PDA

View Full Version : PS CS3 Beta Benchmarks- PowerPC vs. Intel




longofest
Dec 15, 2006, 04:56 PM
Hey guys-

I'm noticing that a bunch of people have been having success having CS3 run well on their Intel-based Macs. I haven't heard to much from PowerPC users though, and I'm thinking about compiling some of these benchmarks you guys are getting into a story.

The thing is, I'd like us to run relatively equal benchmarks (everyone using the same settings). Since people seem to like the Retouch Artists test, lets run that as our benchmark and use the following settings:

RAM= 70%
History=1
Cache=6

Please be sure to run the test just after restarting your Mac, with no applications running in the background. Of course, include your Mac's specs in your post.

Although thread is in the PowerMac/Mac Pro forum, MacBook and MacBook Pro users are welcome to join in the benchmarking.



longofest
Dec 15, 2006, 05:16 PM
RAM= "1 GB" (for a 1GB machine, set to 100%; for a 2GB machine, 50%; etc...)

I may be incorrect in how this RAM setting works... It looks like according to this article (http://www.adobe.com/support/techdocs/322829.html), the amount of RAM photoshop uses will be different anyways based on the amount of memory the system uses. So lets just all set the level to 70%. (I will change the first post to reflect this).

Banalltv
Dec 16, 2006, 05:48 AM
Hey.

I'm using a MacPro Quad Xeon yaddayadda 2.0Ghz, 4x512 (2GB) Ram as supplied by Apple and the Ati x1900xt as supplied by Apple.

I just ran the RetouchArtists Speedtest, I think they say to set memory usage at 100% but I used 70% for this post like you asked and got the same first pass times either way.

Restart machine.

CS2:
First pass: 79 seconds

Quit and restart CS2: second pass: 77 seconds

Not quit CS2, close test image and re-open: third pass: 68 seconds.


Restart computer, start CS3:
First pass: 45 seconds

Quit and restart CS3: second pass: 44 seconds

Not quit CS3, close test image and re-open: third pass: 38 seconds.

I notice a nice sharpening of response and brush sensitivity with CS3, very happy.

longofest
Dec 16, 2006, 09:21 AM
thanks Banalltv! Good starter numbers...

Willis
Dec 16, 2006, 09:56 AM
wow... so a 2.66 would probably shave a few seconds of those on CS3, and a 3ghz would take a few more off that. *supposedly*

very good results though!

jhero
Dec 16, 2006, 02:41 PM
Ya, nice results :) All 4 cores maxed out?

massiv
Dec 16, 2006, 10:00 PM
ok..so a MP 2.66 3 gigs 74G Raptor with the required settings...

28 secs.

HibyPrime
Dec 16, 2006, 11:30 PM
dual g4 867 (MDD) 2gb ram, boot disk as scratch
3:16

MBP 2.0ghz CD 1GB ram
4:18

When I lowered the ram available to photoshop on the G4 to the same amount as the MBP (~710mb), it finished in 5:33

longofest
Dec 17, 2006, 09:15 PM
thanks guys! Now just looking for a Quad's numbers...

Karpfish
Dec 17, 2006, 09:54 PM
"
Quote:
Originally Posted by simie View Post
Quad G5 2.5 Ghz
5 GB DDR 2
500 Gigabyte HDD
OSX 10.4.8

Photoshop CS2
1 History State
100% Memory

Time 36.03 seconds

Stopwatch Used

Just run the test under CS
Time 35.92 seconds
CS 3 Beta with 1 history state and cache level 6
1st run 53.35 seconds
2nd run 40.50 seconds

The second run, is always faster than the first on my Quad G5, I have done this test multiple times and get the same results. Restarting the Mac makes no difference to the test"

jimmy2gig
Dec 18, 2006, 05:47 AM
Hi All!

I ran the Retouch Artist test on my Mac Pro and was to say the least blown away...

My Specs:
Mac Pro Quad 3 Ghz, 3 GB RAM, 74 GB Raptor, 500 GB 7200 rpm Scratch Disk and an ATI X1900 XT 512 MB.

When I ran the test with default settings in PS CS3, it took over 4 minutes! No idea how that happened but after a restart the results were a lot different when settings changed to 100% RAM, 1 history state and 4 Cache levels. It finished the test in just under 27 secs. I tried again with default settings and got 1 min 55 secs.

I had tried with CS2 but sorry, don't have those figures to hand. Hope this helps.

Banalltv
Dec 18, 2006, 05:54 AM
Ya, nice results :) All 4 cores maxed out?

Activity Monitor says CPU gets up as high as 92.25% User during this, and mostly stays around 60, 70, 80% range so I guess that's a yes.

If there's a more detailed way of checking let me know.

sbaik
Dec 18, 2006, 06:34 PM
I run the test on my new mac pro 2.66 and also a new 24" imac and finally my poor but lovely 15" powerbook 1.33 they are all running 10.4.8

Mac pro 2.66/2g/500wd/256m NVIDIA 7300

CS2: 1:02
CS3: 36 sec

24" imac 2.16/2g/250g/128m NVIDIA 7300

CS2: 1:55
CS3: 58 sec

15" Powerbook G4 1.33/1g/60(4200rpm)/64m ATI9700

CS2: 6:55
when connected to an external firewire800 as scratch the result was 6:12

I still have 1.8 Dual G5 untested and next week I'm buying a macbook so next week I can give u a full list of benchmarks.
I just want to tell u that the Mac pro is so fast that it runs CS2 like the 24" imac running CS3 as u can see from the results (only 4sec difference) also worth mentiong that the powerbook did run well until the Gaussian blur action which took him almost 3 minutes to finish:( but maybe its a ram issue.

clemn
Dec 18, 2006, 10:32 PM
copied from my post on original PS retouch artist test thread:

quad g5 (quadro fx, 16Gb non-ECC, 2 raptor 74Gb on seritek 2Se2 as boot, barracuda 750Gb as scratch)
hist level =1, cache =8, RAM 100%

CS2: 40'' 1st time, 33'' 2nd time
CS3beta: 41'' 1st time, 34'' 2nd time

so CS3beta is a little slower than CS2 on my quadG5, and overall no significant difference between MacPro and G5 (expecially at the same clock speed, see previous posts). Waiting for Leopard and its 64bit (will CS3final address more of my 16Gb of RAM ?!!), this CS3beta is not really interesting for the time being at least in terms of speed as far as PPCs are concerned.

longofest
Dec 18, 2006, 11:11 PM
http://www.macrumors.com/images/macrumorsthreadlogo.gif (http://www.macrumors.com)

Since Adobe posted a beta of Photoshop CS3 (http://www.macrumors.com/pages/2006/12/20061214171952.shtml), many users have been eager to see what kind of performance enhancements the universal binary software would bring to both Intel Macs and existing PowerPC systems.

Informal MacRumors.com benchmark testing was performed by various forum members with access to both Intel-based and PowerPC based systems and the beta Photoshop software using the RetouchArtists benchmarking test. These tests have shown significant improvement in Photoshop CS3 speeds vs. Photoshop CS2 speeds on Intel hardware. A few selected results are displayed for your convenience (full results are in this story's thread (http://forums.macrumors.com/showthread.php?t=261046)):

Model - CS2 Time (s) - CS3 Beta Time (s)

Mac Pro 2.66 GHz, 2GB RAM - 62s - 36s
Quad G5 2.5 GHz, 5GB RAM - 36s - 36s
Quad G5 2.5 GHz, 2GB RAM - unavailable - 42s

Smaller time is better.

The 2.66 GHz Mac Pro showed a nearly 60% speedup by upgrading to the CS3 beta. However, PowerPC systems appear to have not felt any boost in speed, and at least one PowerPC user had speeds actually decrease when upgrading to the CS3 beta.

Similar trends have also been noticed by BareFeats (http://www.barefeats.com/quad16.html), which posts a similar comparison and also includes their own benchmarks. Of note, BareFeat's 'MP6' test shows a greater disparity between the Mac Pro and Quad G5 than the Retouch Artists test.

However, in the end the Photoshop CS3 software is still BETA software, and firm conclusions about the performance of the product should be avoided until the final version is shipped in Spring 2007.

[ Digg This Story (http://www.digg.com/apple/Photoshop_CS3_Beta_Benchmarks) ]

longofest
Dec 19, 2006, 12:27 AM
pardon me for concluding some obvious points, but it is interesting to point out a few notable things:

-This benchmark is very RAM sensitive. So, the QuadG5 that had 5 Gigs of RAM and ran at the same clip as the 2.66 GHz Mac Pro with 2 GB of RAM probably had artificially inflated numbers. If both systems had the same RAM, the Mac Pro would have most likely won out.

Westside guy
Dec 19, 2006, 12:54 AM
I'm not sure that it's news when a native Intel app runs faster than its Rosetta predecessor. :D But if CS3 is no slower than CS2 on PPC, that's certainly okay.

However it doesn't necessarily mean a lot when some functionality is currently disabled on Intel (don't know if rendering brush size etc has any meaningful impact though).

Has anyone benchmarked Rosetta CS3 versus CS2? That might be a worthwhile additional datum.

gugy
Dec 19, 2006, 01:19 AM
wow, the Quad G5 still performs very strong. I am surprised.
I guess I can get good money if I put it on ebay as soon as the octo-core comes out.;)
Hopefully by February, the Octo is here.

Multimedia
Dec 19, 2006, 01:43 AM
Quad G5 is still a winner. Mac Pro shows no advantage over the original Quad G5 in this test. Still focused on getting 8 cores next. :) And I won't be selling my Quad G5 when I get the OctoCore. It's a keeper.

poppe
Dec 19, 2006, 04:07 AM
Does anybody have a serial code that I could use to download the beta. I really don't use photoshop all that much but have been doing a lot of DVD menu's lately and would really like to have it. However I don't have the money to buy it/its not universal yet, so if anyone would like to pass me there code so I could dl the beta.

(Please delete this post if it is against policy, and my accept my apologies)

nicksoper
Dec 19, 2006, 04:11 AM
My Macbook Pro (2.16 CD1) doesn't die anymore!

Photoshop CS3 beta seems much friendlier without Rosetta holding his/her hand!

japanime
Dec 19, 2006, 05:54 AM
I'm thrilled with the speed boost when using PS CS3.

However, because the brush cursors don't display properly, this beta is not something I can work with on a daily basis.

I applaud Adobe for getting this out to the public, but I wish they had fixed the cursors first.

Photek
Dec 19, 2006, 06:43 AM
Does anybody have a serial code that I could use to download the beta.

no.

..download it and you can run it for 30 days... no serial needed.

koobcamuk
Dec 19, 2006, 06:50 AM
no.

..download it and you can run it for 30 days... no serial needed.

I heard it was a 2 day trial upon re-opening the applicaiton.

Snark
Dec 19, 2006, 07:32 AM
I heard it was a 2 day trial upon re-opening the applicaiton.
It's a two day trial. The thirty day information was a mistake on Adobe's end; supposedly it has been corrected. Of course it's been out for more than twenty-four hours, so I'm sure there's a krack already out. ;)

Snark

xUKHCx
Dec 19, 2006, 07:34 AM
Of course it's been out for more than twenty-four hours, so I'm sure there's a krack already out. ;)

I too am sure there is a crack out there.

dante@sisna.com
Dec 19, 2006, 07:52 AM
Quad G5 is still a winner. Mac Pro shows no advantage over the original Quad G5 in this test. Still focused on getting 8 cores next. :) And I won't be selling my Quad G5 when I get the OctoCore. It's a keeper.

Amen, I have said this before in response to your comments in this area but I totally agree.

The Quad G5 is a very strong Box.

TheFuzz
Dec 19, 2006, 08:08 AM
I'm thrilled with the speed boost when using PS CS3.

However, because the brush cursors don't display properly, this beta is not something I can work with on a daily basis.


for some reason running cs3 under rosetta fixes this. i'm still not sure how they missed it though:confused:

failsafe1
Dec 19, 2006, 08:27 AM
Must have...final version....:eek:

japanime
Dec 19, 2006, 08:29 AM
for some reason running cs3 under rosetta fixes this. i'm still not sure how they missed it though:confused:

You are correct. However, the fact that it's a UB is the reason I downloaded it in the first place. If I'm forced to run it under Rosetta, I might as well use my CS2 version.

In any case, no matter which one I run, if it's under Rosetta, the fan on my MacBook is gonna be spinnin' like crazy...

Spagolli94
Dec 19, 2006, 08:51 AM
"
Quote:
Originally Posted by simie View Post
Quad G5 2.5 Ghz
5 GB DDR 2
500 Gigabyte HDD
OSX 10.4.8

Photoshop CS2
1 History State
100% Memory

Time 36.03 seconds

Stopwatch Used

Just run the test under CS
Time 35.92 seconds
CS 3 Beta with 1 history state and cache level 6
1st run 53.35 seconds
2nd run 40.50 seconds

The second run, is always faster than the first on my Quad G5, I have done this test multiple times and get the same results. Restarting the Mac makes no difference to the test"

I see you have your RAM set to 100%. If we don't all run the test with the same prefs, the test is useless. See original post... Regardless, I think it's obvious that G5 users won't see much of a speed bump. After all, the G5 has run CS2 very well all along. Especially the Quad G5. The MacPro may be a hair faster, but it's certainly not blowing it away like Apple marketing would have you believe.

lorductape
Dec 19, 2006, 08:54 AM
I don't know where they got those numbers... mine starts up much faster. of course, cs2 was much slower on my machine than their numbers.

Spagolli94
Dec 19, 2006, 09:03 AM
I'm thrilled with the speed boost when using PS CS3.

However, because the brush cursors don't display properly, this beta is not something I can work with on a daily basis.

I applaud Adobe for getting this out to the public, but I wish they had fixed the cursors first.

Agree 100%. I was actually going to give this a shot to do some real work. But then I went to use the brush tool on a mask. Uh oh. I can't work with it until they fix that. It's a shame - that one little bug ruins the whole app for me.

Now, I've tried to go back to a PSD file, get info, and change it's default app back to CS2. When I click "change all" it doesn't take. Any ideas?

longofest
Dec 19, 2006, 09:10 AM
It's a two day trial. The thirty day information was a mistake on Adobe's end; supposedly it has been corrected. Of course it's been out for more than twenty-four hours, so I'm sure there's a krack already out. ;)

Snark

When you first run the program, you are offered the opportunity to put in a serial number or do a "30 day trial". However, if you take the trial route, the next time you open it, it will warn you that you only have 2 days left in the trial. It's a bug. It's really just a 2 day trial.

longofest
Dec 19, 2006, 09:17 AM
I see you have your RAM set to 100%. If we don't all run the test with the same prefs, the test is useless. See original post... Regardless, I think it's obvious that G5 users won't see much of a speed bump. After all, the G5 has run CS2 very well all along. Especially the Quad G5. The MacPro may be a hair faster, but it's certainly not blowing it away like Apple marketing would have you believe.

Yeah... its a bummer that he didn't have his RAM set to what I specified, but its okay in that we got enough people who did that so it didn't matter. I didn't realize when I made the specs in the first post that Retouch Artists had instructions as well, which conflicted with my instructions :p

katanna
Dec 19, 2006, 09:59 AM
Here are my results:
http://show.imagehosting.us/show/1834788/0/nouser_1834/T0_-1_1834788.jpeg

Matthew

Whistleway
Dec 19, 2006, 10:21 AM
Regardless, I think it's obvious that G5 users won't see much of a speed bump. After all, the G5 has run CS2 very well all along. Especially the Quad G5. The MacPro may be a hair faster, but it's certainly not blowing it away like Apple marketing would have you believe.

I thought they touted that Mac Pro runs 4x times faster than comparable G5 or something to that effect?

kothrush
Dec 19, 2006, 10:34 AM
I am running this test on macbook (core duo, 2GB ram) with CS3. Everything goes fine till it hits to Gaussian blur and that part takes over 4 minutes. :(

Anyone had similar problem ? My RAM seems to fine when I do hardware test and CS3 does seem to use up to 1.3GB RAM while doing this test.

tbrinkma
Dec 19, 2006, 10:46 AM
I thought they touted that Mac Pro runs 4x times faster than comparable G5 or something to that effect?


Oh look. Someone who thinks 'up to 4x faster' means 'at least 4x faster' or 'always 4x faster'. Certain CPU operations are 4x faster, some are only 3x, 2x, 1.5x, etc. faster. Non CPU operations (HDD access speed, RAM access speed, etc.) are completely unrelated to the CPU, and have a large effect on application speeds, especially when you're dealing with large data-sets as is common with PhotoShop.

Even during the keynote where the macintels were announced, Jobs mentioned that not all components were that much faster, so the overall improvement would vary. 'Up to' provides an upper limit, not a base-line, or guaranteed minimum.

shawnce
Dec 19, 2006, 10:47 AM
I thought they touted that Mac Pro runs 4x times faster than comparable G5 or something to that effect?

It fully depends on the tasks you are running as how well a given computer core (G5 / C2) will perform. What Apple listed is in fact 100% correct for the measurements they made.

In my experience for the developer tasks I do the Mac Pro easily out performs my Quad PowerMac by a factor of 2 if not more.

kaeptn
Dec 19, 2006, 11:48 AM
Hi.
I've done the results with my MacBook. Core 2 Duo, 2 GHz, 2 GB Ram, 80GB HDD.
Settings as described in post #1.

Restarted OS X (10.4.8).
Launched PS
Opened Image
Started test, result: 63 secs

Closed PS
Launched PS
Opened Image
Started test, result: 62 sec

Quite good for a non-MacPro or Quad G5 or sth like this... :)

I love my baby! *gg*

Bye,
Fritz

Aniej
Dec 19, 2006, 11:52 AM
quick question: I thought the download was supposed to provide CS3 for a while (spring) to use as a demo, but it is only offering 30-day trial. Am I mistaken?
:confused:

SiliconAddict
Dec 19, 2006, 12:15 PM
Gah....The benchmarking systems are all over the freaking board. Tested off of FireWire? 2GB of RAM vs. 5GB of RAM, systems that don't have matched settings. The geek in me is crying right now. :(



Now, I've tried to go back to a PSD file, get info, and change it's default app back to CS2. When I click "change all" it doesn't take. Any ideas?

Yah, don't use beta software to do real work, but I'm pretty sure you meant constructive ideas so... I've got nothing. ;)

failsafe1
Dec 19, 2006, 12:24 PM
The geek in me is crying right now. :(

No! Don't cry remember the people over at Adobe love us all and want us to be happy.;)

iBookG4user
Dec 19, 2006, 03:23 PM
MacBook Pro
2.0GHz Core Duo
2GB RAM
80GB 5400RPM Hard drive used as scratch drive

Same settings as aforementioned in first post

Photoshop CS2
after reboot
3 minutes 9 seconds

Photoshop CS3
after reboot
1 minute 18 seconds

japanime
Dec 19, 2006, 05:13 PM
quick question: I thought the download was supposed to provide CS3 for a while (spring) to use as a demo, but it is only offering 30-day trial. Am I mistaken?
:confused:

For licensed users of CS2, the beta can be registered and will last until this spring, when it is set to expire sometime after the official release of CS3.

For those who don't have a CS2 license, the beta can be used for 2 days.

(Although there is a dialogue box that, on first launch, indicates there is a 30-day trial for unregistered users, the trial is actually limited to 2 days.)

AidenShaw
Dec 19, 2006, 08:16 PM
What Apple listed is in fact 100% correct for the measurements they made.
Apple has a pretty strong record (at least in the "second coming of the Lord God Jobs" era) of making performance claims that are somewhere between suspicious and clearly misleading. In this case, what a typical person hears and what a lawyer would hear from the same ad are in different ballparks.

Sometimes, Apple's claims are simply bald-faced lies ("first 64-bit desktop" for the G5 is on the top of the list for obvious outright falsehoods).

But, we all know that Apple's marketing department has no scruples....

JeffDM
Dec 20, 2006, 10:09 PM
Apple has a pretty strong record (at least in the "second coming of the Lord God Jobs" era) of making performance claims that are somewhere between suspicious and clearly misleading. In this case, what a typical person hears and what a lawyer would hear from the same ad are in different ballparks.

Sometimes, Apple's claims are simply bald-faced lies ("first 64-bit desktop" for the G5 is on the top of the list for obvious outright falsehoods).

It is a bit twisted, but the early G5 Power Macs were somewhere between a desktop and a workstation, other 64 bit machines that came out before were purely workstation type computers. I did have an Alpha Multia though, which was a little more in the vein of a network computer than a workstation or desktop.

Spagolli94
Dec 21, 2006, 08:48 AM
Yah, don't use beta software to do real work, but I'm pretty sure you meant constructive ideas so... I've got nothing. ;)

Haha! That's exactly my point. I don't want to use it for real work! I'm pretty much done trying it out but don't want to delete it off my machine because I never know when I might want to try out some new stuff, do another tutorial, etc.

But now I can't get the thing to stop opening. Every time I double-click a PSD, JPG or TIF in the Finder (or in Bridge CS2/CS3), the new version of PS will pop open. I've tried resetting that preference everywhere I can think of. I set it in Get Info and in the Bridge.

So for now, I'm just right-clicking and choosing "open with" but this technique is getting old FAST. Is anybody else having this problem?

JeffDM
Dec 21, 2006, 10:06 AM
Haha! That's exactly my point. I don't want to use it for real work! I'm pretty much done trying it out but don't want to delete it off my machine because I never know when I might want to try out some new stuff, do another tutorial, etc.

But now I can't get the thing to stop opening. Every time I double-click a PSD, JPG or TIF in the Finder (or in Bridge CS2/CS3), the new version of PS will pop open. I've tried resetting that preference everywhere I can think of. I set it in Get Info and in the Bridge.

So for now, I'm just right-clicking and choosing "open with" but this technique is getting old FAST. Is anybody else having this problem?

What I do when this type of thing happens to me, is make an archive of the offending program and delete the unarchived copy when I don't need it, and unarchive it when I do. I don't think I should have to do this, but that's the best I've found.

Rocketman
Mar 5, 2007, 12:00 PM
My takeaway is that you simply get the SAME SPEED as a good old PMG5 by waiting for this upgrade.

If you are a Pro you have NO INCENTIVE to by a Xeon MacPro at all since a PMG5 is exactly as fast AND far lower price. And more fully debugged.

It seems the PMG5 has really long legs.

http://www.smarter.com/---se--qq-power%2Bmac%2Bg5.html

Rocketman

JeffDM
Mar 5, 2007, 12:53 PM
My takeaway is that you simply get the SAME SPEED as a good old PMG5 by waiting for this upgrade.

If you are a Pro you have NO INCENTIVE to by a Xeon MacPro at all since a PMG5 is exactly as fast AND far lower price. And more fully debugged.

You won't have to wait much longer, apparently Adobe will launch CS3 this month.

That said, if you can use an existing machine, then that's probably going to be serviceable for quite some time. If your uses needs the the fastest that's available, then pretty soon, I think the Mac Pro will be the machine to have. It's also a lot more flexible in terms of storage and is a lot quieter too.

FireArse
Mar 5, 2007, 02:06 PM
My takeaway is that you simply get the SAME SPEED as a good old PMG5 by waiting for this upgrade.

It seems the PMG5 has really long legs.

It does seem those PPC chips were state-of-the-art. I have a Mac Pro sat here infront of me. Part of me regrets selling the Dual G5 for it...

Still, if only H.264 was as well optimised on either PPC or Intel...I wonder how long it took to work through every line of that code. You can bet those engineers at Adobe know PPC & Intel optimisations better than IBM and Intel!