“Rick Bernstein, sent in a link to a great discussion at Luxology’s forums about render tests done with a new Quad. Apparently the company had one for testing purposes. No surprises there, Luxology folks are big Apple supporters. However, the test render results for the Quad were definitely surprising and pleasing to many,” Architosh reports.
“Alan Hastings of Luxology posted a test rendering image taken from the new Quad G5 with modo 201. The scene included 244,000 polygons with 8 sample antialiasing and 200 indirect rays. The new Quad G5 finished the image — which looks extremely good by the way — in 17 seconds flat. That bested a score of 38 seconds flat on a Luxology Power Mac Dual G5 at 2.5 Ghz,” Architosh reports. “Alan also tested the same render on a Dell 530 workstation with dual 2.8Ghz Xeons. The result? 49 seconds. There are no Opteron results to compare to yet so at the moment we are not sure if the new Apple Quad G5 is the new heavy weight champion of the world when it comes to rendering.”
Full article here.
Luxology article and image here.
[UPDATE: 2:15pm ET: Revised headline.]
Related MacDailyNews articles:
AnandTech: Apple new Power Mac G5’s biggest improvement is the move to PCI Express – October 21, 2005
Photos of new dual core Apple Power Mac G5 interior, ports, and more – October 19, 2005
First benchmark tests of Apple’s new Power Mac G5 dual-core machines – October 19, 2005
Apple introduces Power Mac G5 Quad and Power Mac G5 Dual – October 19, 2005
Are you kidding me? Have you even looked at those numbers?
In every test these 3.6Ghz machines (both dual/single core Xeon and dual core Pentium EE) just barely beat out the old 2.7 Ghz dual G5, even though they had faster memory and better video cards. You call this a bitter pill? With a Ghz advantage and a better chipset, I have to wonder why the Dell rigs didn’t blow the doors off the PM … except I don’t have to wonder. I already realize that the weak link in their machine was basing it on Intel technology.
You really need to lay off the Apple tabacco LOL. It’s amazing what it does to your perception. What’s the matter, can’t admit a single Intel chip outclassed a dual 2.7GHz Powermac? (which you call “old”, even though it was released only a month prior LOL). And for much less cost no less.
…At $3496, it’s more than $1000 cheaper than the dual Xeon 3.6 Dell Precision workstation 470 we tested earlier, and it’s also more than $1000 cheaper than the Mac dual G5 2.7GHz machine. It’s remarkable to test a new product with faster performance at a price that’s about 25% cheaper. Compared to the dual Xeon machine, the performance was roughly the same on many benchmarks, but significantly better on others. And compared to the Mac, particularly in the CineBench benchmarks and the TotalBenchmark Comp 2 scores, the Mac fell way behind while still costing over $1000 more. …
“You think the change to Intel (x86) dual cores is all good. Take a look at the following …
Dual/single core G5 PM beats dual core PentiumD 3 out of 5 tests, and is very competive in the other 2:
http://www.barefeats.com/dualcore.html
“
So you think the change is NOT good due to a benchmark based on a low end dual core Intel chip? How stupid is that? Rob Art himself stated the dual core chip he used wasn’t the fastest available… a $250 Pentium D 2.8 Ghz 820. Run those tests again with a $995 Pentium D 840 (like digital video editing link above) and then you’ll see why change IS sometimes better.
Pretty much the same result of last gen G5 PM vs Opteron, among others:
http://www.barefeats.com/macvpc.html
You used this link to support your “change is not good” argument. Problem is, all it does is showcase how its trounced by x86 processors.
“Good luck explaining away how these new G5s ‘got lucky’ in trouncing this, the best Intel has to offer.
Actually I look forward in your responses when upcoming tests pit the new quads against the best of AMD/Intel quads. I can tell people saw only the 49 second mark compared to 17 seconds, and totally overlooked the system in question, an old 2.8Ghz dual Xeon. CAD applications already favor Opterons/Xeons in benchmarks over Powermacs, what makes you think a dual core dual 3.2Ghz Xeon box won’t best this score at all?
You say “If Apple had stuck by IBM, what makes you think the next generation of G6 would have embarrased Intel? Judging by recent history, it would have been more Apple marketing hype …”
Yeah – revisionist history.
LOL
Revisionist history and contemporary rubbish–according to you and Lisa. Both of you claimed the G5 was the fastest computer on the planet, when 3rd party benchmarks showed otherwise. Who’s really laughing now? LOL.
In multiple comparisons the Dual 2.7 (and Dual 2.5) provided dramatically inferior performance on industry standard database functions (serving the database market is much bigger than Hollywood in gross dollars). Apple has tried for years to set the pace and have just not been able to match Intel based machines (except when the “Reality Distortion Generator” is running full blast). I applaud any move Apple makes that will increase the performance of their little Porsche and BMW “like” computers (bitch’n handling, good accelleration, premium priced).
Actually, in test after test after test, relatively GENERIC PC hardware running other OS’s has repeatedly trumped Apples best hardware offerings. I think it will be absolutely GREAT when my Apple purchases have an engine in them that delivers Ferrari street-performance. Mac user since late ’84, PC free since December 4, 2003.
Kerry and Sammy: What can I say? You two are quite a comedy team.
The numbers are the numbers guys. Anyone with an ounce of sense who looks at them, as opposed to your spin-like scribblings, will see them for what they are. And the new G5 PMs will simply drive the point home further. Good luck to both of you in your new careers, though – I’m sure you’ll knock em dead at The Improv.
Macs King: Color me Curious regarding the performance issues you cite. The topic as you state it seems a little obscure – at least for me. I do know that OSX has problems keeping up with Linux and Windows dealing with multiple requests in a commercial server environment (think of having a bunch of customers trying to buy something, and checking out at the same time). AnandTech did a very extensive test on this subject earlier this year. But this is almost totally a function of the limitations/advantages of the OSes in question, not because of the CPU (although the G5’s integer performance has been known to be it’s only ‘average’ aspect, and that would deal with databases). Specifically, OSX isn’t very good at allowing such commands get through to its kernal all at once (which is apparently VERY important).
Quote from Anandtech:
“In the early versions of Mac OS X, only one thread could lock onto the kernel at once. This doesn’t mean only one thread can run, but that only one thread could access the kernel at a given time. So, a rendering calculation (no kernel interaction) together with a network access (kernel access) could run well. But many threads demanding access to the memory or network subsystem would result in one thread getting access, and all others waiting.
This “kernel locked bottleneck” situation has improved in Tiger, but kernel locking is still very coarse…
[So] Mac OS X is incredibly slow, between 2 and 5(!) times slower, in creating new threads, as it doesn’t use kernel threads, and has to go through extra layers (wrappers). No need to continue our search: the G5 might not be the fastest integer CPU on earth – (but) its database performance is completely crippled by an asthmatic operating system that needs up to 5 times more time to handle and create threads.”
Check it here:
http://www.anandtech.com/mac/showdoc.aspx?i=2436&p=7
So, do the “industry standard database functions” you mention have anything to do with that? It sure sounds like it. If not, can you provide a link to another test or two, documenting what you claim?
Your last comment is more of what ‘Martin & Lewis’ were spewing above – good for a laugh – but if you can keep the Wintel marketing to a minimum, I definitely would like to see whatever facts you’ve got regarding the database issue, and how the G5 manifestly contributes to it.
Thanks!
“
The numbers are the numbers guys. Anyone with an ounce of sense who looks at them, as opposed to your spin-like scribblings, will see them for what they are. And the new G5 PMs will simply drive the point home further.”
Yes, numbers are numbers. And you’re EXACTLY right, any one with an ounce of sense will see them for what they are.
For example, barefeat’s comparison of the high end dual 2.5Ghz Powermac to Intel’s low end Pentium 820D. Seems Anandtech users had a field day on that one, actually forcing Rob to post specific notes concerning the “comparison.”
http://forums.anandtech.com/messageview.aspx?catid=27&threadid=1669382&enterthread=y
And yes, earlier dual-core chips of the year really did shed Apple in an embarrassing light. Your obvious snubbing of this fact, and overzealous rants concerning IBM, Cell, DRM & Intel show exactly the kind of chip you have on your shoulder that’s skewing your judgement. Everyone here sees it except you.
For most of this year, Apple was limp and only now does it finally have something that can actually compete against x86. Unfortunately, not enough to “embarass” Intel nor AMD. You can pout and rant all you wish, but IBM doesn’t have much of a future for Apple even if they had stayed. You may think Apple might have been better going with AMD, but you would be basing that on currently shipping products….Intel’s chips of 2006 are poised to reshape the industry, and that’s not my own personal opinion…that’s the expectation from many technical venues (Anandtech included) who fully understand Intel’s roadmap.
Perhaps you’ll change your tune come around this time next year.
Correct link:
http://forums.anandtech.com/messageview.aspx?catid=27&threadid=1669382&enterthread=y
Sammy: Your link – like your arguments – keeps coming up empty. It seems odd that Anandtech posters would be flaming Barefeets testing, but there’s no way to verify it anyway since a blank page keeps coming up. I hope this isn’t just more spin on your part. I actually stopped reading your above comments when it got to be too much.
Now that I’ve re-read it all I can see the veneer of legitimate critisicm, but for every valid point you make, there’s more that are out of context or false. Example: You say “You used this link to support your “change is not good” argument. Problem is, all it does is showcase how its trounced by x86 processors. http://www.barefeats.com/macvpc.html “
Lets recap what’s on the link.
Test: AfterEffects 6.5 – Nightflight Render
Result: dual 2.5Ghz G5 PM – 1st place, dual 2.6Ghz Opteron – 2nd place, dual 2.0Ghz G5 PM – 3rd place, and a variety of x86 gear (Athlon FX55, Pentium 4, dual Xeon), all at higher clockspeeds, bringing up the rear.
Test: Cinebench 2003 – CPU Render
Result: 2.6 Ghz Opteron – 1st, 2.5Ghz G5 PM – 2nd, 3.4Ghz Xeon – 3rd, 2.0Ghz G5 PM – 4th, etc …
Test: Bryce 5 – Beach Chair Render
Result: 2.5Ghz G5 PM – 1st, single core 2.6Ghz Athlon FX55 – 2nd, 2.6 Ghz Opteron – 3rd, … the 2.0Ghz G5 beat out the Xenon (well over a Gig faster) handily.
I won’t rewrite all the results, but here’s what the tester had to say:
“I’m often accused of choosing applications that favor the Mac. My only criteria is that the applications run on both platforms. That charge falls flat when you see the fastest Power Mac G5 was fastest in only 2 out of 5 tests on this page… Though the Dual G5 Power Macs were able to run with the Windows “big dogs” in the CPU intensive tests featured on this page, the 3D graphics tests were a whole other matter.”
These were the best rigs offered as of March when the tests were done. G5 was fastest in 2 tests that relied on CPU performance, competitive in two others that revolve around Photoshop optimizations (which are in turn dependent on Adobe’s coders). The dual 2.5Ghz G5 lost a head-to-head with the dual 3.4Ghz Xeon in the the Maya 6.5 Render, which tests both overall system hardware and system software capabilities, and the performance of both of those features are completely in the hands of Apple – not IBM.
I can’t go through each of your critiques – I’d be here forever – but this one is representative of all of them; a claim made (a complete blow out for x86) despite facts showing quite the contrary (G5 either winning outright, or holding it’s own against CPUs with advantages in clockseed/cache size/ what have you). When you do make a valid claim on a particular ‘loss’, you ignore the context of what is primarily being tested (i.e. video system, software, overall system hardware, or CPU).
With little more than a passing knowledge of these things, you can clearly see what the numbers are saying – the 970FX/G5 maximizes what it has to a much greater degree than any other CPU. Even with ‘bloated muscles’ (uber Ghz, twice the cache, and now even dual cores) Intel’s stuff can rarely eak out a victory. AMD is a different story – they are clearly more capable than Intel – but even when they win over G5 it usually isn’t by much. And, of course, Apple isn’t using AMD, which just further proves my “overzealous rants” about how the adoption of Intel on Mac has nothing to do with performance.
You say “For most of this year, Apple was limp and only now does it … have something that can actually compete against x86.”
Half right again. The fact, undeniable, that Apple has been “limp” most of this year is not because better PPC CPUs haven’t been available (see multiple links above). It’s because Apple is cutting the ties, and slowing (IBM) or eliminating (Freescale) further PPC technology adoption. It’s time for the slow march – which will just make the upcoming x86 offerings look that much better. I mean, why else would Apple take so long to introduce a dual core G5? Why not introduce the dual core version of ‘lowly’ G4 in the Powerbook, if the Pentium M versions that are coming – running almost a full Gig faster – are so damn superior? Costs? Development time? Neither makes sense given the availability of these CPUs for many months, Apple’s long expertise in making MoBos for them, or that any investment could easily be amortized by moving these ‘inferior’ products down to the non-pro lines once the Macintel PBs are introduced.
I’ll just end here by saying I don’t mind if you disagree with me – just don’t insult me (or anyone who – god forbid – may be interested in reading) by using the ‘political spin’ school of debate in doing so. Know your subject – all facets of it – and if the chips don’t fall your way, don’t make up a bunch of crap to disguise that fact.
I’ll read your any response, but we’ll have to pick this up on another thread.
Later
I find Sammy more convincing than Odyssey67.
Sammy: Your link – like your arguments – keeps coming up empty. It seems odd that Anandtech posters would be flaming Barefeets testing, but there’s no way to verify it anyway since a blank page keeps coming up.
Seems odd that people with rational would actually object to a low end Pentium dual core pitted against a high end Powermac? Well it would seem odd to you, after all…you tried to convince Macs King that change to the Intel processors would not be good because of that very benchmark…a high end dual powermac pitted against an entry level dual core Pentium D. Do you not see the flaw in that logic? Apparently, plenty of people did…. I can’t post threads directly for some reason, just go here: http://forums.anandtech.com/searchresults.aspx
and type “powermac” in the search dialogue, look for “Dual Core Pentium versus Dual CPU G5 Powermac” headline.
Go ahead and check priceline also, and see how much the Pentium D 820 retails for….right down at the bottom of the barrel, since it’s an entry level dual core chip. That’s like comparing an Athlon64 4000 against a bottom end iMac.
Makes sense to you now?
“G5 was fastest in 2 tests that relied on CPU performance, competitive in two others that revolve around Photoshop optimizations (which are in turn dependent on Adobe’s coders). The dual 2.5Ghz G5 lost a head-to-head with the dual 3.4Ghz Xeon in the the Maya 6.5 Render, which tests both overall system hardware and system software capabilities, and the performance of both of those features are completely in the hands of Apple – not IBM.
“
“With little more than a passing knowledge of these things, you can clearly see what the numbers are saying – the 970FX/G5 maximizes what it has to a much greater degree than any other CPU.”
Funny how you spin around yourself to downplay the G5’s losses as not really CPU related and evangelize the wins as only CPU pertinent cases. Spin spin spin…. LOL
IF you statement had been correct, the opposite would have been shown in the benchmarks… In fact, Night Flight is actually an example of selective benchmarking to shed the G5 in its best light under tight rendering paths. The fact that the results do not carry over to other CPU dependant renders proves this point exactly.
…Before we get started, I’d also like to remind some readers who in the past have had objections to using After Effects on the Mac as a benchmarking tool that Apple itself, on its own Web site, declared After Effects optimized for both OS X and Windows XP when it cited benchmarks using a comp called Nightflight, a narrowly-hewn rendering routine that tests just one specific aspect of the system, one that—no big surprise here—favors the Mac’s rendering characteristics.
http://www.digitalvideoediting.com/articles/viewarticle.jsp?id=32951
In fact, I just love how you conveniently ignore the only benchmark in barefeats suite that truly represent CPU power, Cinebench. The Powermac 2.7Ghz is virtually identical to the dual Xeon 3.4 (and lower compared to a Xeon 3.6). These results carry over to the wider range of 6 After effects benchmarks, where a dual Xeon 3.6 wins 3 out of 3 benchmarks against the 2.7Ghz, virtually a tie as suggested by Cinebench (and just as Cinebench indicated, the Powermacs gets slaughtered by Dual Core Pentium EE’s and Opterons).
“Half right again. The fact, undeniable, that Apple has been “limp” most of this year is not because better PPC CPUs haven’t been available (see multiple links above). It’s because Apple is cutting the ties, and slowing (IBM) or eliminating (Freescale) further PPC technology adoption. It’s time for the slow march – which will just make the upcoming x86 offerings look that much better. I mean, why else would Apple take so long to introduce a dual core G5?
Dual core versions of the POwerPC 970FX was not announced by IBM unti mid summer of this year. Yes you can say IBM had engineering samples from much prior, just as AMD likely has engineering samples of quad core CPU’s now….high reliable yields are all that matters, IBM needed this much time to ramp up and supply enough dual core chips that Apple can now introduce thier new line.
Apple had no control over this. AMD/Intel were already a full cycle into dual core CPU’s. Apple, together with IBM, was losing the technology race. Dual core G4’s would have been a step backwards to “move forward”, like AMD using dual core K7’s instead of Athlon 64’s to further thier business.