BareFeats Has A DP G5 Cinebench Benchmark

[Makosuke]

The title says it all; looks like the guy over at BareFeats got his hands on a DP G5 to test and put it in a chart with the number from the 1.6 benchmarks posted yesterday:

http://www.barefeats.com/g5sum02.html

Considering Cinebench is completely unoptimized for the G5, those are impressive numbers. Heck, even if it were optimized, they're still pretty good; we're looking at a near doubling of speed from the previous top of the line, and a performance-per-clock about the same as the Intel Xeon (and although you can now get a Xeon clocked quite a bit higher than a G5, that's still a heartening sign).

[idea_hamster]

If there isn't a typo, someone's going to have to explain to me why the 2GHz G5 performed at 40% of the 1.6GHz G5.

G5 2GHz: 66
G5 1.6GHz: 158

:confused:

[Makosuke]

Originally posted by idea_hamster
If there isn't a typo, someone's going to have to explain to me why the 2GHz G5 performed at 40% of the 1.6GHz G5.

G5 2GHz: 66
G5 1.6GHz: 158

Read the caption on the graph: "SHORTER bar means FASTER" -- those are times in seconds, not abstract benchmark numbers.

[Mr. Anderson]

So that's not that bad a score and they also mentioned that it was not optimized for 64 bit (which most software also isn't optimized for yet).

So eventually, we'll get an even better score.

Its really nice to see Apple competing head to head with the PCs again ;)

D

[cr2sh]

Love of god.. I know that the g5's are new, but I wish they'd optimize the benchmarking software so we see some results that:

1. Are faster than Dual Xeons
2. Actually mean something

As it is now, this benchmark is pointless to even look at... let's see the unoptimized score for the dual xeon.. then I might be interested. :)

[Abstract]

Agreed. Its like comparing a Bimmer to an Infinity G35, but only giving the G35 three wheels instead of four. We don't even know how the G5 will do when everything is optimized for it, which is all that really matters. These numbers don't represent the performance of the G5 PM's in full working conditions. I want to see a Xeon put against an optimized G5.

[Makosuke]

Think of it this way; not all software is optimized, or will be in the near future, so even unoptomized benchmarks tell you something. Heck, I'm still running Photoshop in OS9 because I'm too cheap to upgrade, so I'm sure there'll be unoptomized software I'm using a year or two from now.

But on the optomization front, AccelerateYourMac has an interesting comment on their front page (http://www.xlr8yourmac.com/) referencing a PostForum thread (http://www.postforum.com/forums/read.php?f=6&i=87458&t=87424) that has some Cinebench benchmarks of a single 1.8Ghz. But what is really interesting is there's a quote from a Maxon guy in there saying that they've been experimenting with G5 optimization, and that when (or if) they ever ship a tweaked version, a DP 2GHZ G5 should score pretty much the same as a DP 2.4Ghz Xeon.

Cousin of a friend of a friend of a guy I work with story, I know, but it sounds reasonable, particularly considering how poorly Cinebench seems to perform on PPC chips in general.

[agreenster]

Cinema 4D is a raytrace renderer, where the top half of the image is rendered by one processor and the bottom half is rendered by the other processor. Then, whichever processor completes its half first, it picks up the slack for the rest---I used to render w/C4D all the time, and it was kinda slick to watch.

Anyway, its interesting to note that the Single processor G5 is just a touch slower than the DP 1gHz G4. That leads me to believe that the SP 1.6gHz G5 is actually rendering twice as fast as a 1Ghz SP G4 because it is doing a whole frame (image) with one processor, instead of two.

Xeons are still mighty fast chips.....I use em and love em.

Cant wait until the G5 grows to its full potential. It may actually catch up.

[MrMacMan]

Cool.

But the 1.6 is slower then a Dual 1 GHZ!

noooooo.

:(

[agreenster]

Originally posted by MrMacman
Cool.

But the 1.6 is slower then a Dual 1 GHZ!

noooooo.

:(

Not necessarily. The chip is faster (nearly twice as fast, read my above post), but two processors are better than one when it comes to rendering. The G5 can do with one processor what the G4 can do with two......

Why Apple didnt release all DP systems is still beyond me. :confused:

[idea_hamster]

Originally posted by Makosuke
Read the caption on the graph: "SHORTER bar means FASTER" -- those are times in seconds, not abstract benchmark numbers.
Oh, right :rolleyes: Sometimes I shock even myself!

It seems that this makes some fair sense given that the program doesn't make use of 64-bitted-ness:

The 2xXeon is clocked at about 20% faster than the 2xG5 and benchmarks at about 20% faster; and

the 3Ghz Pentium is clocked about 53% faster than the 1.6GHzG5 and benchmarks at about 55% faster.

[eric_n_dfw]

Sweet!

You'd think, however, that they would have had the ability to get a G5 compile out the door by now. Hell, you'd think Apple's FCP and other apps would have the same. (or maybe they do and we just don't know it!)

[mactastic]

Freakin awesome... Can't wait to see some 64 bit optimized benchmarks.

[pwm519]

Waiting for people to start complaining?


I think these are excellent results!

[Fender2112]

The part I find interesting is that the 1.6 G5 does about as well as a Dual 1.0 G4. I'm not sure if that's good or bad. But it is interesting. I'm anxious to see how the 1.8 G5 does as this is the one I will be getting.

If some of you folks don't mind, how about posting some comments about noise levels, heat, size, and any other details you think might be useful.

[pwm519]

I think to really be competative Apple shoud have made dual 1.6, 1.8 and 2.0 at the current price or maybe dual 1.8 and 2.0 with a low end single 2.0. I think that would have put the low end model just slightly above the 1.42 DP machines in performance and would have been a really good deal across the entire line.....that said the 2.0 GHz seems to still be an excellent deal.

[Neuro]

Maybe the performance of the 1.6 and 1.8 is actually disappointing in a DP configuration...

Or perhaps Apple feels the need to make G4 DP 1.42 buyers feel like they bought someting worth having...

From my perspective, the only Mac worth owning at the moment is a DP 2Ghtz G5. (I have a slooooow G4 667 TiBook).

I'd love to upgrade from my PB, but only if the PowerMacs get faster, or the PowerBooks move to a G5 with faster disk.

[cc bcc]

I read on a 3D forum (postforum) that maxon is already optimizing Cinema 4d (and therefor Cinebench) for the G5 just this week, they didn't get a developers machine! Anyway, they expect the dual 2 GHz to come close to the dual xeon (much more expensive). A Maxon engineer posted this info.
The xeon will be a bit faster because of hyperthreading, especially in radiosity. They make no use of Altivec, MMX, SSE2 etc. because of the unpedictable nature of raytrace calculations.

edit: link (http://www.postforum.com/forums/read.php?f=6&i=87424&t=87424)

[idea_hamster]

Show me the Dual 3.0 GHz G5 and the G5-optimized benchmarking software!

Ooops. Drool alert!

:cool:

[NavyIntel007]

BIG CAVEAT: Cinebench 2003 is in no way optimized for 64 bit processors or the G5... yet. Word on the street is that Maxon is looking seriously into G5 optimizations using the new Xcode compiler. Predictions are that the G5 will easily surpass the Dual Xeon once the G5 optimizations are in place.

I think this might bring the dual G5 ahead of the Xeon and give a speed adjustment to the 1.6.

I agree, I was surprised to see Apple cheap out and not offer all duals. They should of left it as an option, all those motherboards are the same. All they would have had to do was do the pin out for the second processor. They could have scaled the prices so that a dual of one machine was actually more expensive than it's upgrade.

When it comes out, the speed differences between the singles and the duals is going to be rediculous. I mean, if I was in the market... I consider the Powermac to be an investment... The usability of the Dual G5 is, dare I say, twice what the singles are but not twice the price.

[windwaves]

pretty cool, I guess my Dual 1.25 still has good life left in it as it is only beaten by the current Dual G5. Needless to add that the performance difference between the D 1.25 and the D 1.42 is pathetic. Great, this allows me to patiently wait for the dual G5 at 3 Mhz :)

Lets go Apple, after the G4 debacle got to move quickly !!!

[fred]

Even if optimized this would still put into doubt Apple's claims that Dual G5 2 ghz smokes a dual Xeon 2.8 ghz..... Why doesn't this surprise me ?

[crazzyeddie]

I really wish people would figure this out...

When people say the program has not been optimized for the G5 that DOES NOT mean 64-bit!!!!!!

The G5 has different Altivec limitations and some of the commands slow it down greatly. Therefore, optimizing for G5 means changing the Altivec commands.