[Zuffenhausen67]

Here's a quick summary of the results from Macwelt's test of the 1.6GHz "baby" G5. Keep in mind that these results are without the new Photoshop plugin and often with less RAM than any normal user would use and far less than the Athlon they used. In other words, the tests seem VERY limited, but (if you're like me and you're looking everywhere for B.Marks)...

Configs:
G5 Standard Config from the Apple Stores - 256 MB DDR333-RAM
Athlon PC - Athlon XP 2200+ with 1,83 GHz, 768 MB DDR266-RAM, 80 GB Ultra-ATA-100-HDD

In a Photoshop (non-G5-optimised) test using a script to work on a 10MB filewith various filters:
G5 256MB RAM: 1min 27 sec
G5 512MB RAM: 56 sec.
Athlon: 1 min

Cinebench tests
CPU-intensive Render-Test using the "CB"-value (proprietary--don't ask me)
Apples G5 with 1,6 GHz = 150 CB-Points
G5 withbei 512 MB = 165
Dual-1-GHz-G4-Mac =96
Athlon = 221

UT2003
"Antalus" Botmatch and Flyby
G5 with Geforce FX 5200
Athlon with ATI Radeon 9700 Pro
1024x768, 32-Bit, max details

Botmatch on G5 = 19 fps avg.
Flyby on G5 = 56 fps
Athlon Botmatch = 31 fps
Flyby = 124 fps

B4 any harsh remarks are submitted--yes these test seem flawed due to limited RAM and dramatically different GPUs, but i am providing them as they are the 1st indep. tests on the 1.6GHz that i have seen...in other words--purely informational:-)

[idyll]

Sounds good. But it looks like my 1.6 G5 is going to be as fast as the PC it is replacing (a 2.2GHz P4). Ah well, at least I won't have to use Windows XP anymore.

[blackwind]

The 1.6-GHz G5 outperforming the Athlon in Photoshop (even unoptimized) is expected.

I don't know why, but Cinebench 2003 is slow on Macs compared to Cinebench 2000. My Dual 1.25-GHz G4 matched a 2.66-GHz Pentium 4A in Cinebench 2000. Now, my Dual 1.25-GHz G4 just barely beats a 2.2-GHz Pentium 4 in Cinebench 2003. As such, the Athlon outperforming the G5 was expected. (By the way, a single 1-GHz G4 would score 96... a Dual 1-GHz G4 scores 176.)

Unreal Tournament 2003 relies more on the graphics card than the processor speed, so the Athlon with a Radeon 9700 Pro (the fastest graphics card up until only fairly recently) should beat the G5 with the GeForce FX 5200 (a pitifully bad graphics card that is just barely a match for the GeForce 4 MX).

Adobe is going to release a G5-optimized plug-in for Photoshop very soon now, and Maxon engineers claim to be working on an optimized version of Cinema 4D XL for the G5. As such, don't be disheartened by these initial results.

[Mediaman_12]

Originally posted by blackwind:
Maxon engineers claim to be working on an optimized version of Cinema 4D XL for the G5.

There was a C4D Update only yesterday, I wonder if that includes the G5 Optimisation? The Photoshop one being due any day now.

[mac freak]

Ok, you can't pit a GFFX5200 against a 9700 Pro. It's not fair.
Additionally, as already stated, the Athlon has 3x the RAM.
The PS plugins are Athlon-optimized but not G5 optimized.

In short, these are almost useless benchmarks.

[slider]

Originally posted by mac freak:
Ok, you can't pit a GFFX5200 against a 9700 Pro. It's not fair.
Additionally, as already stated, the Athlon has 3x the RAM.
The PS plugins are Athlon-optimized but not G5 optimized.

In short, these are almost useless benchmarks.

A this was already said, did you read the thread?

[PixelPete17]

Having less RAM in the G5 makes no difference to Photoshop in this case as the file size was only 10MB. Unfortunately OS X probably benefits from more RAM so that would've yielded (theoretically) an increase in speed, but the G5 was tested with 512 MB RAM as well, and sure enough a speed increase was seen, but any more RAM than that should not make a difference with only a 10MB file. The only reason you saw an increase in speed going from 256 to 512 was because of the bloat-o-matic OS X code that is now over two years old and still turtle-ish to many.

So far, to my own shock, the G5 1.6 is looking to be a dog.

Well see soon enough,

Pixel Pete

[idyll]

Looking to be a dog as in a good thing or a bad thing? I think it's pretty good for a 1.6 GHz processor, but could be better.

[blackwind]

If there was a C4D update yesterday, then it is not the version that is G5 optimized. The update yesterday simply allows the G5 to be able to run C4D without crashing (as it did with Cinebench 2003). I am pretty sure that a G5-optimized version is still in the works (although I suspect that it won't get very far, considering the amount of optimizing Maxon engineers did for the G4... that is to say, very little).

As for the G5 being a dog, it still will take a while to determine whether it is or not. Consider that applications are still in the process of being updated to take better advantage of the G5. Furthermore, consider the fact that most G4 optimizations slow the G5 down to molasses (except for Altivec without using vec_dst).

As previously mentioned, the G5-optimized version of Photoshop will give the G5 a much-needed performance boost. As such, the 1.6-GHz G5 should comfortably outperform the 1.83-GHz Athlon XP. As for RAM, the more RAM that Mac OS X has, the better, even for only a 10 MB file.

Cinema 4D XL is fairly optimized for X86, but is not that optimized for Macs. As someone on Ars Technica pointed out, the most time consuming code used by Cinema 4D XL is actually 32-bit, unlike what Maxon engineers claimed earlier.

A better raytracer comparison would probably come from Newtek's Lightwave, which is heavily optimized for both Macs and PC's. I am certain that a G5-optimized version of Lightwave is in the works right now.

Finally, as everyone else has pointed out, the Unreal Tournament 2003 test has just shown us that we must avoid the GeForce FX 5200 at all costs and go for the Radeon 9700 Pro.

As such, of the two more useful tests:

1) G5-unoptimized Photoshop: The 1.6-GHz G5 performs like a 1.96-GHz Athlon XP. Considering that the fastest Athlon XP currently runs at 2.2-GHz, that is not so bad. When the G5-optimized Photoshop is available, the 1.6-GHz G5 will probably beat the 2.2-GHz Athlon XP 3200+. (This also means that a 1.8-GHz G5 will defeat a 2.2-GHz Athlon XP 3200+, assuming linear scaling, even with the unoptimized Photoshop.)

2) Cinebench 2003: The 1.6-GHz G5 performs like a 1.65-GHz Pentium 4. Something tells me that there is something wrong with this benchmark...

[Severed Hand of Skywalker]

Originally posted by mac freak:
In short, these are almost useless benchmarks.

Just like all the other benchmarks that are flying around.

[Leonis]

Originally posted by blackwind:
If there was a C4D update yesterday, then it is not the version that is G5 optimized. The update yesterday simply allows the G5 to be able to run C4D without crashing (as it did with Cinebench 2003). I am pretty sure that a G5-optimized version is still in the works (although I suspect that it won't get very far, considering the amount of optimizing Maxon engineers did for the G4... that is to say, very little)....

First. Where the hell is that update? The latest one I found is the Wacom tablet patch for the intergated BodyPaint that was posted two weeks ago.

Second. There had been THOUSANDS of talks in the Post Forum (where Maxon people talk to their users) regarding G4 vs Athlon vs P4 issue.

They all told the real truth:

In the last 4 years the x68 processor had advanced a lot in terms of raw speed while the PPC stalled (thanks to Moronola )

Keep these in mind:

G4 is very limited as a "high performance" chip. AltiVec can't do double precision FP (so it's useless for rendering). G4's FPU unit is weak. System bus is too slow compare to PC.... etc etc etc etc.

You can't blame them for not optimizing the app (actually they did optimize the renderer to roughly 14% faster on G4 while over 40% on P4) because there's not much they can do for such a limited processor.

The Mac/PC ratio for Cinema is very close to half half. G5 is a much more promising chip than the Moronola's G4. Therefore I believe they will do whatever they can to make Cinema "fly" on G5 systems.

Oh yeah. In case if you don't know....Cinema came from Amiga platform, just like Lightwave.

Also. These two apps are the very first to be OS X native.....and you say they don't commit to the Mac platform?

[saru boy]

Originally posted by blackwind:
1) G5-unoptimized Photoshop: The 1.6-GHz G5 performs like a 1.96-GHz Athlon XP. Considering that the fastest Athlon XP currently runs at 2.2-GHz, that is not so bad. When the G5-optimized Photoshop is available, the 1.6-GHz G5 will probably beat the 2.2-GHz Athlon XP 3200+. (This also means that a 1.8-GHz G5 will defeat a 2.2-GHz Athlon XP 3200+, assuming linear scaling, even with the unoptimized Photoshop.)

I don't know how accurate your scaling between the Athlon processors would be. You can't really compare the megahertz ratings between an Athlon XP2200 and an XP32000. The Athlon XP2200 used in this test is built on the Thoroughbred core, whilst the Athlon XP3200 is built with the Barton core (ie better performance per mhz than Thoroughbred).

[biposto]

I wonder how the Athlon64 performs compared to a G5 1.6.
Guess we will find out next month.

[axlepin]

Originally posted by mac freak:
Ok, you can't pit a GFFX5200 against a 9700 Pro. It's not fair.
Additionally, as already stated, the Athlon has 3x the RAM.
The PS plugins are Athlon-optimized but not G5 optimized.

In short, these are almost useless benchmarks.

maybe not completely *useless*...it shows that, even with both arms tied behind its back, the G5 puts in a very respectable showing... 256 MB of RAM for OSX? So then, how much was left over for the apps? Imagination RAM?

C'mon! for it to score so well with so little RAM demonstrates what a killer the G5 is.

IOW, it wasn't well-matched to that Athlon, and did NOT get its heinie kicked. I think that's part of the value of the G5...apple KNOWS that Macs are judged in areas where software is not optimized for the Mac, so this chip comes along like Godzilla, eats the Empire State Building, starts mawing on New Jersey, etc..

Imagine, therefore, the same shootout between comparably-equipped computers!

axle

[sleepyrenderer]

The botmatch score may be telling though. That's almost completely CPU bound.

[blackwind]

Firstly, I will have to eat crow about implying that Maxon did not support the Mac platform. Indeed, Cinema 4D XL was one of the first Mac OS X applications. (I can't believe that I actually forgot about that...)

Also, I believe that you are correct about Maxon not really wanting to put too much work with the G4, since the G4 did have a very weak double-precision floating-point implementation, especially compared to the Athlon.

In any case, I humbly apologize for offending Maxon engineers and users of Cinema 4D XL.

In response to the Throughbred and Barton cores, I would like to note that while the per-MHz performance of Barton is slightly better than Throughbred (about 5% better per MHz), that would still allow a 1.8-GHz G5 to perform approximately the same as a Athlon XP 3200+ and beat an Athlon XP 3000+ using the G5-unoptimized Photoshop, at least using the same testing procedure as German Macworld.

As for Athlon 64, it is supposed to have an additional 5% performance boost per-MHz over the Barton core, similar to the Opteron. (Opteron's actual performance boost is much higher due to much larger caches and such. Opteron's core, by itself, gives a 5% boost over Barton, according to AMD.)

As for "imagination RAM", I believe the more popular term would be "virtual memory".

Even with equal graphics cards, though, there is no guarantee that the G5 would defeat the Athlon XP simply because OpenGL drivers for Mac OS X are currently inferior to those used by other PC's.

[tooki]

Am I the only one that wants to see comparisons with older Macs (G4s and G3s), not with PCs that mean nothing to me?!?

tooki

[osiris]

Originally posted by tooki:
Am I the only one that wants to see comparisons with older Macs (G4s and G3s), not with PCs that mean nothing to me?!?

tooki

I'd say Yes... and no. The same app/game compared with different platforms/systems can tell us whether or not we 're blowing hot air, or, if in fact they are faster. Apples & Oranges, but with the same task. The faster fruit wins.

Didja see Barefeats comparison with a dual G4 vs G5?

Barefeats G4/G5 comparison

personally I'd be too embarrassed to see my G4 400 crushed..

[Doc Juansinn]

Originally posted by tooki:
Am I the only one that wants to see comparisons with older Macs (G4s and G3s), not with PCs that mean nothing to me?!?

No, you're not alone. I too would like to see benchmarks against G4 and G3 processors. These would have more relevance to my own computing experience than would comparisons against a PC that I have no intention of ever using. I want to know how much faster it is than what I have so that I can decide if it's a reasonable expense now for me. Just being able to say that my computer is "faster" than a PC isn't reason enough.

[Powaqqatsi]

Originally posted by osiris:
Didja see Barefeats comparison with a dual G4 vs G5?

Barefeats G4/G5 comparison

I think there is something weird with those results

[awcopus]

Originally posted by tooki:
Am I the only one that wants to see comparisons with older Macs (G4s and G3s), not with PCs that mean nothing to me?!?

tooki

No, not at all. I have no interest in it. I mean, why would I or any Mac user who has endured years buffetted by the incessant "hardware superiority" claims of PC users want to be able to substantiate, at long last, a dramatic rebuttal?

"No. My new Mac is faster than your PC. Here's the proof. How do you like them Apples!? Wanna buy one now, dontchya?"

--and--

"Boss, now's the time to upgrade our systems to Macs. They're faster and less expensive to maintain than any PCs." "Faster? No way." "Yes way. Way way."

That said, it will be nice to have more real-world application performance comparisons. Those are more compelling than these academically fascinating but otherwise irrelevant statistics.

As a Dual 1.42 owner, I am definitely also interested in comparisons between my machine and the G5s. Once software has been optimized for the G5 and both the G4 and G5 are running on Panther, these will be even more illuminating.

2003 has been by far the coolest year for Mac users in a long time.

[Dan-the-Man]

Originally posted by blackwind:
The 1.6-GHz G5 outperforming the Athlon in Photoshop (even unoptimized) is expected.

I don't know why, but Cinebench 2003 is slow on Macs compared to Cinebench 2000. My Dual 1.25-GHz G4 matched a 2.66-GHz Pentium 4A in Cinebench 2000. Now, my Dual 1.25-GHz G4 just barely beats a 2.2-GHz Pentium 4 in Cinebench 2003. As such, the Athlon outperforming the G5 was expected. (By the way, a single 1-GHz G4 would score 96... a Dual 1-GHz G4 scores 176.)

Unreal Tournament 2003 relies more on the graphics card than the processor speed, so the Athlon with a Radeon 9700 Pro (the fastest graphics card up until only fairly recently) should beat the G5 with the GeForce FX 5200 (a pitifully bad graphics card that is just barely a match for the GeForce 4 MX).

Adobe is going to release a G5-optimized plug-in for Photoshop very soon now, and Maxon engineers claim to be working on an optimized version of Cinema 4D XL for the G5. As such, don't be disheartened by these initial results.

---------------------------------------

I have to disagree with this, blackwind. UT2K3 is mostly CPU bound, hence one of the reasons why there is such a big difference in FPS between "flyby's" and "bot-matches" on both Mac and PC. It also shows up when comparing higher end systems like the dual 1.42GHz G4 (check out xlr8yourmac.com), there is virtually no FPS difference on this system with a GeForce 4MX/Ti, Radeon 9000/9700 (EXCEPT FOR FLYBY'S OF COURSE), so the limiting factor is indeed the CPU.

On another note, the GeForce FX 5200, while it is not as fact as the GeForce FX5900 Ultra, has similar core specs to the GeForce 4 Ti, and in benchmarks which stress the graphics card, shows similar results.

[Peder Rice]

Originally posted by Dan-the-Man:
---------------------------------------

I have to disagree with this, blackwind. UT2K3 is mostly CPU bound, hence one of the reasons why there is such a big difference in FPS between "flyby's" and "bot-matches" on both Mac and PC. It also shows up when comparing higher end systems like the dual 1.42GHz G4 (check out xlr8yourmac.com), there is virtually no FPS difference on this system with a GeForce 4MX/Ti, Radeon 9000/9700 (EXCEPT FOR FLYBY'S OF COURSE), so the limiting factor is indeed the CPU.

On another note, the GeForce FX 5200, while it is not as fact as the GeForce FX5900 Ultra, has similar core specs to the GeForce 4 Ti, and in benchmarks which stress the graphics card, shows similar results.

http://www6.tomshardware.com/graphic...x_5900-14.html
http://www6.tomshardware.com/graphic...ournament_2003

CPU bound?

[Dan-the-Man]

Originally posted by Peder Rice:
http://www6.tomshardware.com/graphic...x_5900-14.html
http://www6.tomshardware.com/graphic...ournament_2003

CPU bound?

Perhaps I didn't make myself clear. If you look at the benchmarks from this URL, they are all run under the "flyby" test for UT2K3 - this doesn't stress the CPU, as the graphics card is doing all the work. If you look at the "botmatch" tests (which are more relevant to how the game actually plays), you will see (especially on the Mac, where the CPU is the bottleneck) that there is little or no difference in average frame rates between a GeForce 4MX/ti or radeon 9000/9700 on eg; a Dual G4 1.42GHz. This is my point; on the Mac the limiting factor is the CPU (ie the game is CPU bound in actual game play - NOT FLY BY TESTS). The G5 (dual 2GHz), may change this of course, we''l have to wait and see.

[blackwind]

Hmm, so "botmatch" tests are actually CPU bound? Eh, the saying that "You learn something new everyday" strikes again for me!

In any case, now that I contemplate the name, it seems to test the speed of artificial intelligence execution (whatever that means) based on the number of bots, which would indeed be mostly CPU bound (well, CPU and RAM bound).

Am I correct this time, or am I mistaken again?

Also, is "botmatch" CPU limited on Athlons and Pentium 4's as well? Or do they all get about 34 fps?

In any case, thank you in advance for enlightening me.