G5 2.7GHz

[Nicolas]

The one place I didn't look, and that is where you posted results...

Corporate help , thats why I like this forum!

TZ,

if you reply to a topic, there is nothing more to add, all questions covered!
Mostly

Seriously, the tremendous amount of work you've done here
Your truly a MacGuru!

Best regards

Nicolas

[TZ]

Well, thanks, but I just dabble when it comes to actually using a computer! To say nothing of having only an academic idea of Photoshop and what people do and use a Mac for. Sort of strange.

If you shave 3 seconds off "Step 10" is that something that happens dozens, hundreds, thousands of times in the course of a day, an hour? Do nanoseconds matter?

Rob feels that the Quad is going to be the machine to have if someone primarily uses Adobe Suites. Along with the other side dishes.

When I "run into someone" with 32-bit system wanting to run PS but not interested in RAID or getting a new system.... and has already "maxed out" RAM to 2GB... time to move on, or realize there is a brickwall ahead. Yet, a lot of people are happy or content and their needs are "modest" enough that a $7500 workstation setup isn't economical or feasible, and they have more time and not enough "clients" for their work to justify such a setup.

I am proud to see Apple put a Quad out there. Maybe it is "revision A" to test the waters and see what feedback they get. I'll bet not only graphics and video/audio production will like it, but programmers that use to have to wait for their compile, will get more done, or not have to use multiple systems. Which means we should benefit the fruits down the road. What is Tiger other than the work of better compilers and code if anything?

but would you believe that FW800 PCI suffers from slow writes, like 25MB/sec (half what it should be)? Will have to see if that is only one event, or "endemic" to the system.

It would be nice to know or have word on how the UL5D is moving forward. I was looking around and noticed that Maxtor's SAS drive is a different connector/interface, sports 16MB cache in their 10K model, 600MB/sec (Ultra600 or Ultra620?). Only "SAS" isn't a search word on ATTO.

But right now, the 2.7DP is "mature" and a solid system, and the fastest cpu Apple offers.

[kaye]

Thanks for that info Nicolas. I copied and printed it and am going to experiment.

I too just dabble at Pshop TZ. In fact, when I decide I need to run it, usually my RAID or RAIDs are shutdown and it is a small file anyway. BTW, I updated the PshopTest thread with my large file size tests. Some of the Tribal tests were with using 2 separate RAIDs which were slower so I did not include them. Worked from my notes and the test forms and not from the Tribal posts.

I have no idea how much time a heavy Pshop user can save but you will see in my latest summary of tests post that a heavy user of large files can save a bunch of time depending on what and how many functions and how many times they use particular functions. For those folks, max RAM affordable and more drives is better. As you say, progress on the Quad will be very interesting as will some Pshop CS2 tests on that machine. k

[kaye]

Nicolas,

I think I will stick with MTU 1492. That works good. I used Cocktail to set in and out to 65536. I forget what they were set at but much, much lower. I'm not sure but I think IPv6 in Prefs might have been on. It is definitely off now.

Running the DSLreports I get slightly better download speed 157.0KB/s and definitely better upload 36.8KB/s. In Terminal at 1492 on my G5:
round-trip min/avg/max/stddev = 0.062/0.074/0.085/0.006 ms
That's better than before. I'm a happy camper. k

[Nicolas]

Nicolas,

I think I will stick with MTU 1492. That works good. I used Cocktail to set in and out to 65536. I forget what they were set at but much, much lower. I'm not sure but I think IPv6 in Prefs might have been on. It is definitely off now.

Running the DSLreports I get slightly better download speed 157.0KB/s and definitely better upload 36.8KB/s. In Terminal at 1492 on my G5:
round-trip min/avg/max/stddev = 0.062/0.074/0.085/0.006 ms
That's better than before. I'm a happy camper. k

Kaye,

nice results!

BTW Some German ISP's are using 1488.



What test you are using?

Try this one (download only) and do three runs:
http://web.zdnet.de/mobile/speedtest...est500-wc.html

There are some tests which are not "that" accurate so check with two better three different tests.

Regards

Nicolas

[kaye]

Nicolas,

Tnx, typo fixed. I am using http://www.dslreports.com/stest and trying Seattle, SF, Los Angeles, and New Jersey. Usually LA is the fastest but obviously it varies. They have a link to other locations too, 327 locations.

This morning I fixed the iBook, IPv6 was on, and Cocktail for in and out to 65536. Default was set at 32768 for both. Slightly better download and much better upload at dslreports at all four locations but all were pretty busy. Similar improvement in Terminal.

I will try your link. k

[Nicolas]

Nicolas,

Tnx, typo fixed. I am using http://www.dslreports.com/stest and trying Seattle, SF, Los Angeles, and New Jersey. Usually LA is the fastest but obviously it varies. They have a link to other locations too, 327 locations.

This morning I fixed the iBook, IPv6 was on, and Cocktail for in and out to 65536. Default was set at 32768 for both. Slightly better download and much better upload at dslreports at all four locations but all were pretty busy. Similar improvement in Terminal.

I will try your link. k

Kaye,

I tried the link you posted and guess what:
Seattle: 53KB/s- down and 5.4KB/s upstream
LA: 180KB/s- down and 12KB/s upstream

he he he but, I get 280KB or 0.280MB/s when I download something like the new Ubuntu Linux image

So, all this depends on:
1. traffic
2. Backbone connection from your ISP
3. Yes, Browser (try for yourself using Safari and lets say Firefox) also on the Cache settings of those.
4. some ISP are using line sharing which means not the full throughput all the time.
EDIT: 5. Java version and speed of running Applets
Best

Nicolas

[kaye]

Had my first full power failure early this morning. In the past I have had momentary failures like when the utility is switching power grids. Just enough to only get a chirp from the APC UPS and probably an LED but it happens so fast that by the time I look at the UPS there is no indication. But this one lasted somewhat less than a minute. Got very dark except for the monitor, then a PowerChute warning window indicating AC power failure. Quit Safari and Mail while watching the percentage display in the Menu Bar go down from 100%.

After less than a minute, maybe 50 seconds, the room light came back on, another PowerChute message appeared indicating that AC power was restored. The percentage was at 90% and starting to climb back up. No Burlys were powered up. I wish they were. Was running about 390W according to KILL A WATT before the AC failure.

Love this APC BACK-UPS RS 1500VA. k

[kaye]

Nicolas,

I tried the test 3x on your link and keep getting this:

Sie sind verbunden mit:
510000 KBit/s
62500 KByte/s

Daten (from the 3rd try):
Wenn Sie die reinen Daten interessieren, mit denen Ihre Bandbreiten berechnet wurde, finden SIe sie hier:
Download-Start: 1133115004694
Download-Ende: 1133115004702
Dauer: 0.008
Download-Größe in tausend Byte: 500
Geschätzte Geschwindigkeit (Kilobit pro Sekunde): 510000
Geschätzte Geschwindigkeit (Kilobyte pro Sekunde): 62500

The Spedometer thermometer bar is pegged red all the way at the top.
What does it all mean? k

[Nicolas]

I was testing this APC on a friends G3 server (G3MT 366MHz now) plus HW RAID (six ATA HD's) plus 15" LCD plus two FW drives, by pulling the powerplug, can't say the wattage but, powerchute pops up and after 17 minutes and 22% left I plugged it back in.

It is also a APC BACK-UPS RS 1500VA (BR1500I).

Regards

Nicolas

[Nicolas]

Nicolas,

I tried the test 3x on your link and keep getting this:

connected with:
510000 KBit/s
62500 KByte/s

Time: 0.008
Download-size in thousand Byte: 500
estimate Speed (Kilobit per Second): 510000
estimate Speed (Kilobyte per Second): 62500

The Spedometer thermometer bar is pegged red all the way at the top.
What does it all mean? k

Wow that is speedy!!!!

Some tests are not working from every location.
Try again, just click on the link and wait till you see results, if this happens again, it is not working from your location.

did you hit the "back" button of your browser?
than this happens:


Should look like this if it is running OK.


Try this one:
http://www.speedcheck.arcor.de/index.html
You have to start it with the left button "Speed-Check Starten"
the upper is download speed, the lower is upload speed

Regards

Nicolas

[TZ]

There was a 15 minute period today where MacGurus went unreachable, then came back. Other sites I tried were okay... So maybe the west coast was doing weekend maintenance/upgrade perhaps.

Two weeks ago two main backbone hubs were down, one in Washington DC and the other in New Orleans, that had a lot of people without service.

10% down so quickly! But nice that you could sale through unharmed. Which is why I like having the extra battery pack add-on. I think it would take an hour or two before it even got to using the main APC.

On Photoshop: anyone have a test with just one Raptor or 15K drive out there? See the difference between that and normal RAID.

[kaye]

Nicolas,

After each test, I quit Safari and rebooted. I just ran the test again once after emptying Safari cache and got exactly the same results. I can see constant activity on the router LEDs from beginning to end of the test. G5 has been shutdown since late morning. I'll try the other test tomorrow. k

[kaye]

TZ I'll see if I can run a single Raptor as scratch tomorrow. Oh wait, that 6GB file gets up to 76GB during the test. How about 2x Raptors? k

[TZ]

Kaye,

Wow! that is a lot of storage getting eaten up. Reason I asked is to help resolve what single disk drive would make the best candidate for PS scratch.

Boot drive OS, apps, and scratch with the PS file only.
2nd drive with a partition for scratch
dedicated drive for scratch

MLIII 300GB
Hitachi 500GB
Raptor 74GB
Even $800 Atlas 10K V SAS or SCSI

If someone is just doing small files, it probably isn't worth asking.

Some folks want the cheapest solution. Or try to go with Adobe's guidelines.

[kaye]

TZ,

My scratch drives are 18GB X15.3, 36GB Atlas 15K II, 74GB Raptor. The 2 internal G5 drives are OEM WD Caviar SE SATA 250GB-8MB, 7200 RPM which is slow by comparison to my main boot drive Maxtor MaXLine III 300GB-16MB. Best I can do is run the 6GB test on the Maxtor. Be interesting tho to see how long that takes, I have no clue. Maybe today but if not, then tomorrow. k

[kaye]

Nicolas,

I did the ZDnet test once again this morning and like I said previously, I am seeing constant activity on my router LEDs so I know the test is being done but the results are way too high unless ZDnet.de is located in my crawl space under the house with about a 4" pipe directly connecting. This time it is showing even faster.

k

[Nicolas]

Kaye,

so this test is useless if your not in germany, good to know.

The only thing why this is not working is when your ISP is using a really fast proxy.

Would be great if this happens with the second test too.

And once again, those tests are not "that" accurate.

Thanks for testing!

Best

Nicolas

[kaye]

I posted to the Sample PshopTest Results 6GB file single drive performance with my boot MaxLineIII drive. It is in the B (lower) drive position. Before rebooting I checked its temperature at 40.0C. After completion of the test, launched Hardware Monitor, still at 40.0C. Was curious because the poor baby went thru a humongous amount of work during the test as the file sizes grew to 79+GB. k

[Nicolas]

Hello Kaye,

new results for Seattle, no answer why (the traffic bar was full, two points left)

Regards

Nicolas