[blackwind]

Hello. In any case, here are results that were posted on Ars Technica by Ashby, who remembered to set the performance to "Highest" in the Energy Saver panel. No other applications were running, a 50-MB file was used, and 1 GB of RAM was installed.

1 90 Degree Rotate .4 .8 .3
2 9 Degree Rotate 2.4 2.2 2.2
3 .9 Degree Rotate 2.1 2.1 2.1
4 1 Gaussian .5 .5 .5
5 3.7 Gaussian 1.9 1.9 1.9
6 87 Gaussian 2.8 2.8 2.8
7 1 pix USM .8 .9 .9
8 3.7 pix USM 2.4 2.4 2.3
9 10 pix USM 2.9 2.9 2.9
10 Despeckle .5 .5 .6
11 RGB-CMYK 3.6 3.5 3.5
12 60% Reduction .6 .5 .5
13 LensFlare 5.5 5.5 5.4
14 Color Halftone 3.6 3.6 3.5
15 NTSC Color 3.6 3.6 3.6
16 Accent Edge 14.0 14.0 13.9
17 Pointilize 21.4 21.4 21.4
18 Watercolor 30.8 30.8 30.8
19 Polarize Coordinates 4.2 4.2 4.2
20 Radial Blur 47.4 47.3 47.3
21 Lighting 2.8 2.8 2.8

Average total time: 153.4 seconds
Normalized score: 344

The normalized score is relative to a 1-GHz Athlon Thunderbird running Photoshop 6.0.1.

The 1.8-GHz G5 is, through the normalized score, faster than a dual-channel DDR-fed 1.8-GHz Opteron (scores 332), faster than the 2.2-GHz Athlon XP 3200+ (scores 332), and almost as fast as a 3.06-GHz Pentium 4 with Hyperthreading (scores 358).

With the total time, the 1.8-GHz G5 is faster than a 2.8-GHz Pentium 4 and almost as fast as an Athlon XP 2700+.

These scores, however, somewhat invalidate the 1.6-GHz G5 scores in the past, which were probably done without the user knowing about the performance settings in Energy Saver. Unless, of course, a 12.5% clock speed increase gives a 24.6% performance boost.

[businezguy]

It's looking promising. Things will get even more interesting when Photoshop 8 comes out (performance improvements expected) and Panther comes out (stil more performance improvements).

It's going to take awhile before we get the real picture. I think, for the most part, we will be pleased with the results. The only disappointing part is how long it has taken for us to find out what the REAL performance of the G5 is going to be like.

By the way, keeping the Mac Addict tests in mind, I bet it would have scored even better with 2 gigs of memory installed. Even with just a 50 meg file.

So, let's just say there are really 3 factors to determine the performance of the G5 for Photoshop:

1) We need Photoshop 8

2) We need a G5 with lots of Ram (1 gig may not be enough)

3) We need Panther (10.2.7 is clearly a temporary fix to get the G5 out of the door)

[fireside]

normalized score? explain please.

[jrod7350]

Originally posted by blackwind:
These scores, however, somewhat invalidate the 1.6-GHz G5 scores in the past, which were probably done without the user knowing about the performance settings in Energy Saver. Unless, of course, a 12.5% clock speed increase gives a 24.6% performance boost.

Yah, I had no idea what the effects of bus slewing would be (I heard it was just a rumor and didn't know that there was a control panel feature to set it to "maximum" performance), so it was set to default when I ran the tests on the 1.6Ghz G5.

[Eug]

1) We need Photoshop 8

2) We need a G5 with lots of Ram (1 gig may not be enough)

3) We need Panther (10.2.7 is clearly a temporary fix to get the G5 out of the door)

1) Photoshop 8 is a long way off.

2) 1 GB of RAM is enough for this bench. 2 GB would make no difference.

3) Some people are saying that Photoshop 7's PSBench doesn't run noticeably faster in Panther beta.

Remember people USE Photoshop and X.2.7 in the real world. No point in waiting for Panther and Photoshop 8.

normalized score? explain please

It gives each test equal value, so that one test that takes a long time won't totally skew the numbers. Some tests take 3 seconds while others take 40 seconds. One may argue that 40 seconds is real world and so that one shouldn't normalize it down, but OTOH, quite often the 3 second tests are used much more. ie. Nobody ever uses pointillize (a filter that takes a long time), but everyone uses unsharp mask (a test which is only 3 s in this bench) over and over again.

Remember also that these are 50 MB files. People quite often use several hundred MB files, or dozens of 50 MB files with batch actions.

[MindFad]

People are already beta testing the next Photoshop. It might not be that far off.

[TC]

Originally posted by Eug:
1) Photoshop 8 is a long way off.

October should bring major Adobe upgrades, Photoshop 8, Illustrator 11, Go Live 7, In Design 3.
Creative Studio bundles know as CS to match up with Macromedia using MX.

If they get the G5 optimization done in time they will take some ground on Macromedia and Quark.

[Mr Heliums]

Originally posted by Eug:
1) Photoshop 8 is a long way off.

In one way you're right: there will never be a Photoshop 8. But Photoshop CS is closer than you think.

[Eug]

I know the Photoshop 8 (or whatever it's called) is in beta, but most reports I've read said it was supposed to be a November/December release.

Anyways, Photoshop 7 is here, and that's what should be tested. Sure Photoshop 8 will have G5 speed improvements, but likely so will PCs and possibly G4s too. You can wait, but if you can wait, you could always just wait until February for your dual 2.4 GHz G5. There is always going to be TheNextBestThing™.

[businezguy]

I'm just saying, it's going to take awhile before we see the potential of the G5 platform. If "Photoshop 8" took 2 years to come out, then that is how long it would take for us to see the true performance of the G5.

In the meantime, I did not say we shouldn't benchmark and test the G5. If I did, let me know about it and I'll get some medical help for multiple personalities.

[businezguy]

By the way, those scores were impressive as things stand.

[KidRed]

Originally posted by Eug:
1) Photoshop 8 is a long way off.

2) 1 GB of RAM is enough for this bench. 2 GB would make no difference.

1) A few months is not that far off

2) Do you mod these forums? http://forums.macnn.com/showthread.p...hreadid=176556

[Eug]

Originally posted by KidRed:
1) A few months is not that far off

2) Do you mod these forums? http://forums.macnn.com/showthread.p...hreadid=176556

Not sure what you're talking about with the mod comment.

If you read the contents of the benches...

PSBench as performed by Ashby uses a 50 MB file. It is well-established that a machine with 1 GB is sufficient for this test, as long as Photoshop is set up correctly, and there isn't too much stuff running in the background. Indeed, one of the reasons for using the 50 MB file size is because most people do not have more than 1 GB and thus with a larger file size it'd be difficult to get sufficient numbers of people with the proper hardware to perform this bench properly. ie. They'd be thrashing the hard drive too much with say a 250 MB file.

OTOH, the Macaddict bench is a 512 MB machine using a 115 MB file. It is said you need approximately 5X the memory as the size of images you'll be using in Photoshop. 512 MB is clearly less than 5x115. 1.25 GB for example would have been sufficient however.

ie. Macaddict likely would have gotten the exact same score with 1.25 GB as they did with the 2 GB on the dual 2.0.