G5 2.7GHz

[kaye]

The 11 drive RAID screenshot results dumbed down for the web (ZoneBench length x width also reduced):



k

[kaye]

Ran the Pshop tests on the 11 drive RAID.
512MB, 1.5GB, and 3.0GB file sizes the same results, just 0-2 seconds faster then my previous bests.

512MB, previous best 76s, this time 75s.
1.5GB, previous best 239s, this time 239s.
3.0GB, previous best 593s, this time 591s.
6.0GB, previous best 1859s, this time 1815s.
If you were working with a lot of that size file or if much larger file size, then it would make sense to build the fastest RAID affordable. k

[kaye]

There is a new version of Cinebench 9.5, optimized for the latest Mac OpenGL.

I have combined my previous test with the latest I'll call rev1:

CINEBENCH 9.5//9.5rev1
*************************************************
Tester : kyum//same

Processor : G5-2.7DP//same
MHz : 2700//same
Number of CPUs : 2//same
Operating System : 10.4.5//10.4.6

Graphics Card : X800XT//same
Resolution : 1920x1200//same
Color Depth : Millions//same

*************************************************
Rendering (Single CPU): 414 CB-CPU//413
Rendering (Multiple CPU): 753 CB-CPU//749

Multiprocessor Speedup: 1.82//1.81

Shading (CINEMA 4D) : 405 CB-GFX//411
Shading (OpenGL Software Lighting) : 1177 CB-GFX//1374
Shading (OpenGL Hardware Lighting) : 2150 CB-GFX//3263

OpenGL Speedup: 5.31//7.94
*************************************************
Note that the Shading results are where the speedup occurs. k

[kaye]

New version of QuickBench and it is much improved and changed. Addresses about all of the requests I and others asked for on the Intech Forum. Here is the text readout (multiple tests are possible before QB3 completes):
QuickBenchª 3.0 Test Results
©2000-2006 Intech Software Corp.
Test file created on Tuesday, May 9, 2006 at 6:39:11 AM
Test Volume Name: 4xRap1500
Test Volume Type: MacOS Extended
Test Volume Size: 558.902 Gigabytes
Test Volume Free Space: 558.811 Gigabytes
Allow Write Cache Purges: Enabled

Standard Test Results:
Test Cycles: 1

Xfer Size Sequential Read Sequential Write Random Read Random Write

1 KByte 0.841 MB/Sec 0.747 MB/Sec 0.256 MB/Sec 2.476 MB/Sec
2 KBytes 13.49 MB/Sec 4.275 MB/Sec 3.754 MB/Sec 4.013 MB/Sec
4 KBytes 9.239 MB/Sec 16.52 MB/Sec 26.83 MB/Sec 5.385 MB/Sec
8 KBytes 34.12 MB/Sec 10.40 MB/Sec 42.86 MB/Sec 9.861 MB/Sec
16 KBytes 4.945 MB/Sec 33.69 MB/Sec 59.71 MB/Sec 23.37 MB/Sec
32 KBytes 20.709 MB/Sec 43.09 MB/Sec 7.359 MB/Sec 39.96 MB/Sec
64 KBytes 97.11 MB/Sec 27.375 MB/Sec 86.45 MB/Sec 75.56 MB/Sec
128 KBytes 112.78 MB/Sec 72.751 MB/Sec 84.000 MB/Sec 90.817 MB/Sec
256 KBytes 176.97 MB/Sec 136.28 MB/Sec 47.039 MB/Sec 140.62 MB/Sec
512 KBytes 230.31 MB/Sec 201.71 MB/Sec 79.549 MB/Sec 220.23 MB/Sec
1024 KBytes 221.90 MB/Sec 269.80 MB/Sec 153.13 MB/Sec 252.07 MB/Sec

Extended Test Results:
Test Cycles: 5

Ext. Xfer Size Extended Read Extended Write

20 MBytes 274.50 MB/Sec 433.05 MB/Sec
30 MBytes 340.15 MB/Sec 433.19 MB/Sec
40 MBytes 339.20 MB/Sec 438.05 MB/Sec
50 MBytes 333.04 MB/Sec 433.59 MB/Sec
60 MBytes 280.91 MB/Sec 457.55 MB/Sec
70 MBytes 278.79 MB/Sec 461.69 MB/Sec
80 MBytes 270.29 MB/Sec 460.76 MB/Sec
90 MBytes 269.81 MB/Sec 461.53 MB/Sec
100 MBytes 267.10 MB/Sec 464.96 MB/Sec

Custom Test Results:
Transfer Size: 1 Gigabytes
Transfer Count: 3
Test Cycles: 3
Contiguous File Only: No

Maximum Measured Read Speed: 246.82 MB/Sec
Average Measured Read Speed: 243.67 MB/Sec
Minimum Measured Read Speed: 235.81 MB/Sec
Maximum Measured Write Speed: 343.22 MB/Sec
Average Measured Write Speed: 333.07 MB/Sec
Minimum Measured Write Speed: 317.74 MB/Sec

[kaye]

The Custom Test:

1024MB is the max. Plus more selections such as purging the write cache (slight pauses during the test).

The Extended Test:

Bar or line graph choices plus more selections. ZoneBench also is a new incremental version but it must all be under the hood because it looks the same to me. k

[kaye]

AltiVec Fractal Carbon v1.3 has been replaced with Power Fractal v1.4. So I ran a comparison test on my G5.
AFC
0.7sec/18433.1MF
PF
0.8sec/16897.0MF

Huh? I checked all of the default settings for both, Color Speed, Maximum Count, Zoom Factor, Window Size, all the same, but I could see a momentary pause at the start of the PF test. So I launched AFC and like as far back as I can remember the CPUs go to max during the test and retest and stay at max between tests. On the other hand, PF the CPUs go to near but not full max during test and between retests the CPUs go back to idle. That explains the difference. k

[kaye]

An update to OpenGL utilizing OpenGL Extensions Viewer v2.25 which displays OpenGL accelerator info, and apparently has had a Benchmark function that my old version of the Viewer did not have, which tests not only Status but also FPS.
http://www.realtech-vr.com/glview/download.html

G5-2.7GHz DP, 10.4.6, Apple 23" Cinema Display, Radeon X800XT ME 256MB set 1920x1200, Millions and within Viewer 1920x1200x32bpp:

Total score = 2251 adding it up.

Same settings as above except Viewer set 1920x1200x16bpp:

Total score = 2405

No idea why, but enabling Quartz Extreme and Quartz 2D Extreme in Quartz Debug results in a dramatic increase in FPS. The above tests are with both enabled. I routinely use Quartz Debug. Most of the time those settings still do not survive a reboot or shutdown/startup but sometimes they do now. Latest version of Debug is 2.1.3. I downloaded the latest version of Xcode Tools v2.3 to get that and other tools. It's a big download, 915MB. Or you can just download CHUD v4.3.3, about 33MB.

I also suspect that a CRT would be faster than an LCD due to the inherent Response Time delay. Mine is 14ms. k

[TZ]

No idea why, but enabling Quartz Extreme and Quartz 2D Extreme in Quartz Debug results in a dramatic increase in FPS. The above tests are with both enabled. I routinely use Quartz Debug. Most of the time those settings still do not survive a reboot or shutdown/startup but sometimes they do now.

I thought they were pulled totally from OS X or Xcode because of problems and bugs. I use to enable it also but after reading about problems, left it off. I'll have to look and see.

[kaye]

TZ,

It is in the latest Xcode 2.3 download update and in CHUD 4.3.3, both on the Developer site. k

[kaye]

G5 with 10.4.6. I ran the original 1080p BBC Motion Gallery Reel and the new 720p BBC Motion Gallery: Andes. Enabling Quartz Extreme and Quartz 2D Extreme does eliminate the peak readings but there are fewer FPS below 24. Pretty much a continuous 24FPS.

http://www.apple.com/quicktime/guide/hd/
BBC Motion Gallery Reel
1080p
1920x1080
H.264
AAC
1:33
93.3MB download

BBC Motion Gallery: Andes
Andes is 720p, 1266x720
Video Coded H.264
Audio Coded AAC
4min 22sec
195.8MB download

I ran both at Full Screen 1920x1200, Millions. As soon as Full Screen in QuickTime is selected from the Menu Bar, the movie starts and there is black surrounding the movie. I did a Get Info (command-I) to get the Movie Info to overlay the movie and then clicked that window to get a screenshot later.

Without Quartz Debug, BBC Motion Gallery Reel, example:

A few as low as 23.33FPS and two at 24.67FPS, remainder 24FPS.

Without Quartz Debug, Andes example:

One 23.80FPS, two 24.67FPS, remainder 24FPS.

Apple recommended Hardware Configurations for H.264 High Definition (HD) Playback

To play high definition video, a large amount of data must be processed by your computer. A powerful system will deliver the best playback experience.

For 1280x720 (720p) video at 24-30 frames per second:
QuickTime 7 for Mac OS X:

* 1.8 GHz PowerMac G5 or faster Macintosh computer; 1.83 GHz Intel Core Duo or faster
* At least 256MB of RAM
* 64 MB or greater video card

For 1920x1080 (1080p) video at 24 frames per second:
QuickTime 7 for Mac OS X:

* Dual 2.0 GHz PowerMac G5 or faster Macintosh computer; 2.0 GHz Intel Core Duo or faster
* At least 512MB of RAM
* 128MB or greater video card
k

[kaye]

Comparing my original MLIII (7B300S0) with a new 7LMLIII (7L300S0). Both drives 300GB and exactly the same size, SATA 150, journaled, and Spotlight disabled when tested.

Empty test for MLIII in May, 2005 was QB2.1 and OS10.4. All other tests are QB3.0 and OS10.4.6. 7LMLIII was zeroed out and and initialized in DU and exercised thoroughly before testing the next day. For the QB3.0 test, I used settings exactly the same as QB2.1, none of the extra bells and whistles. 7L is in the A (upper) bay, old drive in the B (lower) bay.

MLIII - empty:
QuickBench 2.1 Test Results File
2003 Intech Software Corp.
Test file created on Friday, May 13, 2005 at 6:16:44 AM
Test Volume name: MLIII-16MB
Extended Test Size: 100 MB Read: 55.749 MB/sec Write: 71.874 MB/sec

7LMLIII - empty:
QuickBench 3.0 Test Results
2000-2006 Intech Software Corp.
Test file created on Wednesday, June 14, 2006 at 1:12:33 PM
Test Volume Name: 7LMLIII-16MB
100 MBytes 67.735 MB/Sec 71.505 MB/Sec
So for this test the new 7L drive tho new is definitely faster in Read and the same in Write.

Continuing with the 7L still empty (and no comparison with the old drive since it was my boot drive), the QB3.0 1GB test:

Custom Test Results:
Transfer Size: 1 Gigabytes
Transfer Count: 3
Test Cycles: 3
Contiguous File Only: No
Allow Write Cache Purges: Enabled

Maximum Measured Read Speed: 60.706 MB/Sec
Average Measured Read Speed: 60.474 MB/Sec
Minimum Measured Read Speed: 59.087 MB/Sec
Maximum Measured Write Speed: 63.608 MB/Sec
Average Measured Write Speed: 63.368 MB/Sec
Minimum Measured Write Speed: 63.065 MB/Sec

With SuperDuper, I cloned the old drive to the 7L. I had used iDefrag on the old drive a few days earlier so both were pretty much the same after cloning which took 58:35 for about 104GB, both now about 37% full.

MLIII - 37% full:
Test Cycles: 1
Allow Write Cache Purges: Enabled
100 MBytes 59.866 MB/Sec 64.927 MB/Sec
7LMLIII - 37% full:
Test Cycles: 1
Allow Write Cache Purges: Enabled
100 MBytes 60.350 MB/Sec 65.126 MB/Sec

MLIII - 37% full, 1GB:
Custom Test Results:
Transfer Size: 1 Gigabytes
Transfer Count: 3
Test Cycles: 3
Contiguous File Only: No
Allow Write Cache Purges: Enabled

Maximum Measured Read Speed: 54.491 MB/Sec
Average Measured Read Speed: 53.764 MB/Sec
Minimum Measured Read Speed: 53.439 MB/Sec
Maximum Measured Write Speed: 57.357 MB/Sec
Average Measured Write Speed: 56.644 MB/Sec
Minimum Measured Write Speed: 56.257 MB/Sec


7LMLIII - 37% full, 1GB:
Custom Test Results:
Transfer Size: 1 Gigabytes
Transfer Count: 3
Test Cycles: 3
Contiguous File Only: No
Allow Write Cache Purges: Enabled

Maximum Measured Read Speed: 55.692 MB/Sec
Average Measured Read Speed: 55.392 MB/Sec
Minimum Measured Read Speed: 53.579 MB/Sec
Maximum Measured Write Speed: 58.237 MB/Sec
Average Measured Write Speed: 58.160 MB/Sec
Minimum Measured Write Speed: 58.051 MB/Sec
k

[rwm]

That helps post 170 confused me.

[TZ]

Soho Tank 3.5" USB 2.0/FireWire 400 Aluminum Drive Case

Turn your 3.5-inch IDE drive into a high speed USB 2.0 and FireWire 400 external IDE drive. It features a hot swappable tray module and Oxford 911 Plus chipset. It can support a Serial ATA (SATA) hard disk simply by installing an upgrade kit (not included).

This SohoTank External drive case is compatible with Windows and Macintosh systems and comes with USB cable, FireWire cable, AC adapter, and keys.

$105 with FW and SATA. Extra drive sled $30. But the one MacGurus sells uses an Oxford922 Chipset, this one does not, so hopefully they'll be back in stock soon.

[Nicolas]

Soho Tank 3.5" USB 2.0/FireWire 400 Aluminum Drive Case

Turn your 3.5-inch IDE drive into a high speed USB 2.0 and FireWire 400 external IDE drive. It features a hot swappable tray module and Oxford 911 Plus chipset. It can support a Serial ATA (SATA) hard disk simply by installing an upgrade kit (not included).

This SohoTank External drive case is compatible with Windows and Macintosh systems and comes with USB cable, FireWire cable, AC adapter, and keys.

$105 with FW and SATA. Extra drive sled $30. But the one MacGurus sells uses an Oxford922 Chipset, this one does not, so hopefully they'll be back in stock soon.

Made by http://www.raidsonic.de/en/pages/home/home.php

The SR6600-5S-U3 is truly a nice unit:
External RAID Subsystem with power supply for 5x SATA HDD, RAID 0, 1, 3, 5, Spare, JBOD, U320 interface, managment via Network & GUI Software

Also is the SR2900-5S-U3 internal unit: Internal RAID Subsystem for 5x 3,5" SATA HDD, RAID 0, 1, 3, 5, JBOD, Spare, U320 interface, managment via Network & GUI Software

I am using for of those at the moment:
http://www.raidsonic.de/en/pages/pro..._objectID=2546



Regards

Nicolas

[kaye]

TZ and Nicolas, I've been searching everywhere. In the USA it looks like WiebeTech has obtained exclusive rights to sell those units and everyone else is either out of stock or running out. The unit I need is model U6-1S-WBC (with SATA tray). I saw that cooldrives page and have visited 5 other sites. For instance
http://fwdepot.com/thestore/product_...oducts_id/1021
Cooldrives has this
http://cooldrives.stores.yahoo.net/e...-external.html
but it is not hotswap. I need something on the small side on top of my FW400 case and both on top of my G5 and still be under my desk. The U6 is only 1.25" height which would be ideal for me.

I want FW800 so that I am not using a SATA port. I have an unused FW800 port on the rear of this box.

Randy, what have I confused you on the QuickTime post? k

[rwm]

Randy, what have I confused you on the QuickTime post? k

Oh, I see it was not post 170. It had to be another thread because it had talk of SuperDuper and screen shots. When I was reading things it sounded that with the new drive a copy/backup went from 50+ minutes to under 5 minutes. I been busy this week. Had several windows open.

I enjoy following along these threads I know little about ... I'll never need this kind of equipment but I learn a lot from you guys and can can pass info onto forum members.

[Nicolas]

TZ and Nicolas, I've been searching everywhere. In the USA it looks like WiebeTech has obtained exclusive rights to sell those units and everyone else is either out of stock or running out. The unit I need is model U6-1S-WBC (with SATA tray). I saw that cooldrives page and have visited 5 other sites....

http://www.raidsonic.de/en/pages/pro..._objectID=2555

Mayb you can contact sales via email:
http://www.raidsonic.de/en/pages/contact/sales.php

Or mybe Rick is interested in sellin' those units too.

Regards

Nicolas

[kaye]

While I wait for the U6-1S-WBC, if Rick can get it, I'm going to pick up this Granite external which is quite small but will give me another backup with my removed OEM WD drive installed and will be offline except for backups or if necessary to be used. I don't even have room for the 2-drive model. k

[kaye]

Decided I would get a fresh start with this SATA OEM 250GB WD Caviar SE, Model WD2500JD series, 7200RPM and 8MB cache.

Ordered the Granite ALUM HS 1 bay External, an extra hot swap tray, and padded carrying case for tray. With UPS ground, total $112.85 and it arrived the next day since Granite is just down the road from me. Opened it up to look over everything and the included manual with good tips on how to hot swap which I don't do anyway. I picked the hottest day of the year to get it running and test it out. Observations and results:

1. The Granite page http://granitedigital.com/catalog/pg...swap1drive.htm says
These units come complete with everything needed including a 3' shielded SATA cable and power cable (no software needed for Macintosh, Win 98SE, ME, 2000, and XP).
I wondered about that when the order was placed since Granite external cables run $15-30 and depend on what your needs are. There was no cable in the shipment so yesterday morning I called Kenny at Granite to give him a heads up. Whoops, he said they would have to get the webmaster to fix that.
2. Single function blue Power ON LED, no drive activity LED.
3. Takes about 4 minutes to install the drive in the tray, screws included, and insert the tray into the external bay.
4. Hooked up the inline power supply and an extra external SATA cable I have to a SeriTek/1VE4. BTW, changed these 3 inline PS's I now have so that when shutdown they are not still powered up by the UPS. Constant 6-9 watts by them before, 24/7.
5. Hardware Monitor sees it thru the case, SeriTek, and drive, 30C/86F after on the desktop.
6. SMART status verified in DU.
7. Rebooted to it no problems, 35.0C/95.0F, boot seemed significantly faster than when drive was on mobo SATA.
8. In DU, zeroed out entire drive and initialed journaled, 37.0C/98.6F.
9. Tested QB3.0-100MB with no options
Ext. Xfer Size*******Extended Read*******Extended Write
100 MBytes*********52.821 MB/Sec*******60.718 MB/Sec
Now 40.0C/104.0F, room temp 85F, no air conditioning running
10.After lunch while G5 and Granite case were running, did a full Backup - all files in SuperDuper. Took 57:56, room temp still 85C, outside temp in shade 110F, 46.0C/114.8F.
11.Case never seemed more than warm to the touch even after SuperDuper. Shutdown, pulled the tray out part way for 10 seconds for the hard drive spindown. Drive was definitely warm but not hot in the tray. Let it cool down before putting it and the manual in the carrying case. Now extra backup off site in a fire proof safe. Works great. k

[TZ]

kaye, thought it better to discuss Maxtor drives over here ;-)

Latest entry today on SR forum:

I got VA111680 from Maxtor. I have not decided it is fixed this problem, but so far my 7 7V300F0 drives have know an uptime of 7 x 25 days 23.5 hours = 4365 hours of trouble free operation. Looks definitely better than VA111630 and it has now also passed the maximum uptime before failure I had with VA111670.

If you had waited and then gone with PCIe Quad, you would have been 'safer' to wait until NOW for controllers (other than maybe ATTO UL5D, maybe not). And, if you had waited this long, surely would have waited a few more months to see what and when (plus three months from SJ's best guess) for Mac Pro?

Notice how long it has taken to get even that far? and unfortunately it uses its own driver and RAID utility. Not good. Much like another PM bridge that uses java utility and damages boot blocks?

Barefeats tested a PCIe 2-channel SATA. In earlier articles we reported on the Sonnet and HighPoint 4 port SATA 3G host adapters that cost $300. Add the cost of the 5 bay enclosure with drives and we're talking big storage bucks.

CalDigit's FASTA-2e PCIe host adapter offers two external eSATA ports for $79. (There are other 2 port cards available but most of them have one 1 external and 1 internal port.)

We used two Maxtor MaXLine 7V300F0 drives (300GB 7200rpm 16MB cache) in the tests we ran. (A FirmTek SeriTek/2EN4 enclosure was used.) We ran the drives as single drives and as a software RAID 0 (striped) pair.

The FASTA-2e host adapter uses a Mac driver that comes with the card or can be downloaded (along with the user manual) at their website.

The original 300GB Maxline III purchased for my G5 is 7B300S0. The 300GB Maxline III recently added to replace the OEM WD drive is 7L300S0, both on internal SATA. I was thinking of adding a 300GB 7V300F0 to my new Granite Alum SATA case since I have an extra tray. I have an email into George at FirmTek with all of the specifics, G5 model, Granite case, and SeriTek/1VE4 controller. Sounds like I should just forget about it however. k

How both Maxtor and Seagate are having problems with drives dropping off with some controllers and trying to deal with various firmware on their drives. SATA drive firmware problems