, ultimately moving to "point to point bus" rather than a shared bus. Intel's Penryn and Nehalem (http://www.geek.com/news/geeknews/2007Apr/bch20070406004050.htm)

Mac Pro 8-core vs 4-core:
http://barefeats.com/octopro1.html
... speculation that the 8-Core Mac Pro maybe bottlenecked by the memory bus and also considers the possibility that Mac OS X Tiger may not be well optimized for the 8-Core Mac Pros.

"My workstation was a PowerMac G5 Quad until March, 2007. It offered excellent performance for most tasks, but the Mac Pro outperforms it handily for almost everything, so long as the program being used is a Universal Binary (one that contains Intel code)."
http://diglloyd.com/diglloyd/free/MacPro/

Apple introduced an 8-core Mac Pro model today, utilizing dual 3.0GHz, Quad-Core Intel Xeon 5300 series processors in a build-to-order configuration carrying a $1498 premium over the standard configuration.

Apart from the new processor option, we didn't note any changes in prices or specifications.
http://www.apple.com/macpro/specs.html
8-core processing: Two 3.0GHz Quad-Core Intel Xeon 5300 series processors
Intel Core micro-architecture
8MB of L2 cache per processor (16MB total; each pair of cores shares 4MB)
128-bit SSE3 vector engine
64-bit data paths and registers
Energy efficiency optimization
1.33GHz, 64-bit dual independent front-side busesQuad-channel 4Gb Fibre Channel - Apple has added a new option for fibre channel cards: Apple Quad-Channel 4Gb Fibre Channel PCI Express Card, price = $999. It shows up as a new option for the Mac Pro configure-to-order.

Re: "Does anyone know if the Mac Pro can boot over Fibre Channel? I'd like to run a Mac Pro with no internal (or local) drives for even booting/applications."

At WWDC the storage project manager said that Mac Pros and Intel Xserves could boot off the 4Gb PCIE Fibre Channel card, but not off the 2 Gb card.
----------------------------
Mac Pro (8-Core): Memory and hard drive kit compatibility (http://docs.info.apple.com/article.html?artnum=305352)
Mac Pro Developer Doc (http://developer.apple.com/documentation/HardwareDrivers/Conceptual/Mac_Pro_0704/index.html)
Intel Xeon Processor 5300 Series (http://www.intel.com/products/processor/xeon/index.htm#5300) support site)
Ram Expansion Developer note (http://developer.apple.com/documentation/HardwareDrivers/Conceptual/HWTech_RAM/index.html#//apple_ref/doc/uid/TP40003031)
Thermal considerations for Mac Pro FB-DIMMs (http://developer.apple.com/technotes/tn2006/tn2156.html)
How to identify a Mac Pro (8-Core) computer (http://docs.info.apple.com/article.html?artnum=305349)
Power consumption remains unknown, but it should between 120W (Quad Core 2.66GHz Xeon 5355) and 160W

Intel's well-defined roadmap Xeon (http://www.intel.com/xeon)

"We are indeed shipping a 3.0GHz Xeon 5365 version," Bill Kircos from Intel said. "For now, the product is in limited production and Apple has chosen to adopt it. We will introduce another 3.0GHz Xeon SKU later on as well."

Preliminary Barefeats tests and benchmarks:
http://www.barefeats.com/octopro.html
http://www.barefeats.com/octopro1.html
http://www.barefeats.com/octopro3.html

Memory: The difference in the memory is that all memory modules have a little program embedded in them. Apple changed the program in the 8 core memory modules to make them play nicely with the new 8 core processor. Also, the new 8 core memory module has a larger heat sink.

Hard Drives: Notice that Apple uses the word "Kit" in the Knowledge Base Article. A "Kit" consists of a hard drive mounting sled and a SATA II hard drive. The sleds from the quad core Mac Pro are larger than the sleds in the 8 core Mac Pro. The smaller 8 core Mac Pro sled allows for greater air circulation, hence better cooling.

(http://www.xlr8yourmac.com/systems/8-core_mac_pro/8-core_Mac_Pro_reports.html)
8-core Mac Pro versus 4-core Mac Pro (http://www.barefeats.com/octopro3.html)

Normally the CS3 action file consumes as many cores as you have to give. We saw 796% usage on the 8-core when Photoshop was the only busy application. This action file normally takes only 37 seconds to complete on *BOTH* the 8-core and 4-core Mac Pro. In other words, it's a tie if Photoshop is the only active application.

But if three other CPU hungry apps are fighting it for CPU cycles, we see a whole different picture...

CONCLUSION If your work flow means doing more than one thing at a time on your Mac Pro, then you will see significant gains if you spend extra to get the 8-core version. Our Photoshop CS3 actions were completed 39% faster on the 8-core when we had 3 other apps busy crunching. This advantage emerges in spite of the memory bus limitations of the 8-core Mac Pro.

TZ
01-05-2007, 07:13 AM
Hardware Monitor is an application to read out all available hardware sensors in Macintosh computers. The program can display and visualize measured values in a large variety of fashions. It can also store and export readings.

The application Temperature Monitor is available for free, but is limited to accessing temperature sensors only. The application Hardware Monitor is commercial shareware, but can access additional sensors if your computer is equipped with them. This includes the following sensor types:

http://www.macupdate.com/info.php/id/16609
Version 4.2:
Added support for a large number of voltage, current (amperage), power, and light sensors for Intel-based Macintosh systems.
Added full support for the new Apple Xserve Quad Xeon.
Added support for S.M.A.R.T. temperature sensors in SATA drives connected via an SAS bus.
Removed support for the internal Intel CPU sensor monitoring the target value of the core voltage. This is now superseded by the voltage sensors monitoring the actual core voltage supply.

Added new "disk saver" preference which allows users to stop monitoring of internal hard drive sensors when the system is idle. Idle drives can now enter sleep mode and be held in sleep mode, independently of S.M.A.R.T. monitor time interval settings.

Almost a must to monitor FB-DIMMs, Risers, PCIe slots and PSU fans and temperatures. Add smcFanControl (http://www.macupdate.com/info.php/id/23049) and you are good to go!

TZ
01-06-2007, 11:43 AM
Sun recently ended an AMD-exclusive x86 server relationship, partnering with Intel to produce a new line of Xeon servers. In exchange, Intel will promote Solaris as a mainstream x86 server operating system (see our coverage (http://www.geek.com/news/geeknews/2007Jan/bch20070122001754.htm)).
Sun admits Intel pays for Xeon server engineering (http://www.geek.com/news/geeknews/2007Feb/bch20070214002609.htm)


Virtual realities: all the world on a Mac (http://www.computerworld.com/action/article.do?command=viewArticleBasic&articleId=9011577&intsrc=hm_list)
Mac virtualization: VMware and Parallels side by side (http://www.computerworld.com/action/article.do?command=viewArticleBasic&articleId=9011579&intsrc=hm_list)


45nm-Engraved Xeon in 2007 based on the Penryn core will be available into production for the second half of 2007." Intel's latest desktop guidance claims 45nm desktop SKUs will also launch in late 2007, with volume shipments occurring in 2008. As it stands right now, only the mobile 45nm SKUs are expected to launch in 2008.

Skaugen also confirmed that Penryn-based Xeon processors will utilize the same server platform as Xeon 5000, 5100 and 5300. Nehalem, Intel's next-generation micro architecture on the 45nm node slated for 2008, will require new platform technology and is not compatible with the Penryn platform.

45nm quad-core Harpertown and dual-core Wolfdale were originally slated to spearhead the next-generation Xeon launch in Q1 2008. The existing Bensley platform, Intel 5000P chipset, will still provide the heavy lifting for volume dual-socket on 45nm Xeon. A new platform, Cranberry Lake, will replace Bensley-VS for value dual-socket Intel platforms, and will support Harpertown and Wolfdale.

So, price of the current Quad Core Xeon might quickly decrease and while higher clocked models should appear.
_______________________
Future Xeons, Itaniums to share a common platform
Cyril Kowaliski- 02:05 am, February 27, 2007

During an interview with ZDNet, Intel VP and Digital Enterprise Group co-manager Pat Gelsinger revealed (http://news.zdnet.co.uk/emergingtech/0,1000000183,39286049-1,00.htm?r=1) that future Xeon and Itanium processors will eventually share common platform and architectural elements.

According to DailyTech (http://www.dailytech.com/Intel+Moves+Towards+Itanium+and+Xeon+Convergence/article6236.htm), convergence between the Xeon and Itanium lines will be aided by Intel's Common System Interface, which will be the company's answer to AMD's HyperTransport interconnect. The site says both Tukwila and upcoming "Tigerton" Xeons, which it says are due later this year, will use CSI.

Crucial FB-DIMMs Xserve (http://www.crucial.com/store/listparts.aspx?model=Xserve+%28FBDIMM%29)

According to DigiTimes, Intel plans to introduce a set of "Tigerton" processors intended for four-way servers some time at the start of September.

Tigerton CPUs will essentially be spinoffs of Intel's Core-based Xeons tweaked for operation in 4P systems.

DigiTimes says the Tigerton lineup will include quad-core models clocked between 1.86GHz and 2.93GHz with either 4MB or 8MB of cache as well as dual-core models running at 2.4GHz and 2.93GHz with 8MB of cache.

Prices will range from $856 for the cheapest dual-core model to a whopping $2,301 for the 2.93GHz quad-core part.

Today, Intel's only 4P-capable parts are still based on the old Netburst architecture.

TZ
01-16-2007, 09:51 AM
Intel updates compilers for multicore era (http://arstechnica.com/news.ars/post/20070605-intel-updates-compilers-for-multicore-era.html)
By Jon Stokes (http://arstechnica.com/authors.ars/hannibal)

Intel has announced major updates for its C++ and Fortran tools, updates that are aimed at making it easier for programmers to exploit thread-level and data-level parallelism in multicore processors.

On the data parallelism side, the Intel C++ Compiler and Fortran Professional Editions both sport improved auto-vectorization features that can target Intel's new SSE4 extensions.

For thread-level parallelism, the compilers support the use of Intel's Thread Building Blocks for automatic thread-level optimization that takes place simultaneously with auto-vectorization.

The version 10.0 compilers are actually the first from Intel that can do both auto-vectorization and thread-level optimization simultaneously in one pass.
"Combining these capabilities in a cooperative and coordinated manner brings our customers added performance benefits and compiler reliability, which in turn gives greater PC responsiveness for gamers and everyday PC users," said Kevin J. Smith, director of the Intel Compiler Products.

Intel is encouraging the widespread use of its Intel Threading Tools (http://www.intel.com/support/performancetools/tbb/sb/CS-023131.htm) as an interface to its multicore processors. As the company raises the core count with each generation of new products, it will get harder and harder for programmers to manage the complexity associated with all of that available parallelism.

The Thread Building Blocks are Intel's attempt to insert a stable layer of abstraction between the programmer and the processor so that code scales less painfully with the number of cores.

Cache and memory in the many-core era (http://arstechnica.com/articles/paedia/cpu/cache-mem-many-core.ars)

As CPUs gain more cores, resource management becomes a critical performance challenge. Here's how Intel is addressing the problem of resource contention for our multi-core, virtualized future.

Multi-core overhaul
Converting single-threaded software to be multi-threaded is a highly specialised skill and a challenge for developers. The extensions initially focus on performance analysis via the company's new Light-weight profiling (pdf) (http://developer.amd.com/assets/HardwareExtensionsforLightweightProfilingPublic200 70720.pdf) specification. This technology will enable real-time feedback for performance optimisation that can be directly used by software, said Earl Stahl, vice president of software engineering at AMD.

"Developers will incorporate [the technology] into the software, and the software can utilise it to at runtime to discover, for example, cache contention with their memory allocator and reallocate that memory," Stahl said.

Over time, other hardware extensions could be added in such areas as software transactional memory, high-performance message-passing, and fast context-switching for lightweight parallelism. AMD plans to offer the full gamut of specifications released under the plan up for public review.

Software that would leverage the extensions could be applications based on platforms like Microsoft's Common Language Runtime or the Java runtime, Stahl said. Access to the extensions by developers could be done either through compilers or via a small API layer.

The extensions, however, will not be in AMD chips for an estimated three to four years. They will not be in the upcoming AMD Barcelona and Bulldozer chips.

Intel has focused on software tools to help with multi-threaded development, while AMD is offering support in hardware itself, he said. http://www.techworld.com/opsys/news/index.cfm?newsid=9793 (http://www.techworld.com/opsys/news/index.cfm?newsid=9793)Intel readies research papers (http://eetimes.com/news/latest/showArticle.jhtml?articleID=201800194) on programmable multicore architectures
provide details on how the company expects future microprocessors with simplified parallel programming models to evolve.

What's different in developing software for multicore environments is the need for parallel programming, which is the divvying up of tasks from an application among multiple processors, and having them perform the work simultaneously. The complexity of such an environment requires different development tools than the ones typically used today.

Intel is proposing the in-hardware implementation of a function (http://www.techweb.com/encyclopedia/defineterm.jhtml?term=function&x=&y=) called task scheduling, which is the mapping of work to cores for execution. The software-based methods used today introduce too much overhead for use in highly parallel workloads.

Multicore Processors Are Reshaping Computing (http://www.informationweek.com/shared/printableArticle.jhtml?articleID=198701707)
"The bus-based multicore system will fade in the next year or two,"

TZ
02-05-2007, 04:33 PM
An answer to Charlie's quad-core challenge (http://www.geek.com/news/geeknews/2007Feb/bch20070205002150.htm) -
Charlie Demerjian at The Inquirer and I go way back ... back to the heady days of discussions and post and debates on a website called Real World Technologies ... back to a time when my primary software background coupled to an extreme interest in all things x86-based in hardware was found running up against his primary hardware background with an almost indiscernible interest in software. Those backgrounds used to come to loggerheads over issues of theory, not substance. I'm not sure if either of us won any of those arguments, though I'm sure we both have positions on that question. But the truth is I've always respected Charlie, and his articles have conveyed much that would probably otherwise go missing in the semiconductor world were it not for his interests and his reporting the same (the same is also true for Mike Magee). As such, in response to an article he published last Thursday, I thought I would write a brief retort.

In Definition of true quad cores proposed (http://theinquirer.net/default.aspx?article=37361), Charlie poses the question,


"What is the true definition of a quad core processor?"
He goes on to boil down his position to the following:
First, it must not be able to have functional sub-cores taken out of it, such as pulling two Woodcrests out of a Clovertown.

Second, it must be physically able to communicate from core to core without hitting the main bus, regardless of whether or not it actually does so.Okay, that's a pretty straightforward definition, and is one that makes all quad-core offerings available today "untrue" quads. Still, I believe another facet needs to be considered.

The reality is this: what is a processor? Is it a physical package? Is it the die that used to exist only as a single-core model within the package? Does a Hyper-Threaded model count as two separate cores when they are provided to software logically? In short, does physical implementation or logical representation define the characteristics of a quad-core?

I would like to offer the following criteria for meeting the definition of being truly quad-core:

First, does the software recognize it as quad-core?

Second, when operating on all cores, is the maximum theoretical computing performance four times that which would be seen on a single core? This, of course, refers to computational speed and not throughput onto a shared bus for memory requests, but rather simulated, artificial workloads that deal specifically with code and data that can be stored entirely in the L1 cache for maximum computational prowess.

If the answer to both of those questions is yes, you have a quad-core processor. And the same would be true at dual-core, oct-core, and so on.

See, as someone with primarily a software background I do not see the physical implementation as being anything of real substance, apart from the actual increases in performance a truly hard-wired (by Charlie's definition) quad-core would be. To software that's of no consequence. The fact of the matter is that what we see as x86 opcodes and exposed abilities from within the core are not always operating at a 1:1 ratio with what happens inside the core. The core hides much of the limitations found in the instruction set architecture, for example, and operate in manners that software designers would probably truly hate were they to find out more about them.

As such, the definition of being quad-core should not, in my opinion, mandate that it be a physical implementation of any particuluar kind, but rather be visible to software as a quad-core, coupled with the commensurate maximum theoretical performance achievable, thereby discounting dual Hyper-Threaded processors as quad-cores even though four visible cores are exposed.Anyone care to offer an alternate definition? - RickGeek
(February 5, 2007 - 2:53pm EST)
_________________________
A good read:
http://techreport.com/articles.x/13224

What is 'quad-core' mean? physical? logical?
http://www.geek.com/an-answer-to-charlies-quad-core-challenge/

Penryn testbed:
http://www.legitreviews.com/article/521/1/

Intel IDF conference:
http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3105

TZ
02-12-2007, 04:11 PM
I can't believe though that MaxMenus and MagicMenu brought my computer to a crawl. This thing was dying! What's in these programs that would allow such a thing to occur?

I found that it is far too easy to run into a problem with pesky background processes (Desk Accessories and Inits are still in the building along with 3rd party extensions).

TZ
02-16-2007, 06:37 AM
* got a stock mac pro 2.66 with a motu 828 (original) and it mysteriously loses sound output without warning. Logic still sees it and acts as if nothing's wrong - but no sound until I reboot. I just read a handful of similar accounts on xlr8yourmac.com

* same boat, brand new Mac Pro (x1900 video card!) 30" ACD and a MOTU 828 mk 1 going along fine and the sound will drop out, iTunes stops, the progress bar stops, Logic plays fine but no sound. Tried resetting the clock source and the changed the sampling frequency to 48000. Also tried with a friends Traveler and encountered the same problems...

* I have loss of sound, too. MOTU 828mkII.

Happened with a clean 10.4.8 install with only logic pro and motu drivers installed and with two different 828mkII interfaces (no other devices connected).

Most times it happens when I close an application with MOTU sound output. Happens with Logic, games, Quicktime etc. It never happend during work with an application, only when closing it.

MOTU MIDI I/O is disappearing, too when ist happens.

Restart is the only way to get MOTU output back. No power cycling, unloading/loading kernel extensions, etc.

* Motu put me thru standard troubleshooting and said it appears to be a problem with my MacPro, but I believe its a problem with the Apple firewire audio driver or something like that.

There are some posts at www.xlr8yourmac.com (http://www.xlr8yourmac.com)

* a similar problem using the MOTU 828 on my Mac Pro with 2GB of Apple RAM and OS 10.4.8. Every once in a while, I lose the audio output (the 828 is the system output device). Even weirder, once when I was playing back multiple tracks in Logic Express, one track stopped playing but the other continued! Unlike some people, I have always been able to fix the problem by power cycling the 828, even when the application generating the audio is running. When the 828 comes back up, the sound is back on.

After a brief email conversation with MOTU tech support, they wanted me to send my unit in for repair. It seems to be a Motu thing, but when i call motu they say there are no probs

* Are you using your FW interface as main audio output, or only in apps like Logic, GarageBand or Cubase ? I use it as main output.

* Are you using Entourage (under Rosetta) ? I use it and, each time I see that I've lost the sound on MOTU 828, when I switch on another output, only Entourage sounds are played from the buffer.

* some news : one person I know has the same problem with a RME interface. His Mac Pro is older than mine and, as I only know OS X 10.4.8 on my new Mac, he used his Mac Pro with 10.4.7. And he never had a problem with the audio via FireWire before the 10.4.8 update.

* So it seems that the problem is not with the interface, neither with the driver, neither with the hardware (FireWire bus).

* I have exactly the same problem with my 828 mkII. Audio via FireWire is lost after several hours. When I switch to another output, all the sounds kept in the buffer (from Entourage) play. So I can have an idea of the duration of lost output.

And I have to restart (with MOTU unplugged) to recover the FW audio output. Because reset on the MOTU doesn't work. The FireWire must be resetted.

* I thought it was because I installed Wiretap Pro, a good Ambrosia app which had problems with FireWire audio management (I helped them for that). Wiretap is not native. Wiretap uses an extension kept in the system folder. So I disinstalled it to see if, without the extension, the audio output would be kept.

* i've had the same problem with my tascam fw-1082... i just use the fw-1082 for recording. what happens is that it loses firewire connectivity arbitrarily, usually
for about 2-3 seconds. it has caused quite a few problems with recording,

i have too a motu 828 mk2 connected to my macpro,and in osx have no problem at all, very stable.Instead under xp (bootcamp) sometime (not often)happens your same problem, the audio is cut for 1-2 seconds.

* using a Motu 828 Mk II. Well since the new memory, my sound keeps cutting off every once in a while. The computer still recognizes the Motu as the sound device (through firewire), but no sound, and I have to reboot to get it back.



Versions of Logic prior to 7.2.3 are not compatible with the built-in audio hardware on the Mac Pro — they do not allow the outputs to be assigned in Logic's Audio > Audio Hardware & Drivers window.
http://a248.e.akamai.net/7/248/51/3043901180996615/www.info.apple.com/images/kbase/304390/304390_1.jpg
Logic Pro and Logic Express 7.2.3 provides access to all three outputs available on the Mac Pro:

Built-in Output (the speaker)
Built-in Line Output (the line out jack on the back of the Mac Pro)
Built-in Digital Output (the optical audio output on the back of the Mac Pro)
http://a248.e.akamai.net/7/248/51/3043901180996615/www.info.apple.com/images/kbase/304390/304390_2.jpg
http://docs.info.apple.com/article.html?artnum=304390

TZ
04-06-2007, 02:19 PM
Intel's Nehalem to feature eight cores, 16 threads
by Scott Wasson (damage@techreport.com) — 1:13 PM on September 18, 2007

Intel Developer Forum — Nehalem is now complete, and Otellini claimed it's on track for delivery in the second half of 2008.
Chief architect Glenn Hinton showed a working Nehalem-based system, just three weeks old, running Windows XP. Otellini also claimed the team managed to get Mac OS X booting just today.

Nehalem chip will be able to execute 16 threads at once.
Nehalem will integrate a high-performance memory controller and a new chip-to-chip interconnect known as QuickPath—both provisions similar to AMD's Opteron processors. Intel plans to release processors based on its upcoming "Nehalem" is scheduled for a launch in the fourth quarter of 2008. Desktop Nehalem variants, which are dubbed Bloomfield.

Nehalem will bring (http://techreport.com/discussions.x/12130) a brand new, 45nm-based architecture packing features like an integrated memory controller, point-to-point interconnects à la HyperTransport, a built-in graphics core, simultaneous multi-threading, and more. Intel will introduce (http://www.fudzilla.com/index.php?option=com_content&task=view&id=3757&Itemid=1) a new LGA1366 socket, Tylersburg-DT chipset, and ICH10 south bridge along with desktop Nehalem processors.

Nehalem CPUs due in 2008 will be dual- and quad-core only with no integrated graphics controllers. Models with eight cores and integrated graphics are said to have slipped into 2009.

In addition to the release of Penryn chips for servers later this year, Intel is expected to release the Yorkfield and Wolfdale desktop variants of Penryn in early 2008, with some speculating that high-end versions of these desktop chips could arrive before the end of this year.

Intel will release the high-end 1666MHz FSB Xeon models for Q3 2007 with a TBD of 80W, and will come along with a new chipset named Seaburg.

Intel's Penryn and Nehalem (http://www.geek.com/news/geeknews/2007Apr/bch20070406004050.htm)
http://www.intel.com/technology/quad-core/server/index.htm

Penryn is already ticking away at 3.0GHz (http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=2972)The first Penryn processors come Nove 12th. Intel makes Penryn samples available early - Intel will offer manufacturers samples of its Penryn server chips before the planned launch later this year.

Wolfdale is a dual-core desktop Penryn core with a 6MB L2 cache shared between the two cores. Yorkfield is two Wolfdales on the same package, giving the quad-core CPU a total of 12MB of L2 cache. Wolfdale is faster clock-for-clock, you won't see Wolfdale until Q1 of next year and the performance advantage simply isn't great enough to justify delaying a purchase by 6+ months. AnandTech (http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3069)

Several 32-bit and 64-bit DIVIDE operations on Penryn will be notably faster than on Intel's current x86 processors. The 32-bit, 64-bit, and even 80-bit floating point square root computations will all be about 66% faster, with the 80-bit version being the most improved. All of this is thanks to a new technology Penryn brings to the table called "Radix-16."

Penryn Virtualization will be greatly improved over the Core architecture, on the order of 25-75% faster on enter/exit calls from the OS to the HyperVisor layer--something that's key when in virtualization mode. Shared caches in the dual-core Penryns will be 6 MB, and quad-core beasties will see a healthy 12 MB of shared cache. The associativity will (most likely) also be 24-way, up from current 16-way implementations for L2 access, a 50% increase.

Penryn 3.2GHz with 1.67GHz FSB will outperform an existing 3.0GHz Conroe (sitting atop a 1.33GHz FSB) by about 20%. That's about what you would expect from the increase in clock and FSB speeds, however, and that was only on gaming. Some of the biggest performance gains may not be seen in desktop spaces, however, but rather on servers with heavy use of FP.

Penryn represents the most advanced CPU Intel has released since the Pentium Pro. It may exceed that release, however, because not only will it leverage a high-k 45 nm manufacturing process with the potential of reducing leakage by 1,000% over current models, but it will also incorporate redesigned core logic components that will reduce execution latency and increase throughput overall--even on existing applications without any code changes whatsoever.

Intel Peyrin 3.33GHz LGA775 45NM PENRYN BENCHMARKS (http://www.legitreviews.com/article/521/1/)
is ready to be sent away to system integrators as early as Q3 2007.

Anandtech (http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=2955)Intel Xeon 5300 processors (http://www.intel.com/products/processor/xeon/index.htm#5300)
Penryn Performance Review - anandtech (http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=2972)
Penryn preview probes performance, power performance preview (http://www.hkepc.com/bbs/hwdb.php?tid=837360&tp=Intel-Penryn-Preview&rid=837360) (in Chinese, translate here (http://babelfish.altavista.com/)) of what is apparently a 45nm "Wolfdale" Core 2 Duo running at 2.33GHz. Most notable is a nearly 116% jump in encoding performance with an SSE4-optimized alpha of DivX 6.6. Gains in other apps aren't nearly as striking, but they're consistent across the board and often eclipse 10%.

According to Intel executive, Penryn-based computers will see the following speed improvements.

- 15% for imaging-related
- 25% for 3-D rendering
- 40% for gaming
- 40% for video encoding (SSE4 optimized)
- 45% for "bandwidth intensive applications
- 25% for servers using Java

3.33 GHz
1600 MHz FSB.
dual-core 6 MB L2 cache
quad-core 12 MB L2 cache

Penryn chips are expected to be released later in the fourth quarter of 2007 with quad-core "Harpertown" Xeons based on 45nm process technology and running at speeds of up to 3.16GHz.
http://blogs.intel.com/technology/2007/04/penryn.html

Harpertown Xeons launch on November 11 seven quad-core 45nm. The chip's power efficiency suggest that the new core runs cooler and consumes less power than the E6550 at both idle and under load. In fact, the Wolfdale chip actually has lower temperatures and power consumption under load than the E6550 does at idle.

Harpertown will have a thermal rating of 120W, but Intel is reportedly cooking up low-voltage variants clocked at 2.33GHz and 2.66GHz with power ratings of just 50W. All quad-core Harpertowns will pack 12MB of L2 cache —6MB more than current Xeon 5300-series "Clovertown" models.

Nehalem 45nm microarchitecture to motherboard makers in October with an eye towards release in the second half of 2008 . The next Xeon revision will be an evolution of the current models (based on Penryn), will sport numerous improvements.


Xeon Nehalem will be true quad core CPU and not two dual core associated on the same die. Cores will not only share their cache memory, but the entire cache system. Among other features:
- Leverages 4-issue Intel Core micro architecture technology
- Simultaneous multithreading (SMT)
- Multi-level shared cache architecture
- Performance-enhanced dynamic power management
- Fully unlocks Intel 45nm Hi-K silicon process benefits

"Wolfdale-DP"For server users not keen to make the jump to quad-core, VR-Zone's roadmap says Intel will also have dual-core "Wolfdale-DP" chips clocked at 3.33GHz with thermal ratings of up to 65W.
http://www.vr-zone.com/?i=5116


Tylersburg technology will sport QuickPath technology, Intel's new uniform bus previously known as Common System Interface , which should dramatically speed up communication between components.
Intel's new Common Systems Interconnect (CSI)
- late next year or in early 2009

The initial 65nm and 45nm implementations of CSI are aiming for 12-16GB/s of bandwidth each way, so that's 24-32GB/s per bi-directional link.


CSI, like HyperTransport, is a point-to-point interconnect for multisocket systems that can support an on-die memory controller. When Intel moves to on-die controllers and NUMA in the server market, this will immediately provide a dramatic boost to Xeon systems' aggregate bandwidth—possibly on the order of 8x to 10x. This greatly increased system bandwidth, along with Intel's projected move away from power-hungry FB-DIMMs, will help with Xeon systems' current power/performance deficit versus AMD's Opteron. CSI's impact on the mainstream desktop and mobile markets will probably be minimal.The long-overdue bus bandwidth boost will have effects on the design of Intel's processors themselves, since the company has been cranking up the cache sizes in both its Xeon and Itanium parts in order to relieve system-level bandwidth starvation.

TZ
04-20-2007, 04:58 AM
One of AMUG reviews shows using Disk Utility (http://www.amug.org/amug-web/html/amug/reviews/articles/firmtek/2eEN4/) to first partition a drive, then use the first half only to use in a RAID. Short-stroking the drive platters like this insures that your single partitioned drive RAID will use only the better half of the drive and get maximum performance.

The ExpressWay FireWire800/1394b and USB v2.0/1.1 Combo PCIe Host Adapter is $99.95.

WD Firmware
To updated the firmware on my wd5000YS. I checked the TLER status after the firmware update with WDTLER and TLER is still disabled (both read and write) which is unchanged. New firmware version 09.02E09

they list the following models only:
* WD1600YS
* WD2500YS
* WD4000YS
* WD5000YS

If running RAID, you definitely need to apply this firmware update.

http://www.theinquirer.org/default.aspx?article=37188
WD Support firmware (http://support.wdc.com/download/index.asp?cxml=n&pid=15&swid=57)

EnhanceBox E8-ML SATA Hot Swap Multilane 8-Bay HD enclosure coupled to a HighPoint RocketRAID 2322 to create a very fast RAID 5 setup for the Apple Mac Pro.
Performance is simply amazing: read and write performance of 420MB/sec while redundancy provided by RAID 5 was successfully tested.
Add an Eight Bay RAID 5 to the Apple Mac Pro (http://www.amug.org/amug-web/html/amug/reviews/articles/enhance/e8/)

Belkin F5U602 PCI-Express FW card, FW/SCSI adapter

Fry's has Belkin F5U602 PCI Express card. I bought it and popped it in anyway. To my delight, it works just fine on my fully updated 10.4.8 system. I plugged a LaCie drive into it and did some read wrties and all worked well. I now have a SCSI to FW adapter, plugged into it running out to a Sony DAT DDS drive (ripping audio from a flashed Sony 9000). Previously I could not run that device AND a Firewire hard drive at the same time or bad things happened. Now I have it running to the card, and on the onboard FW800 port a daisy chain of 2 OWC RAIDs (Mercury Elites) and 2 OWC Mercury Elite Classics. Everything smooth so far. Apple System Profiler shows it only as a PCI to PCI bridge, with no mention of Belkin, etc.

Belkin FW800/USB2 combo card. If I end up getting that.


<table border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr valign="top"><td width="97%">RAID 6 for the Mac Pro
: Jul 6, 2007 7:46 AM </td> <td class="jive-rating-buttons" nowrap="nowrap" width="1%"> </td> <td width="1%"> <table border="0" cellpadding="0" cellspacing="0"> <tbody><tr> <td>
</td> <td class="jive-icon">
</td> <td class="jive-icon-label">
</td> <td class="jive-icon">
</td> <td class="jive-icon-label">
</td> </tr> </tbody></table>
</td> </tr> <tr> <td colspan="4" class="themessage">
The Areca ARC-1221x can provide over 350MB/sec average read and write performance while configured with RAID 6 using eight hard drives. RAID users that want crash prorection will crave RAID 6. AMUG tested the Areca ARC-1221x while simulating two hard disk failures. The RAID 6 configuration was able to easily rebuild the eight drive RAID set.
http://www.amug.org/amug-web/html/amug/reviews/articles/areca/1221x/
</td></tr></tbody></table>
AMUG review Sonnet Tempo E4P Mac Pro driver 2.1
AMUG: Mac Pro w/ Sonnet Port Multiplier (http://www.amug.org/amug-web/html/amug/reviews/articles/sonnet/mac-pro/)

16 drives provided
730MB/sec Read rates
600MB/sec. Write rates

Using two E4P cards with 20 drives and 5 SATA PM enclosures.
1060MB/sec Read rates
870MB/sec Write performance

That is enough speed to support 4 HD streams at once. The Sonnet Tempo E4P passes SMART data to Mac OS X so that users can monitor hard drive temperatures and check the SMART status of their hard drives. Those users that would like the capability to create large striped RAID sets may find this article interesting.

WD Caviar RE2 500GB WD5000YS
Hitachi T7K500
Maxtor MaxLine Pro 7H500F0
Maxtor MaxLine III 7V320F0
____________

Mac Pro gets RAID card CTO Option -
there's a new (internal) RAID card option at the Apple store for Mac Pros.

" While the front page of the Apple Store makes no indication in the Mac Pro line up, Apple did manage to sneak in a PCI-E RAID card option. (costs $999) Of note it supports RAID5 and 256 MB of cache with a 72 hour (cache) back up battery.

Apple Mac Pro Expansion page shows/mentions the card now, as does the Mac Pro Specs page
[I]"Optional Mac Pro RAID card with 256MB cache and 72-hour cache battery backup - Supports Mac OS X only."The Expansion page includes some benchmarks with the RAID card and says:
With 256MB of RAID cache, a 72-hour cache protecting battery and hardware RAID levels 0, 1, 5 and 0+1, the Mac Pro RAID card delivers sequential read performance up to 304MB/s in RAID 0 and up to 199MB/s in RAID 5 -- without the need for any external drive enclosures, power bricks, or cables. And of course Apple's RAID Utility software makes setting up and managing the RAID card simple.
[/URL]
Areca ARC-1680x at this link:
http://www.areca.com.tw/products/pcietosas1680series.htm (http://www.scsisource.com/sas_controllers/)

The Areca ARC-1680x is a PCIe 8x card that utilizes the Intel 800 MHz IOP348 I/O processor with SAS controller. It includes 256MB of on board memory and supports an optional battery backup module. The card has two external miniSAS connectors that can be used with miniSAS to Multilane cables or miniSAS to eSATA cables.

The ARC-1680x is $845 instead of $1199 for the ATTO card and it supports more RAID options including RAID 6. [URL]http://www.scsisource.com/sas_controllers/

TZ
04-23-2007, 08:01 AM
4 Port PCIe Host Cards (http://www.macgurus.com/productpages/sata/PCIeSATAHostCards.php)

CalDigit FASTA-4e PCIe SATAII Host Card

New CalDigit FASTA-4e 4 External Port SATAII host card. Your choice of card for the MacPro! ALso compatible with Dual Core G5s and PCs. This speed demon is port multiplier capable making it the choice for speed, versatility and expandability. Able to attach drives either directly for up to 4 external SATA drives or in Burly Port Multiplier Enclosures for up to 20 drive capacity. Expandibility is its thing. This card sets the bar way up there for any competition. $288.00

Sonnet Tempo E4P PCIe SATAII Host Card

Tempo E4P with 4 External SATAII Ports, Port Multiplication capable and superior performance. If you are going to pick a great first product to be released for PCI-Express G5 Macs and PCs, this is the one to choose. Using Port Multiplication Enclosures from MacGurus you can attach up to 20 hard drives to this host card giving awesome capacities and never before seen speeds. Attach drives either directly or through one or more port multipliers, or a combination of both. This card will be hard to match. We at MacGurus couldn’t be happier that this card hits first, a solution for everyone here with speed, capacity and versatility unmatched. $288.00

2 Port Hosts

CalDigit FASTA-2e PCIe SATAII Host Card

New CalDigit FASTA-2e 2 External Port SATAII host card. Works with the MacPro! Also compatible with Dual Core G5s and PCs. $75.00

Host Card Type (http://www.macgurus.com/guides/hostcardtype.php)

Problem with Seritek 2SE2-E eSATA card in 8-core MacPro
Updated - he later wrote he had used a clone of his previous Mac Pro boot drive (dual core CPU model) the card is now recognized:


I have a new MacPro 8-core and the Seritek 2SE2-E. The card has firmware 5.2.0 (nov 18 2006) and cannot get recognized by the computer...

With the application "Expansion Slot Utility" I can see the board as "unknown other mass storage controller" but anyway it's not usable by the computer.

The card is perfectly ok with a previous 3GHz 2x2 (dual core) MacPro. But news of the morning is that it seems a totally software problem.

Starting from a clone of my old MacPro (dual core) disk installed in the new quad core, the card works! If I use a fresh newly installed with factory disk, I cannot get the card working. The kernel and OS X build can be confirmed using sw_vers that it is indeed later than 10.4.9 (and there are some problems with the 2007-004 Security Update).

NitroAV (http://www.nitroav.com) introduced the ExpressWay FireWire800/1394b and USB v2.0/1.1 Combo PCIe Host Adapter. (http://www.nitroav.com/product/363/)

The PCI-Express card is compatible with the PCIe-based PowerMac G5 and the Mac Pro, among others; provides two external FireWire 800 ports, three external USB 2.0/1.1 ports, one internal FireWire 800 port, and one USB internal port.

PCI Express Architecture

The Mac Pro has four internal, 2.5 GHz, PCI Express links connected to the North Bridge IC and South Bridge IC. The PCI Express slots are system and user configurable. The Mac Pros standard configuration is one 16-lane, double-wide graphics slot, two 4-lane expansion slots, and one 1-lane expansion slot.

For information on PCI Express and the configuration options, refer to PCI Developer Note (http://developer.apple.com/documentation/Hardware/Conceptual/HWtech_PCI/index.html#//apple_ref/doc/uid/TP40003027).

PCI Express 3.0 to push 8GT/s
by Geoff Gasior (dissonance@techreport.com) - 01:20 pm, August 9, 2007
PCI-SIG has announced (http://www.pcisig.com/news_room/08_08_07/) that the next, next-gen PCI Express 3.0 spec will offer a maximum bit rate of 8GT/s—double that of PCI Express 2.0. Interestingly, PCI-SIG considered pushing a 10GT/s maximum bit rate, but settled on a slightly slower speed, in part to maintain backwards compatibility.
“Experts in the PCIe Electrical Workgroup analyzed both 10GT/s and 8GT/s as target bit rates for the next generation of PCIe architecture, and after careful consideration of several factors, including power, implementation complexity and silicon area, recommended 8GT/s,” said Al Yanes, PCI-SIG chairman.

“This allows us to satisfy the next generation performance requirements for all existing PCIe applications while maintaining backward compatibility, and at the same time broadening the adoption of this pervasive technology into new and emerging applications and usage models.” Along with additional bandwidth, PCI Express 3.0 adds support for "transmitter and receiver equalization, PLL improvements, clock data recovery, and channel enhancements for currently supported topologies." Don't expect to see PCIe 3.0 anytime soon, though. Core logic chipsets with PCI Express 2.0 support aren't due until the fall, and it could be 2010 before products built for PCIe 3.0 become available. Thanks to Xbit Labs (http://www.xbitlabs.com/news/other/display/20070808225423.html) for the tip.

TZ
04-27-2007, 06:36 AM
Intel V8 platform. If you're prone to speculation, you'll be pleased to hear Intel has concocted an answer to that very question in the form of its "V8" media creation platform. V8 is Intel's workstation/server-class technology to take desktop PCs to new heights. If you want to grab a slice of the future now, with eight cores of glory at your disposal, you're going to have to pay a pretty penny for it. Happily, though, we've tested a V8 system against a slew of today's best desktop processors, and we can give you a glimpse of how the future may look, free of charge. Here's a hint: it's ridiculously fast.

1333MHz Greencreek FSB chipset and 8 FB-DIMM memory slots
800-833MHz memory using DDR3, up from 667MHz
3.3GHz Quad Core Xeon X5365 (Clovertown)
Radeon X2800/2900 XT
One or two 16x PCIe slots
EATX mainboard format
6 storage connectors, as either 6 x SATA3G or 2xSATA plus 4xSAS (serial SCSI for those funky pro drive arrays)
8 FB-DIMM sockets for up to 32 GB RAM.
One PCI-E x16 slot is there for your 3-D graphics card
two PCI-E x4 (for extra 2-D graphics or interconnects)
two 64-bit PCI-X slots
Dual Gigabit, basic sound and plenty of fan connectors
_____________________
The Processor Arithmetic and Multimedia benchmark scores are more than double those of a Core 2 Duo QX6800.

Memory Bandwidth benchmark reported maximum bandwidth scores in the neighborhood of 4GB/s, which is low by today's standards and about 1GB/s off the mark set by the QX6800.

3DMark06 test 13,002 (SM2.0=5,104 / HDR & SM3.0=4,932)
multi-threaded Cinebench v9.5 benchmark CPU score was 6,556
- almost 500 points higher than the rig Intel was showing off at CES.
completed the Cinebench rendering pass in only 10 seconds, which is about 4-5 seconds faster then a quad-core Core 2 Extreme QX6800.

Intel Peyrin 3.33GHz LGA775 45NM PENRYN BENCHMARKS (http://www.legitreviews.com/article/521/1/) is ready to be sent away to system integrators as early as Q3 2007.
The Inquirer (http://uk.theinquirer.net/?article=39175)
HotHardware: "8Core_Intel_Xeon_V8_Test_Rig" (http://www.hothardware.com/News/8Core_Intel_Xeon_V8_Test_Rig__Sneak_Peek/) published a short test of Intel V8, the prototype of a new motherboard.
xbitlabs: Intel V8 Eight-Core Platform Preview (http://www.xbitlabs.com/articles/cpu/display/intel-v8.html)
Conclusion Our experience with Intel V8 platform made an unforgettable impression. The obtained benchmark results are so high that we do not hesitate to call it the today’s world’s fastest system for multi-threaded work. This desktop and workstation platform with two quad-core Xeon processors demonstrates unattainable performance in all SMP optimized applications.

FB DDR2 memory subsystem in the eight-core monster from Intel is implemented in not the most optimal way. Its latency is too high, which affects the system performance in some widespread applications.

Updated V8 testbed: http://www.legitreviews.com/article/527/1/

Intel’s upcoming Skulltrail platform. Skulltrail is effectively a competitor to AMD’s Quad FX platform, it is a dual socket motherboard capable of supporting two quad-core CPUs. Intel lists Skulltrail as having support for “four PCI Express slots”, which we can only assume means four PCIe x16 slots.

Skulltrail won’t make its debut until later this year and Intel isn’t confirming whether or not it uses LGA-775 chips or Xeons.

The predecessor to Skulltrail is Intel’s V8 platform.

Intel's X38 chipset to support SLI?
by Cyril Kowaliski (ckowaliski@techreport.com) - 10:06 am, June 27, 2007

At Computex earlier this month, motherboard makers made no secret of the fact that Intel's upcoming X38 Express chipset will support AMD CrossFire multi-GPU setups. That may not be the whole story, though. The folks at The Inquirer say the X38 will also work with Nvidia SLI configurations (http://www.theinquirer.net/?article=40615), making it a sort of multi-GPU jack of all trades. Nvidia has vehemently denied that the X38 will support SLI, but The Inq claims to know for a fact that the contrary is true. Anonymous "engineers" tell the site that SLI drivers for the X38 are "in the final state of polishing with a few bugs left to quash, but plenty of time remains to finish things up."

Also,

We already know about Intel's upcoming X38 Express chipset, which is scheduled to replace (http://techreport.com/discussions.x/13052) the aging 975X in October. However, the folks at DigiTimes have word that Intel is prepping a second enthusiast chipset (http://www.digitimes.com/NewsShow/MailHome.asp?datePublish=2007/9/14&pages=PD&seq=207) for a launch not long after the X38's introduction.


DigiTimes says this second chipset is dubbed X48, and that it will be based on the same design as the X38, with a few notable changes. The maximum front side bus speed will be bumped up from 1333MHz to 1600MHz, and the maximum supported memory speed will similarly climb from DDR3-1333 to DDR3-1600. Board makers will reportedly be able to use the same PCB for either X38 or X48 mobos, and X48-based products should start hitting stores some time in the first quarter of 2008.

Also, OCZ intros a new CrossFire-certified power supply. has launched a new PC Power & Cooling PSU for enthusiasts with multi-GPU systems (http://www.ocztechnology.com/aboutocz/press/2007/226). The new PC Power & Cooling Silencer 750 Quad CrossFire Edition (http://www.pcpower.com/products/viewproduct.php?show=S75CF) is "tested and qualified under AMD's renowned certification program" and features four PCI Express power connectors, including two 8-pin PCIe power connectors, allowing it to support two Radeon HD 2900 XT or GeForce 8800 Ultra graphics cards out of the box.

TZ
05-09-2007, 08:10 AM
Samsung Electronics is putting the finishing touches on its DDR3 memory chips ahead of their commercial release later this year, 21 of its DDR3 memory chips and modules have been validated by Intel. This validation, which certifies the memory chips are compatible with Intel's PC chip sets, is one of the final steps toward commercial release of the chips.

The newer chips will offer data transfer speeds up to 1.6Gbps, twice the memory bandwidth of DDR2. That means better performance for both 3-D graphics and multithreaded applications that tap the power of multi-core processors. The chips will also consume less power -- around 1.5 volts compared to 1.8 volts for DDR2.

DDR3 specifications.
It is supposed to double the bandwidth to further increase data transfer speed and communication. One will have to wait for at least half a year before seeing competitive DDR3 chips with more aggressive and lower latency, speed is nothing without responsiveness. Nevertheless computer companies should quickly adopt the DDR3 SO-DIMM modules as it consumes less power than current DDR2 SODIMM. However, Apple should not adopt it before Penryn is released in the beginning of 2008.

Apple could quickly move to DDR3 with the Mac Pro, however, which uses FB-DIMM memory modules feature a controller to optimize memory chips management, but led to high latency. This well-known weakness could be partially solved by FB-DIMM DDR3, such modules should be directly compatible with Mac Pro. Indeed, the controller does not exchange data with memory chips, so it should not be a compatibility issue.

Volume production of DDR3 chips before the end of June, in time to start commercial sales during the second half of the year. Hynix Semiconductor. said its DDR3 chips, which go into production during the third quarter, had been validated by Intel.

While DDR3 gets close to its commercial debut, most PC users won't see the chips in their systems until early 2009, when they replace DDR2. The crossover point where shipments of DDR3 exceed DDR2 will come in early 2009 or late 2008.

<style type="text/css"> <!-- @page { size: 8.5in 11in; margin: 0.79in } P { margin-bottom: 0.08in } --> </style>
DDR3 SDRAM: Revolution or Evolution?



Platforms with DDR3 SDRAM support are already available in the market. So, it is high time we performed a comparative test session of DDR3 SDRAM against DDR2 SDRAM. The main goal of our today’s investigation is to find out what memory technology is more suitable for today’s computer systems, and what prospects each of them has ahead.

http://www.xbitlabs.com/articles/memory/print/ddr3.html

According to Digitimes (http://www.digitimes.com/bits_chips/a20070704PD212.html), Intel could drop the exclusive support for the FB-DIMM with its next generation of chipsets dedicated to servers, the same type of chipset Apple is currently using in Mac Pro and Xserve. The upcoming Intel 5100 chipset (San Clemente) is expected to support only RDIMM (registered DIMM) memory modules instead of FB-DIMM (fully buffered DIMM). The price of FB-DIMM did not really drop overtime, remaining several folds higher than registered memory modules. In addition RDIMM modules are currently available with higher capacity.


- Intel will then install a memory controller directly in the processor, as AMD does it.
- RDIMM modules will be faster than FB-DIMM which was slow down by its poor latency
- RDIMM modules are and will be cheaper than FB-DIMM

TZ
05-26-2007, 07:10 AM
Apple Mac Pro Memory up to 4GB FBDIMMs (http://store.apple.com/1-800-MY-APPLE/WebObjects/AppleStore.woa/wa/RSLID?find=FB-DIMM&btn.x=0&btn.y=0)
Memory DDR2 PC2-5300 • CL=5 • Fully Buffered • ECC • DDR2-667 • 1.8V • 512Meg x 72

* Eight FB-DIMM slots on two memory riser cards (4 slots per card) supporting up to 16GB of main memory
* Up to 256-bit wide memory architecture
* The width of each 667 MHz memory bus is 72 bits (including ECC)
* CAS latency of 5
* Maximum number of devices per FB-DIMM is 36

Memory is connected to the North Bridge via two 128-bit channels 256 bit). Each channel services two memory slots via the Advanced Memory Buffers (AMB) using FB-DIMMs. Every Mac Pro comes with two memory riser cards, each with four fully-buffered DIMM slots. With a total of eight DIMM slots available, you can install up to 16GB of memory. To help dissipate heat, every Apple DIMM you purchase for your Mac Pro comes with its own preinstalled heat sink. This unique heat sink lets fans run slower and quieter yet keeps the memory cool enough to run at full speed.

Apple Instruction (1) "Always shut down the computer before opening it". (2) "Wait 5 to 10 minutes to allow the computer's internal components to cool. Warning: After you shut down the system, the internal components can be very hot. You MUST [emphasis added] let the computer cool down before continuing."

The internal components of a Mac Pro do get very hot. This includes the FB-DIMMs, which consume about twice the electrical power of normal DDR2 DIMMs and which are rather exposed to the air when the access panel is removed. The stress imposed on these components from non-uniform heating and cooling could be considerable and add up over time.

Mac Pro's memory architecture consists of two memory riser cards organized by branches, channels, and DIMMs. The memory controller hub (MCH) on the North Bridge has two branches with Branch 0 going to Riser A and Branch 1 to Riser B.

The North Bridge IC connects to an Advanced Memory Buffer (AMB) device on the FB-DIMM modules using a high speed serial link running at 4 GHz DDR, with 14 northbound (read) lanes and 10 southbound (write) lanes. The North Bridge IC has two high speed serial links, each capable of addressing four FB-DIMMs. The DIMMs are arranged in pairs operating in lock-stepped, dual channel mode to provide a 128-bit wide datapath, plus 16 bits of ECC.

Each memory riser card has four DIMM slots labeled 1 to 4.
On Riser A, Channel 0 serves DIMMs 1 and 3 and Channel 1 serves DIMMs 2 and 4.
On Riser B, Channel 2 serves DIMMs 1 and 3 and Channel 3 serves DIMMs 2 and 4.

The AMBs on DIMMs 1 and 2 connect to the MCH via a high speed serial bus. If installed, the AMBs on DIMMs 3 and 4 connect to the AMBs on DIMMs 1 and 2, respectively. The high speed serial bus consists of 14 unidirectional northbound lanes (towards the MCH) and 10 unidirectional southbound lanes (towards the DIMMs) to carry command, address, and data traffic.

The MCH operates each branch independently. At DDR2-667 speeds, each branch provides a peak theoretical thruput of 10.6 GBps of northbound (from memory to North Bridge) traffic and 5.3 GBps of southbound (from North Bridge to memory) traffic.

FB-DIMMs are wired differently making it easier to route them on the motherboard. Fewer wires means lower cost, fewer pins per socket, and fewer traces (but slightly higher latency). FB-DIMMs also have dedicated and different paths for data transmission and reception along with additional memory channels.

Newly introduced FBDIMMs offer virtually unlimited scalability of density, a significantly reduced number of routed motherboard signals, and high bandwidth solutions, all with an extremely reliable channel protocol.

One strength of Intel's FB-DIMM architecture used in Xeon servers is that you can execute read and write requests to the AMB simultaneously. With standard DDR2 memory, you can do one or the other, and there's a penalty for switching between the two types of operations. If you have a fairly random mixture of reads and writes you can waste a lot of time switching between the two rather than performing all of your reads sequentially then switching over to writes. FB-DIMM is a server-class technology that adds some memory access latency and power draw in return for better signal integrity and potentially more bandwidth. Even in the server world, it's a controversial trade-off, but on the desktop, FB-DIMMs just don't make sense right now. The additional 5W per module that FB-DIMMs add over DDR2 isn't especially welcome, but the memory latency is an even bigger drawback.

Spike in reported FBDIMM temperature.

We occasionally see this situation with the FBDIMM modules for the Mac Pro. Surprisingly often the cause of the problem is actually a firmware issue.

Apple is aware of the problem and has released the Mac Pro EFI Firmware Update 1.2, that is likely to solve your problem.You can find it at the following page:
http://www.apple.com/support/downloads/macproefifirmwareupdate12.html

Another possible cause for the seeming spike in temperature would be a faulty sensor in the DIMMs' AMB (Advanced Memory Buffer). When the temperature sensor malfunctions your Mac Pro assumes the worst and reports the highest temperature available. If a firmware upgrade doesn't solve your problem we will assume the temperature sensor is faulty and in that case we will of course replace the memory kit for you.
Micron: FB-DIMM Technology technotes: (http://www.micron.com/products/modules/fbdimm/technotes)
TN-47-21: FBDIMM Ð Channel Utilization (Bandwidth and Power) (http://download.micron.com/pdf/technotes/ddr2/tn4721.pdf)
Fully Buffered DIMMs: Improve server performance (http://images.crucial.com/pdf/fbdimmWhitePaper.pdf)
TN-47-16: Designing for High-Density DDR2 Memory (http://download.micron.com/pdf/technotes/ddr2/TN4716.pdf)

Note: On Mac Pro systems, the distribution of memory modules may affect the performance characteristics of your video applications.

Final Cut Studio: For best performance on Mac Pro
install memory in risers symmetrically (http://docs.info.apple.com/article.html?artnum=304492) To achieve optimal performance when running Final Cut Studio applications, memory DIMM pairs should be installed evenly on both risers.

Note: DIMMs must be installed in pairs of equal size from the same vendor. For instance, you must not have one or three DIMMs on either riser at any time. Additionally, two DIMMs from different vendors should not be combined and used as a pair.

Other configurations of DIMM pairs are still compatible with Final Cut Studio, but may not achieve the same performance levels as when DIMM pairs are installed on both risers evenly.

For best performance on Mac Pro, install memory in risers symmetrically, DIMM pairs should be installed evenly on both risers.
DIMMs must be installed in pairs of equal size from the same vendor.
DIMMs from different vendors should not be combined and used as a pair. Other configurations of DIMM pairs are still compatible, but may not achieve the same performance levels as when DIMM pairs are installed on both risers evenly.
http://docs.info.apple.com/article.html?artnum=304492For additional information:

TechReport: Intel V8 (http://techreport.com/reviews/2007q2/intel-v8/index.x?pg=1)
Anandtech Page 7: A Faster Memory Controller (http://www.anandtech.com/printarticle.aspx?i=2939)
HardwareSecrets.com (http://HardwareSecrets.com) -
introduction to the memory technology (http://www.hardwaresecrets.com/article/266)
FB-DIMM Explained (http://www.valueram.com/fb-dimm/default.asp)
- Memtest will not see ECC corrected (Soft) errors - Check Apple System Profiler memory section for status of those.

Intel quad-channel memory throughput benchmarks are over 40% higher when set up to use 256-bit.
Tom's Hardware: Quad-Channel Memory Bandwidth (http://www.tomshardware.com/2006/06/26/xeon_woodcrest_preys_on_opteron/page10.html#dual_vs_quad_channel_memory_bandwidth)
RAM Expansion Developer Note (http://developer.apple.com/documentation/HardwareDrivers/Conceptual/HWTech_RAM/index.html#//apple_ref/doc/uid/TP40003031)
Developer Note: RAM Implementation (http://developer.apple.com/documentation/HardwareDrivers/Conceptual/HWTech_RAM/Articles/RAM_implementation.html)
http://www.apple.com/macpro/expansion.html
http://developer.apple.com/documentation/HardwareDrivers/Conceptual/HWTech_RAM/Art/060634001312_05.jpg
Thermal Considerations (http://developer.apple.com/technotes/tn2006/tn2156.html)
Mac Pro Memory DIY Guide (pdf) (http://manuals.info.apple.com/en/MacPro_MemoryDIMM_DIY.pdf)
http://www.apple.com/support/manuals/macpro/

Support 32GB
XLR8YOURMAC Report 32GB RAM (http://www.xlr8yourmac.com/feedback/32GB_RAM_Mac_Pro.html)
http://www.legacyelectronics.com/
Crucial CT685422 (http://www.crucial.com/store/listparts.aspx?model=Mac+Pro+%284%2Dcore%29) l

Each memory slot can hold DDR2 PC2-5300 with a maximum of 4096MB per slot. *Not to exceed manufacturer supported memory.
Mac Pro FB-Dimm Memory Reports/Tests (http://www.xlr8yourmac.com/systems/Mac_Pro/mac_pro_ram.html)

TZ
05-30-2007, 04:48 AM
We used the following versions of our test applications:

SiSoft Sandra XI.SP4a 64-b (http://www.sisoftware.net/)it
CPU-Z (http://www.cpuid.com/) 1.40
SPECjbb2005 (http://www.spec.org/jbb2005/) with Sun Java 6 Update 2 (http://java.sun.com/javase/downloads/index.jsp) Windows x64 edition
Valve (http://www.valvesoftware.com/) VRAD map build benchmark
Cinebench (http://www.cinebench.com/) R10 64-bit Edition
POV-Ray (http://www.povray.org/) for Windows 3.7 beta 22 64-bit
CASE Lab Euler3d CFD (http://www.caselab.okstate.edu/research/euler3dbenchmark.html) benchmark multithreaded edition
MyriMatch (http://www.mc.vanderbilt.edu/msrc/bioinformatics/Bumber.php) proteomics benchmark
picCOLOR (http://www.fibus.org/fibusimg.htm) 4.0 build 598 64-bit
The Panorama Factory (http://www.panoramafactory.com/) 4.5 x64 Edition
Windows Media Encoder 9 (http://www.microsoft.com/windows/windowsmedia/9series/encoder/default.aspx) x64 Edition

Intel's Stoakley platform and 45nm Xeons
http://techreport.com/articles.x/13224/2
____________________________

All three tests proved to CPU (bound) intensive - making the Mac Pro the best choice for
(http://www.macintouch.com/readerreports/benchmarking/index.html#may24)Aperture (http://www.barefeats.com/aper01.html) against G5 Quad.

OTHER ARTICLES ON THE MAC PRO by BARE FEATS (http://www.barefeats.com/aper01.html)
Mac Pro with X1900 XT running Motion 2 and iMaginator (http://www.barefeats.com/quad12.html)
Mac Pro versus Quad-Core G5 --
Updated Photoshop CS2 results under 10.4.8 (http://www.barefeats.com/quad13.html)
Original Test Results on Mac Pro
Photoshop CS2, After Effects 7, iMovie HD, Final Cut Pro, FileMaker 8.5, and Cinebench 9.5 (http://www.barefeats.com/quad06.html)
Fastest Mac Pro Boot Drive (http://www.barefeats.com/quad08.html)
Cinebench 10 and Motion graphs (http://barefeats.com/imacal2.html)

What You Should Know About Mac Pro Memory (http://www.barefeats.com/quad09.html)

Macintouch (http://www.macintouch.com/readerreports/benchmarking/index.html#may24) Cinebench Results

Rendering (Single CPU): 439 CB-CPU
Rendering (Multiple CPU): 1381 CB-CPU
Multiprocessor Speedup: 3.15

Shading (CINEMA 4D) : 527 CB-GFX
Shading (OpenGL Software Lighting) : 2109 CB-GFX
Shading (OpenGL Hardware Lighting) : 4439 CB-GFX
OpenGL Speedup: 8.42

Xbench Results

Ric Ford [MacInTouch]
Mac Pro 2.66GHz/2GB RAM/Seagate 250GB hard drive
Results 185.99
There are a couple programs on the site. Lloyd Chambers wrote a $20 utility that is favored by Barefeats and AMUG in their drive testing.

MacIntouch Benchmarking Mac Pro (http://www.macintouch.com/readerreports/benchmarking/index.html#may24)

http://www.aja.com/html/support_kona3_swd.html
http://www.aja.com/ajashare/AJA_KONA_System_Test_v2.app.tar

Eight Core Mac Pro Benchmarks (http://www.geekpatrol.ca/2006/12/eight-core-mac-pro-benchmarks/)
Anandtech: Quad Core Intel Xeon 53xx Clovertown (http://www.anandtech.com/IT/showdoc.aspx?i=2897)
How to upgrade the processors in the Mac Pro (http://www.anandtech.com/mac/showdoc.aspx?i=2832&p=6)
Mac Pro (http://www.apple.com/macpro/)
quad-core Xeons (http://processorfinder.intel.com/details.aspx?sSpec=SL9YL)
Geekbench (http://www.geekpatrol.ca/geekbench/)
GeekPatrol - 8-core MacPro Benchmarks (http://www.geekpatrol.ca/2006/12/eight-core-mac-pro-benchmarks/)
GeekPatrol Benchmark Redux (http://www.geekpatrol.ca/2006/02/geekbench-comparison-redux/)

Geekbench (http://www.versiontracker.com/dyn/moreinfo/macosx/32043) with GUI for testing cpu performance, comparison to other Macs and platforms.

Product Description: Geekbench is a completely cross-platform benchmark that measures the performance of your computer's processor and memory. Geekbench also offers the Geekbench Result Browser which lets you compare your Geekbench scores with other Geekbench users.

What's new in this version: Adds Linux support to Geekbench; users running Fedora Core 6 (or later) on Intel-compatible hardware can now benchmark their computers. Also fixes a number of system information bugs on both Mac OS X and Windows. Geekbench costs $20.
Mac Pro Geekbench Results (http://browse.geekbench.ca/geekbench2/similar/5623/5)

To run benchmarks:
http://www.primatelabs.ca/geekbench/
browse results:
http://browse.geekbench.ca/

TZ
06-07-2007, 06:47 AM
Avid to Expand Delivery of Content Creation Tools for Applefs Intel-based Macs (http://www.avid.com/company/releases/2006/060816_macintel_avid.html)

The Mac version has been updated with a Universal version that fully supports both Intel and PowerPC-based systems.

Universal Mac support. With version 5.7, every Avid Xpress Pro system now contains a version of Avid Xpress Pro for the Mac and Avid Xpress Pro for the PC. The Mac version has been updated with a Universal version that fully supports both Intel and PowerPC-based systems

Avid Online Store (http://www.avidstore.com/us/index.cfm?page=templates/product_detail&PartNumber=7010%2D20036%2D01&manufacturerid=17&productid=61069) $345 upgrade price.

Macintosh

For all systems listed below, the following minimum specifications are required:

* Mac OS X 10.4.6, 10.4.7, or 10.4.8*
o Earlier versions of Mac OS X are not supported
*Mac OS X 10.4.8 requires Avid Xpress Pro version 5.6.3 or later software
*Mac OS X 10.4.9 not qualified as of Avid Xpress Pro v5.7 release
* Dual 2.0 GHz Power Mac® G5, single 1.8 GHz Power Mac G5, or MacBook Pro 2.16 GHz, or Mac Pro with Dual-Core Intel Xeon Processor

http://www.avid.com/products/xpresspro/specs.asp
Forums (http://www.avid.com/exchange/forums/42/ShowForum.aspx)
</td></tr></tbody></table>

TZ
06-14-2007, 06:17 AM
Support for 8-core Mac Pro

Mac OS X 10.4.9 System Requirements (http://www.digidesign.com/index.cfm?langid=100&navid=54&itemid=22904)
June 8, 2007 — Mac OS X 10.4.9 now qualified with Digidesign products with Pro Tools 7.3.x software. If you choose to update to Mac OS X 10.4.9, or your Mac came with Mac OS X 10.4.9 pre-installed, please follow the instructions here to avoid problems such as failure of Pro Tools to launch:

Do not use Apple's automatic Software Update or the Mac OS X 10.4.9 Delta Update. Instead, install the Mac OS X 10.4.9 Combo Update for Intel-based or PowerPC-based Macs:
Mac OS X 10.4.9 Combo Update (Intel) (http://www.apple.com/support/downloads/macosx1049comboupdateintel.html)
Mac OS X 10.4.9 Combo Update (PPC) (http://www.apple.com/support/downloads/macosx1049comboupdateppc.html)
If you have already updated using Software Update, you should download and install the Mac OS X 10.4.9 Combo Update.
If your Mac came with Mac OS X 10.4.9 preinstalled, you should still download and install the Mac OS X 10.4.9 Combo Update if you are experiencing issues such as Pro Tools not opening.
In some newer Macs, such as MacBook Pro laptops with the new Intel Santa Rosa chipset, you may not be able to apply the 10.4.9 Combo update to the existing 10.4.9 preinstallation.
If you are not able to run the Combo Update, please try the Mac OS X Prebinding Script below. Please Note: As detailed on the Digidesign website, Software Update should not be used to update your Mac OS, always use the Combo Update:
How to Get Mac OS X 10.4 Upgrades & Updates (http://www.digidesign.com/index.cfm?langid=100&navid=54&itemid=23369)With Pro Tools and many other software applications, issues with a program not opening or crashing may have to do with prebinding that is not properly handled by using the automatic software update program, particularly programs such as Pro Tools that use PACE-protected software. Both PACE and Apple are aware of this issue and are working to provide updates to remedy the issue.

The following download from the PACE website may help without having to download the Mac OS X 10.4.9 Combo Update, or in addition to the Combo Update. The included script resets the prebinding on your Mac OS X system:
Mac OS X Prebinding Script & Instructions (http://www.paceap.com/prebind_script.dmg)For additional troubleshooting, you may need to delete corrupted Pro Tools preferences and database files. A handy utility created by Jean-Charles Deshaies that does this for you is available here:
ProTools Pref and Database Helper (http://web.mac.com/jcdeshaies/iWeb/DeleteProToolsPrefs/Bienvenue.html) Pro Tools CS Updates (http://www.digidesign.com/index.cfm?navid=54&itemid=23300)
* Pro Tools HD 7.3.1cs3 Update (Mac OS X Universal Binary) (http://www.digidesign.com/index.cfm?langid=100&navid=54&itemid=25128) 115.11 MB Download
* Pro Tools M-Powered 7.3.1cs3 Update (Mac OS X Universal Binary) 98.39 MB Download
* Pro Tools Academic 7.3.1cs3 Update (Mac OS X Universal Binary) - Hardware Authorized
Downloads (http://www.digidesign.com/index.cfm?langid=100&navid=54&itemid=23375)

Pro Tools LE 7.1.1cs2 Update for Intel-based Macs (http://www.digidesign.com/index.cfm?langid=100&navid=54&eid=1185&categoryid=36) - October 5, 2006
For Digi 002, Digi 002 Rack, Mbox, or Mbox 2 Systems with Pro Tools LE 7.1.1 for Intel-based Macs Only

Who Should Use Pro Tools 7.3.1? All Pro Tools 7.3 Users
— This Update is Required to Avoid Potential Data Loss
Pro Tools CS Updates (http://www.digidesign.com/index.cfm?langid=100&navid=54&itemid=23300)
February 12, 2007 Note: The same important fixes in the Pro Tools 7.3.1 updates are also included in the Pro Tools 7.3.1cs1 and higher updates, available from the Pro Tools CS Updates page in the Support section Quick Links. Pro Tools 7.3 users can skip the Pro Tools 7.3.1 update and go directly to the latest Pro Tools 7.3.1 "CS" release if they choose, without having to install the 7.3.1 updates below.

Mac Pro 3GHz 8-core - MacWorld Review (http://www.macworld.com/2007/07/reviews/macpro8/index.php)
New processor option doubles cores, but not performance

- machines running ProTools HD systems (with PCIe cards)
Digidesign and Apple are still trying to find out what is going on,

The theory I have is with the power supply and the power needed for startup

Mac Pro configuration:

4 WD RE2 500 GB disks
X1900 ATI
Digidesign ProTools HD2 and HD3 system (using 2, resp. 3 PCIe cards for the DSP power).

With this configuration, 9 out of 10 systems had the symptom. The funny thing is, the same system in bare-bones, i.e. with only 1 HD and no PCIe cards - only the x1900 - did not show the problem. Changing the PCIe configuration (with the PCIe expansion slot utility) sometimes helped, giving less speed to the PT HD cards, and full speed to the X1900 - but not always.

TZ
06-15-2007, 02:33 PM
Update to Final Cut Pro 5.1.4 for best performance
Apple has posted a Knowledge Base article indicating that performance of real-time functionality in Final Cut Pro 5.1.2 may be diminished on Mac Pro systems after the update to Mac OS X 10.4.9. The remedy is to update to Final Cut Pro version 5.1.4 (http://www.apple.com/support/downloads/finalcutpro514.html).

Final Cut Pro: Additional RAM may be recommended for High Definition video formats (http://docs.info.apple.com/article.html?artnum=305177)

Compressor 3 is 3x faster.

With a Mac Pro 8-core with (this) sample it will take us 16:45 in Compressor 2. In Compressor 3 it's down almost 3x to 6:02. I've encoded a 10 minute clip in less than ten minutes. Compressor 3 demo time.

Apple's new ProRes 422 video format (http://www.appleinsider.com/articles/07/04/18/a_closer_look_at_apples_new_prores_422_video_forma t.html)

Going into extreme detail, Apple has said its new HD video standard was designed not just to beat opposing formats but to clear bottlenecks -- some of which are stifling for video editors.

Those fortunate enough to work at a Mac Pro are poised to see even more of a benefit, Apple boasts. An improved half-resolution decoding method means that Final Cut Pro 6's unlimited real-time preview sees an exponential increase in the number of active previews: where only a single 1080i, 60Hz video would be visible on an eight-core Mac Pro in uncompressed form, the use of ProRes 422's high quality mode allows four. Dropping the quality to 720p at a film-level 24Hz sees the number of streams jump from three to fourteen on the same system and increases further with the normal 145Mbps ProRes bitrate.

this performance boost is explicitly adapted to multi-core systems. ProRes is said to scale almost linearly with the number of cores available to process video, cutting the latency in drawing a finished frame in half every time the available processing power doubles.

A Mac Pro with an HD SDI video card, used for capturing live broadcasts, can automatically transform raw HD into a ProRes 422 clip without losing visual data, optionally sending it back to other hardware at the same quality.

Final Cut Pro & Compressor use about 450% of the CPUs.
You Control installed and there are 8 little CPU monitors in my menu bar.
They all are running when I use compressor, but only at around 50% each.
Activity Monitor shows Compressor using around 350% to 370%
Final Cut Pro using around 80% to 100%.

Power Fractals will pop max it out pretty well, but it is only for a few seconds. Even with the 30" Cinema display, it doesn't take long to fill the screen when it is doing 62 GigaFlops!

AMD: two quad-core chips running side by side. AMD didn't show any benchmark results, but according to the report, the second system was "able to convert 720p video from one format to another (the company declined to comment on file formats) in real time – while maxing out all eight cores at the same time."

Final Cut Pro 6: HD ProRes 422 capture requires Mac Pro Capturing HD resolution video using the ProRes 422 format requires a Mac Pro with Intel Xeon processors and a qualified third-party capture card. Other Mac systems may drop frames when capturing High Definition video in the ProRes 422 format.

For more information about Apple ProRes 422 and devices qualified for use with Final Cut Studio 2, see the following resources:
Final Cut Studio 2 System Requirements (http://www.apple.com/finalcutstudio/specs.html)
Final Cut Studio Partners (http://www.apple.com/finalcutstudio/partners/io.html)
Qualified Devices Guide (http://www.apple.com/finalcutstudio/resources/devicelist.php)

Final Cut Pro: Additional RAM may be recommended for High Definition video formats (http://docs.info.apple.com/article.html?artnum=305177) Final Cut Pro may drop video frames in cases where there is not enough system memory available. For instance, if you're editing an uncompressed 8-bit or 10-bit high-definition codec, and you're monitoring video externally via a PCI or PCI-E video interface, Final Cut Pro may drop frames occasionally if your computer does not have sufficient system RAM.

Solution Install additional RAM in your computer to increase the amount of memory available to Final Cut Pro.

Note: The following table corresponds to the minimum memory recommendations for Final Cut Pro only.
If you run other applications at the same time as Final Cut Pro, they will allocate additional memory from the system, reducing the amount available to Final Cut Pro.

HD formats: Suggested RAM
HDV formats 2 GB
Compressed HD formats 2 GB
Uncompressed HD formats 4 GB

Refer to manufacturer's recommendations for the memory requirements of your video interface:

AJA Video Systems (http://www.aja.com/html/support.html), manufacturer of the KONA family of video interfaces
Blackmagic Design (http://www.blackmagic-design.com/support/detail.asp?techID=60), manufacturer of the DeckLink family of video interfaces

Final Cut Studio: On Mac Pro systems, the distribution of memory modules may affect the performance characteristics of your video applications. For best performance on Mac Pro Install memory in risers symmetrically (http://docs.info.apple.com/article.html?artnum=304492)

Note:

DIMMs must be installed in pairs of equal size from the same vendor.
DIMMs from different vendors should not be combined and used as a pair.
Other configurations of DIMM pairs are still compatible, but may not achieve the same performance levels as when DIMM pairs are installed on both risers evenly.

TZ
06-16-2007, 12:20 PM
Quad 3GHz Mac Pro vs Quad 2.5GHz PowerMac
There's no doubt that both versions on the Mac Pro are faster than the G5 Quad-Core running Universal Binary apps like iMovie, Final Cut Pro.

The Mac Pro 2.66GHz was as much as 62% faster than the Quad-Core G5/2.5GHz. The Mac Pro 3.0GHz was as much as 85% faster.
Barefeats: Quad06 (http://www.barefeats.com/quad06.html)
MacRumors (http://www.macrumors.com/pages/2006/08/20060816233352.shtml)
ForceVMBuffering plug (http://www.adobe.com/support/downloads/detail.jsp?ftpID=3623) is available online for PS CS3

The Mac Pro and Photoshop CS3 (http://www.macworld.com/weblogs/editors/2006/12/mpupdate/index.php)

" stop thinking of system memory as a resource and start thinking of it as a a cache (http://www.codinghorror.com/blog/archives/000688.html) system memory is yet another type of high-speed cache that sits between your computer and the disk drive. ... the most important rule of cache design is that empty cache memory is wasted cache memory. Empty cache isn't doing you any good."

MacIntouch: Adobe (http://www.macintouch.com/readerreports/applications/topic4118.html)

known problems CS2:

Installation troubleshooting
http://www.adobe.com/cfusion/knowledgebase/index.cfm?id=329590
Illustrator CS2 crashing on Intel Based Mac
http://www.adobe.com/cfusion/knowledgebase/index.cfm?id=333449

Pathfinder functions crash Illustrator CS2 on Intel Macs running 10.4.8.

Open Illustrator.
Close the "Appearance/Graphic styles" tool panel.
Restart Illustrator.
Pathfinder tools should now work.

The "Appearance/Graphic styles" tool panel and the crashes don't seem to return. http://www.macintouch.com/readerreports/applications/topic4118.html

CS2 on a 3 GHz Mac Pro, 5 gibibytes RAM, dual 30" ACDs, and they all run faster than they ever did on my dual 2 GHz G5.
http://www.adobe.com/support/techdocs/333449.html

4GB of RAM and a 1.5TB RAID 0. The hardware was easy to install ... my hat is off to the engineers at Apple!

Transcoded 2 hour dual layer DVD movie to single layer that took 2 hours on my G4 ... in less than 2 minutes!
______________
OUR CONCLUSION FROM TEST RESULTS ON THIS PAGE (and others)
In spite of the limitations posed by the memory bus and OS X "Tiger's" handling of core swapping, there are things that the 8-core Mac Pro does signficantly faster than the 4-core Mac Pro. If you are using After Effects CS3 (or any app like it that knows how to use all 8 cores effectively), the 8-core is worth the extra money, in our opinion.

And if you perform multiple tasks within an application (like our QuickTime simultaneous conversion test above) or have multiple apps active (like in our Multi-Processing page), then you are a candidate for an 8-core Mac Pro.

INSIGHTS & RECOMMENDATIONS We think it's awsome what Adobe had done to upgrade and refine After Effects CS3 so that it takes full advantage of all the cores (and all the memory) available in each Mac (both PPC and Intel). It's truly one of those applications that can sell hardware.
http://www.barefeats.com/octopro1.html
http://www.barefeats.com/octopro4.html

What's the story with Photoshop & multi-core? (http://blogs.adobe.com/jnack/2006/12/photoshop_and_multicore.html)

"... not all operations can or should be split among multiple cores, as doing so can actually make them slower.

Because memory bandwidth hasn't kept pace with CPU speed (see Scott Byer's 64-bit article for more info), the cost of moving data to and from each CPU can be significant.

Russell Williams, "The workers run out of materials & end up standing around."

The memory bottleneck means that multi-core can't make everything faster, and we'll need to think about doing new kinds of processing specifically geared towards heavy computing/low memory usage."MacWorld: Multi-Core Computing (http://www.macworld.com/2006/12/features/trends1/index.php)
64-bit Computing for Photoshop? (http://blogs.adobe.com/scottbyer/2006/12/64_bitswhen.html)

Notes on Mac Pro and Photoshop CS2/3:
http://diglloyd.com/diglloyd/free/MacPro/
http://diglloyd.com/diglloyd/2007-03-blog.html
MacWorld CS3 Benchmarks (http://www.macworld.com/weblogs/creative/2006/12/photoshop-cs3-performance/index.php)
Eight Core Benchmark (http://www.geekpatrol.ca/2006/12/eight-core-mac-pro-benchmarks/) multi-thread test
CS3 and Leopard Optimization (http://www.crn.com/software/197700697)
______________
The effects of RAID (http://www.macgurus.com/forums/showthread.php?t=19741)
CPU (bound) intensive tasks making the Mac Pro the best choice (http://www.macintouch.com/readerreports/benchmarking/index.html#may24)
Aperture (http://www.barefeats.com/aper01.html)
Motion 2 and Maginator (http://www.barefeats.com/quad12.html)

<a href=http://www.macgurus.com/forums/showthread.php?t=19792> Photoshop Acceleration Basics Guide (pdf) </a>

TZ
06-22-2007, 08:23 AM
Tutorial (http://www.o0o.it/pro/) on doing a cpu upgrade with Woodcrest (x51xx XEON Socket 771) processors or, more recently, with Clovertown ones (x53xx XEON Socket 771).

They don't draw much more power (on charts) BUT they give you plenty of computational power.

About 2 months ago i found the ideal choice: the nice E5320 (SL9MV).
It's power and thermal specs are quite similar to the 5150: 80W instead of 65W.

Also is the best for try the BSEL mod since it has a default bus of 1066Mhz,
so modding it to 1333Mhz we can push it to 2,33Ghz from the stock 1,86Ghz.

With this guide you'll have an 8 cores Mac Pro from a "normal" 4 cores one.

<a href=http://www.intel.com/cd/channel/reseller/asmo-na/eng/products/server/processors/q5300/feature/index.htm> Intel 53xx </a> keeps improving on its line of chips.
<a href=http://www.intel.com/technology/magazine/computing/quad-core-1206.htm> Quad-core and beyond </a>

On the horizon with 45nm Xeon quads due in '08 (except for a 135W 3GHz slated for November and $999 OEM price apiece).

"upgrade vs new" <a href=http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3069> Anandtech </a> They also did the first <a href=http://www.anandtech.com/mac/showdoc.aspx?i=2832&p=6> review of 8-core "upgrade" </a>

<a href=http://www.vr-zone.com/articles/1600FSB_Harpertown_To_Counter_Barcelona_In_Q4_%270 7/5187.html#Scene_1> Harpertown </a>

TZ
06-23-2007, 01:17 PM
I'm downloading Kubuntu 7.04. With Parallels or VMware you can run Mac or Windows or Linux, native or in a VM.

I would roll over in my grave if a 2.66GHz 4-core was a laggard. One guy has already pulled his 5150 (2.66) and dropped in a 53xx 2.33GHz running a full 8GHz above, with his own benchmarks. PS: I really like how fast and quiet the Mac Pro is.

I use to program (mainframes) but I would think Mac Pro with as much cpu as you can throw in would do wonderful for coding and development. However, trying to educate myself, I also think you might want to be able to dual or triple boot or something, too.

> ... had to use Netbeans 5.0 to develop an application using Windows. I have a Macbook Pro 2.0 w/ 2gig mem. Running XP, using Netbeans...

Apple Developer Connection (free).
They have articles, but their discussion is not appropriate on this board.
http://developer.apple.com/

Some general reference links:
http://java.sun.com/
http://www.netbeans.org/index.html
http://www.ibm.com/developerworks/java/
Performance Inspector (http://www.alphaworks.ibm.com/tech/pi?open&S_TACT=105AGX02&S_CMP=LP&ca=dgr-dw02awpi)
Performance Inspector for Linux (http://perfinsp.sourceforge.net/)
Sun Developer Network:
http://developers.sun.com/downloads/
http://java.sun.com/j2se/1.5.0/docs/api/
http://developer.apple.com/java/netbeans.html
http://developer.apple.com/tools/eclipse.html
http://developer.apple.com/java/creator.html

TZ
06-24-2007, 06:53 AM
salient points regarding 8-core Mac Pro memory as follows:
Most RAM sold for quad-core Mac Pros will work in 8-core Mac Pros
If the RAM is recognized by the 8-core Mac Pro, it will work with the system and should not present any further problems.
There is no change in heat sink size for 8-core Mac Pro RAM.
The only difference in 8-core Mac Pro-spec RAM is a more stringently defined AMB (advanced memory buffer) Mac Pro (8-core) [#3]: More on RAM requirements (http://www.macfixit.com/article.php?story=20070412121335316)

MEMORY BOTTLENECK? Lloyd Chambers thinks there's a memory bus bottleneck for the 8 core, which he discussed back in March on his blog. He wrote,
"Memory bandwidth is inadequate for 8 cores. It's already a limiting factor with the current quad-core 3.0 GHz Mac Pro. Memory copy speed is at best 2.9GB/sec on the Mac Pro, in spite of Apple's highly misleading claims of 21.3 GB/sec (maximum processor bandwidth of up to 21.3 GB/s bandwidth is a bit more than double the memory copy speed). That's a measly 700MB/sec per core on a quad-core machine, and only 350MB/sec per core on an octa-core machine. By comparison, a 6-drive hard disk RAID array can easily perform at over 400MB/sec!"We did run the "test-compute-speed" with digLloydTools (DLT) on the 8 core. Its aggregate rate was 1204MB/sec (versus the 4 core's 601MB/s). That says that if the task is pure CPU, the 8 core is twice as fast as the 4 core. But if your task has to do a lot of interaction with memory, the advantage drops almost to nothing -- as we saw with Photoshop CS3 and Aperture.

TIGER IS "CHASING ITS TAIL" "Yes, the memory bus is weak, but the biggest problem is that OS 10.4 (Tiger) does not know how to deal with so many cores. Perhaps 10.5 (Leopard) will.

OS X Tiger does not know that it should keep a thread on the SAME core instead of swapping it around to any one of the 8 cores. When a thread moves to a different core-group, the cache has to be reloaded on the new core.

Since Intel quad-cores are two Core 2 Duos "duct-taped" together, the problem is worse than it would otherwise have been. When this thread/core shift happens repeatedly, it makes the cache ineffective and floods the memory bus with activity." (Though put forth by a yet-to-be named BF reader, this analysis has been confirmed by Lloyd Chambers where he discussed "core swapping" on his latest blog entries regarding the 8 core Mac Pro.)


On jobs that require modest amounts of memory, performance is remarkable and typically double that of the 4-core model. On the largest jobs, when memory bandwidth is limiting, performance is really no better than the previous 4-core model (perhaps even 2-3% worse), even though all 8 cores are churning away.

Before jumping to an 8-core system, one should consider that, for an application that blindly attempts to use all available cores, the memory required might be nearly double that for 4 cores, which might exceed the available memory. Add the cost of more RAM to the cost equation for a practically configured 8-core Mac Pro. And consider that the 16 GB maximum might be insufficient to support 8 (or even 5) simultaneous threads working on larger problems, in which case an 8-core system might actually be a bad purchase. Some applications may need to be updated to use more than 4 threads, although multiple instances can sometimes be run simultaneously.

The 8-core Mac Pro is quiet, but I would not characterize it as silent like the 4-core model. The 8-core model uses an updated Darwin kernel and possibly other newer OS elements. Simply transfering disks from a 4-core system to the 8-core system might therefore be suboptimal for performance and reliability.The kernel and OS X build can be confirmed using sw_vers that it is indeed later than 10.4.9 (and there are some problems with the 2007-004 Security Update).

MacPro and Third-Party Memory Modules: How to Cool them Down non Apple-certified FB-DIMM modules checking temperature is an average 10-15°C higher than Apple-branded ones. To improve the cooling system add copper-based heatsink from ThermalTake (http://www.thermaltake.com/product/Cooler/VGAnCHIPnMEM/CL-P0025CL-%20%3Cbr%20/%3EP0026/CL-P002526.asp). With such components glued to the chips, the overall temperature goes down to the same level than Apple-branded ones, dropping from 75°C to 60°C.
- [ur=]http://www.barefeats.com/octopro6.html] Memory cooling fan kit from TransIntl [/url]

For the most part, there's no benefit to having all four channels populated, but in some rare cases the performance boost can be tremendous... we can assume that the scenarios where we do see a large performance gain are write bandwidth bound. Anandtech p. 12 (http://anandtech.com/mac/showdoc.aspx?i=2816&p=12)

When is 256-bit is not:

The major benefit that the Mac Pro seems to get from the use of FB-DIMMs is that its memory bus and FSBs can offer identical bandwidths at 21.3GB/s (ignoring the unknowns we discussed earlier about the efficiency of FBD). By using a lower pincount interface, Intel was able to fit four FBD channels on its 5000 series chipset and thus offer the bandwidth equivalent of a 256-bit wide DDR2 memory controller. Anandtech p. 4 (http://anandtech.com/mac/showdoc.aspx?i=2811&p=5)

There is a small but significant performance advantage of about 10% by using dual-rank modules.. ... Keep in mind that the performance of any specific program might or might not benefit from the optimal configuration. LLoyd blog (http://diglloyd.com/diglloyd/free/MacPro/memory.html#PerformanceNumberAndRank)
Diglloyd: Mac Pro Memory (http://diglloyd.com/diglloyd/free/MacPro/memory.html)

Cache and memory in the many-core era (http://arstechnica.com/articles/paedia/cpu/cache-mem-many-core.ars)As CPUs gain more cores, resource management becomes a critical performance challenge. Here's how Intel is addressing the problem of resource contention for our multi-core, virtualized future.

TZ
06-26-2007, 03:58 AM
Overclock of Mac Pro Xeons... via Windows
Hardmac (http://www.hardmac.com/news/2007-06-26/#6940)

One can overclock Mac Pro Xeon via Windows with SysTool allowing you to directly modify the clock generator. SysTool can even force the clock generator to change its frequency without rebooting. As shown below, the Mac Pro generator is identified as an ICS932S401. With SysTool and Apple memory modules we successfully overclocked the FSB from 333 x 4 (1333MHz) to 380 x 4 (1521MHz).

The test machine, a Mac Pro powered by two Dual Core Xeon clocked at 3.0GHz, is now an overclocked to 3.424GHz. The performance gain is proportional to the overclocked Xeon. With Dane-Elec memory modules, we could push the FSB a bit further to 389 x 4 MHz, Xeon are then overclocked to 3.5GHz. The overclocked limit is defined by the memory modules. Indeed, as indicated by SysTool, the CPU - Memory ratio is set to 1:1, and we did not find ways to desynchronize them.

However Apple seemed to have found ways to prevent users to perform such software-based overclock. Indeed when rebooting the overclocked Mac Pro, while the new frequency is saved in a volatile register, OSX refuse to boot and deliver a memory alarm. To bring it back to life, you simply need to switch it off and on.

It is currently not possible to perform such software-based overclock via OS X. The application Reggie SE from the CHUD Tools could have done it, but it has been locked by Apple, and if you attempt to modify registers it immediately generate Kernel Panics.

The Linux command line i2cget might be of help? We might need a way to directly access the clock generator ICS932S401 in OS X.

If you want to join our effort, do not hesitate to react on our Hardmac forum (http://forum.hardmac.com/index.php?showtopic=1639&view=getnewpost).

SysTool Build 730 SysTool will only run on Windows 2000/XP/2003 with or without any Service Pack.
SysTool should work Windows XP 64-bit. Some features might be buggy. Please report them. Vista compatibility, won't Run.

Latest version http://systool.techpowerup.com/
http://www.techpowerup.com/wiki/doku.php/systool/installation_guide

TZ
07-12-2007, 01:55 PM
Re: MAC Pro DUAL QUAD CORE Runing 3DSMAX9

The specs are as follows:

Dual Quad-Core Intel® Xeon® processor X5355 CPUs running at 2.66GHz
Intel® Workstation Board S5000XVNSAS motherboard
Intel® Chipset 5000X
8GB FB-DIMM DDR2-667 RAM
Nvidia Quadro 4600 768mb RAM
2-500gig SATA Drives RAID-0
1-500gig SATA USB2.0 External for Backup of scenes,meshes, matlibs and texture database
1000 watt Power supply from Thermaltake
Full Towercase BLACK ARMOR CASE from Thermaltake

3D Studio MAX 9.0 and Combustion 4.0, Mental Ray

Audodesk on Mac Pro (http://discussion.autodesk.com/thread.jspa?messageID=5588899)

schalliol
08-08-2007, 07:11 AM
Apple now offers a RAID Card for the Mac Pros and xserves that utilizes the existing drives. I was wondering if anyone had thoughts on it, how it works, what it really does, etc. With only 4 internal drives available, I wondered if perhaps using this card you could boot on RAID 5. It sounds pretty steep at $1k.

http://www.schalliol.com/forums/MacGurus/newraid.png


"Ideal for video and creative professionals with demanding storage needs as well as for tower server applications, this hardware RAID option supports RAID levels 0, 1, 5, 0+1, and Enhanced JBOD. It has 256MB of cache and an integrated 72-hour battery for protecting the RAID cache. The card occupies the top PCI Express slot (slot 4) and connects to the four internal drive bays. To enable your Mac Pro for hardware RAID, select the Mac Pro RAID card option and two or more hard drives in bays 1 through 4. Each RAID level has minimum requirements for the number of hard drives"

RAID 0 (striping): Two to four hard drives; Maximum performance and capacity for the most demanding I/O requirements
RAID 1 (mirroring): Two hard drives, Maximum protection for critical data
RAID 5: Three or four hard drives, Data protection, up to 199MB/s of sequential read performance, and efficient capacity utilization
RAID 0+1: Four hard drives, A mirror of striped drive pairs providing performance and data protection

Apple adds:

"For optimal disk utilization in a RAID set, all hard drives should be the same size. Your Mac Pro system ships with each hard drive individually configured in the Enhanced JBOD level with Mac OS X installed on the drive in bay 1. Using Apple's RAID Utility software, you can migrate the drives into a RAID set without reinstalling Mac OS X or reformatting the drives, or you can customize your RAID volumes to meet your exact requirements."

Mac Pro RAID Card and Xserve RAID Card: Frequently Asked Questions (FAQ) (http://docs.info.apple.com/article.html?artnum=306231)
Mac Pro RAID Card: Capacity for four-drive RAID 5 configurations may be different than expected (http://docs.info.apple.com/article.html?artnum=306212)
__________________
The 304MB/sec. quoted by Apple is for a four drive striped RAID set. A striped RAID set will slow as the volume becomes full. By the time the RAID 0 is 80% full performance will be reduced to 215MB/sec. at 90% full, performance drops to 189MB/sec.

A hardware RAID does not provide faster performance than a software RAID. It does allow for optimal RAID 5 performance and it can reduce main CPU requirements by off loading RAID calculations to a secondary processor on the card.

Expect the Apple internal hardware RAID to provide performance very similar to a software RAID and add RAID 5 internal capabilities.

The problem with the Apple card is that an 8x PCIe slot is utilized to support only 4 drives. At a $999 price tag that is expensive and I would rather have a card that supported 8-12 drives. If Apple sold a RAID 6 card that supported the 4 internal bays plus provided two mini SAS connectors on the rear of the PCIe card for supporting 8 more drives for a total of 12, that would be a very cool product and certainly worth the $999 premium.

12 drive RAID 6 performance is awesome and provides twice the redundancy of RAID 5. AMUG replicated this setup using an external 12 bay NORCO DS-1240 and a Areca ARC-1231ML card. RAID 6 average write performance with a hot spare was 596MB/sec. You can see the review here:

http://www.amug.org/amug-web/html/amug/reviews/articles/norco/1240/

Have fun!
- AMUG

TZ
08-11-2007, 04:49 AM
The Mac Pro (http://www.macgurus.com/forums/showpost.php?p=102425&postcount=1)
Reviews, Photos, and Performance (http://www.macgurus.com/forums/showpost.php?p=102436&postcount=6)
Migrating from PPC PowerMacs (http://www.macgurus.com/forums/showpost.php?p=102442&postcount=7)
Classic OS 9: Sheep Shaver (http://www.macgurus.com/forums/showpost.php?p=102569&postcount=8)

STORAGE

Apple Mac Pro RAID Card (http://www.macgurus.com/forums/showpost.php?p=111393&postcount=47)
RAID Options (http://www.macgurus.com/forums/showpost.php?p=108986&postcount=27)
Disk Drives (http://www.macgurus.com/forums/showpost.php?p=102432&postcount=4)

XEON CPU

8-Core Mac Pro (http://www.macgurus.com/forums/showpost.php?p=105935&postcount=19)
Upgrade to X53xx CPU Tutorial (http://www.macgurus.com/forums/showpost.php?p=110381&postcount=37)
Xeon Server (http://www.macgurus.com/forums/showpost.php?p=105999&postcount=21)
Intel Xeon "V8" motherboard design (http://www.macgurus.com/forums/showpost.php?p=109184&postcount=29)
Intel's CPU Roadmap: (http://www.macgurus.com/forums/showpost.php?p=108665&postcount=26)
Intel multi-core/multi-thread compilers (http://www.macgurus.com/forums/showpost.php?p=106283&postcount=22)
Java Development (http://www.macgurus.com/forums/showpost.php?p=110400&postcount=38)
What is "quad core?" (http://www.macgurus.com/forums/showpost.php?p=106814&postcount=23)
Over-clocking Xeon (Windows XP SysTool) (http://www.macgurus.com/forums/showpost.php?p=110471&postcount=40)

MEMORY:

DDR2 PC2-5300 ECC 667MHz FB-DIMM (http://www.macgurus.com/forums/showpost.php?p=109894&postcount=31)
DDR3 Memory (http://www.macgurus.com/forums/showpost.php?p=109398&postcount=30)
Demise of FB-DIMMs (http://www.macgurus.com/forums/showpost.php?p=105774&postcount=16)
Memory Bandwith (or lack) (http://www.macgurus.com/forums/showpost.php?p=110423&postcount=39)

3DSMAX9 (http://www.macgurus.com/forums/showpost.php?p=110912&postcount=41)
Adobe Photoshop (http://www.macgurus.com/forums/showpost.php?p=110271&postcount=36)
Avid Express (http://www.macgurus.com/forums/showpost.php?p=110122&postcount=33)
Final Cut Studio and FCP (http://www.macgurus.com/forums/showpost.php?p=110251&postcount=35)
Motu and FireWire Audio (http://www.macgurus.com/forums/showpost.php?p=107170&postcount=25)
Pro Tools (http://www.macgurus.com/forums/showpost.php?p=110225&postcount=34)

Benchmarks (http://www.macgurus.com/forums/showpost.php?p=109956&postcount=32)

FIRMWARE

Extensible Firmware Interface (EFI) (http://www.macgurus.com/forums/showpost.php?p=105821&postcount=17)
GUID/GPL Partition Table (http://www.macgurus.com/forums/showpost.php?p=104360&postcount=11)
SMC Firmware and Updates (http://www.macgurus.com/forums/showpost.php?p=105823&postcount=20)
Applying OS and Firmware Updates yields k/p on boot RAID (http://docs.info.apple.com/article.html?artnum=304511)

Graphics Cards and Video (http://www.macgurus.com/forums/showpost.php?p=102435&postcount=5)
PCI Host Controllers (http://www.macgurus.com/forums/showpost.php?p=109063&postcount=28)

Power Consumption (http://www.macgurus.com/forums/showpost.php?p=104345&postcount=11)
Maximum Operating Temperatures (http://www.macgurus.com/forums/showpost.php?p=104655&postcount=14)
Hardware Monitor (http://www.macgurus.com/forums/showpost.php?p=105959&postcount=20)
Cisco VPN panic at shutdown (http://www.macgurus.com/forums/showpost.php?p=102620&postcount=9)
Mac Pro downloads search: (http://search.info.apple.com/?search=Go&lr=lang_en&kword=&type=kbdload&newstype=&q=Mac%20Pro)
Mac Pro Technical Specifications (http://support.apple.com/specs/macpro/Mac_Pro.html)
Mac Pro: How to reset the System Management Controller (SMC) (http://docs.info.apple.com/article.html?artnum=304123)
Do-It-Yourself Manuals (http://www.apple.com/support/manuals/)