• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

Anandtech on X360/PS3's "poor" CPU performance

gofreak

GAF's Bob Woodward
Anandtech's been talking with some developers off the record re. X360 and PS3.

http://www.anandtech.com/video/showdoc.aspx?i=2461

Right now, from what we’ve heard, the real-world performance of the Xenon CPU is about twice that of the 733MHz processor in the first Xbox. Considering that this CPU is supposed to power the Xbox 360 for the next 4 - 5 years, it’s nothing short of disappointing. To put it in perspective, floating point multiplies are apparently 1/3 as fast on Xenon as on a Pentium 4.

The Cell processor doesn’t get off the hook just because it only uses a single one of these horribly slow cores; the SPE array ends up being fairly useless in the majority of situations, making it little more than a waste of die space.

We mentioned before that collision detection is able to be accelerated on the SPEs of Cell, despite being fairly branch heavy. The lack of a branch predictor in the SPEs apparently isn’t that big of a deal, since most collision detection branches are basically random and can’t be predicted even with the best branch predictor. So not having a branch predictor doesn’t hurt, what does hurt however is the very small amount of local memory available to each SPE. In order to access main memory, the SPE places a DMA request on the bus (or the PPE can initiate the DMA request) and waits for it to be fulfilled. From those that have had experience with the PS3 development kits, this access takes far too long to be used in many real world scenarios. It is the small amount of local memory that each SPE has access to that limits the SPEs from being able to work on more than a handful of tasks. While physics acceleration is an important one, there are many more tasks that can’t be accelerated by the SPEs because of the memory limitation.

I think he's being a bit sensationalist, and a bit too keen to compare unfavourably to PC architectures. I think it's a given that neither of these systems can have their performance exploited as easily as on a PC. It just seems, on his part at least (not the dev's), to be a big moan re. how they're not PC chips.

I also think the "waste of die space" comment re. Cell SPEs is extraordinary given that he goes on to discuss how they can "accelerate" physics. Even if that's all they could do, it'd be a big win IMO. Saying it only helps with handful of tasks isn't saying much without considering the sizes of those tasks (Physics is a big one). In his very own previous article, he mentioned Tim Sweeney saying that many of the things SPEs aren't useful for only take a small proportion of CPU time anyway..

Worth reading all the way through though. Also comments about the GPUs in there, but that still seems to be speculative territory.
 
Kind of vague. How is the performance being measured? I dont think it's fair to compare the CPUs by running apps of today which definitley dont take advantage of the multi-core architecture.

The recent articles from them all come off as guesswork.
 
oh no! IBM are idiots. Best CPU for consoles is a pentium 4.

How close can I come to calling them fuckwits before I get slapped? Just checking.
 
So basically he seems to be saying noone will be using heavily multithreaded engines for a few years, so until then the CPUs won't be used in the way they were designed and thus won't perform as well as hoped?

That seems kind of shortsighted.
 
The GPUs of the next-generation platforms also proved to be quite interesting. In Part I we speculated as to the true nature of NVIDIA’s RSX in the PS3, concluding that it’s quite likely little more than a higher clocked G70 GPU. We will expand on that discussion a bit more in this article. We also looked at Xenos, the Xbox 360’s GPU and characterized it as equivalent to a very flexible 24-pipe R420. Despite the inclusion of the 10MB of embedded DRAM, Xenos and RSX ended up being quite similar in our expectations for performance; and that pretty much summarized all of our findings - the two consoles, although implementing very different architectures, ended up being so very similar.

as many have been predicting for years :)
 
looking at the PGR3 pics... it's obvious it's not a big a deal as he's making it out to be. or maybe he just doesn't know how to leverage the assets of these new systems?

p.s. Anand is very unbiased... don't worry about any kind of slant. he won't have any.
 
shpankey said:
looking at the PGR3 pics... it's obvious it's not a big a deal as he's making it out to be. or maybe he just doesn't know how to leverage the assets of these new systems?.

He's not discussing his own experience, but those of developers he's talked to.

Barnimal said:
now were they basing those 360 comments on the 30% complete hardware? :D

They're talking directly about the tri-core chip there. Beta kits are out.

midnightguy said:
as many have been predicting for years :)

..and this is now truth..?

I don't think Anand either understands or wants to acknowledge the importance of some of the differences he himself points out. For example, he starts talking down Cell on the one hand, but on the other it seems to me he hands it a sizeable advantage, whether he knows it or not.
 
Well I finished reading this. Long story short summerizing this article for those who don't want to read it:

CPUS suck, GPUs rock, and it'll take developers time to figure out the multi thread archetecture of each processer. Amazing insight. But after reading the article I'm more confused about the potential pitfall of these CPUs. The article basically states the CPUs suck because they aren't great a single threaded support. But isn't the point of these cpus to be multi threaded? Well whatever. :P
 
Right now, from what we’ve heard, the real-world performance of the Xenon CPU is about twice that of the 733MHz processor in the first Xbox. Considering that this CPU is supposed to power the Xbox 360 for the next 4 - 5 years, it’s nothing short of disappointing. To put it in perspective, floating point multiplies are apparently 1/3 as fast on Xenon as on a Pentium 4.

holy shit, *if* true, that sucks really bad.

Intel 733 MHZ CPU in original Xbox - floating point performance: 3 Gflops is the figure usually mentioned. I've seen as low as 1.5 Gflops (recent Xbox vs Xbox 360 comparison chart) - and I've seen 3.5 Gflops as the highest listed rating.


lets take the highest possible FLOPs figure for original Xbox Intel 733 MHz CPU (3.5 Gflops) so that we can be kind & generous to Xbox 360. If Anandtech is even almost correct, that means Xbox 360's triple core Xenon CPU produces less than 10 Gflops in realworld performance ?
 
Vince said:
Power Efficient Processor Design and the Cell Processor - Peter Hofstee, Ph. D.

That's why you dumbass, although it might be too logical for his sensationalist banter about how Sony is deadset in the [wrong] direction and Microsoft is nothing but a personified mindless and gready bastard.

To think this guy is suppose to be a journalist...

the PS3 and XCPU processors are more simplified processors. He doesn't seem to realise that single-threaded code on PS3 and XCPU will only be tapping well... the majority of PS3s power (not even using a single SPE) and using a single core of Xbox 360 without hyperthreading.

He's under the assumption that everyone will be using single threaded code. :/
 
OniShiro said:
How can 3 x 3.2Ghz CPU be only twice as fast as a 733 Mhz? I find it hard to believe.

Ohhh I see that the GHz demon has got you too!

I think he was talking about a single core, and that would be easy to believe. The G5 architecture isn't that impressive in the first place. The Xbox360 CPU has scaled down an already lackluster product.
 
The most ironic bit of it all is that according to developers, if either manufacturer had decided to use an Athlon 64 or a Pentium D in their next-gen console, they would be significantly ahead of the competition in terms of CPU performance.

While the developers we've spoken to agree that heavily multithreaded game engines are the future, that future won't really take form for another 3 - 5 years. Even Microsoft admitted to us that all developers are focusing on having, at most, one or two threads of execution for the game engine itself - not the four or six threads that the Xbox 360 was designed for.


Even when games become more aggressive with their multithreading, targeting 2 - 4 threads, most of the work will still be done in a single thread. It won't be until the next step in multithreaded architectures where that single thread gets broken down even further, and by that time we'll be talking about Xbox 720 and PlayStation 4.

hmmm.
 
gohepcat said:
The G5 architecture isn't that impressive in the first place. The Xbox360 CPU has scaled down an already lackluster product.

The PPE in Cell has nothing to do with the G5, and if the X360 cores are very similar as seems to be the case, I'm guessing they haven't either. Unless you're making a comment generally about the Power architecture..
 
How did the EE and XBox CPU compare with PC CPUs at the time?

I'm guessing not very favourably and yet PS2 is outputting games like GTA:SA and Gran Turismo 4 5 years later and XBox's 'Celeron' 733 is running Doom III, Half Life 2, and Conker.

HALF-LIFE 2 MINIMUM SPECS

Windows XP/2000/Me/98
1.2Ghz processor
256 MB Ram
Net connection
DVD drive

These console vs PC comparisons are getting more and more worthless.
 
Well, maybe this will unite sony and ms fanboys in a show of love.

Maybe we'll get next gen gameplay in 5 years. :D
 
gohepcat,the PPE's in 360 have nothing to do with G5, it is however fairly similar to the PPE in PS3. As for the G5 its not a poor CPU its actually quite the opposite. What the article basically say if Xbox 360 is slow on a single thread, the ps3 will probably be just as bad. ERP over at b3d whose woking on the ps3 stated that single threads on both ps3 and 360 wont touch a P4, but the numbers are not as bad as this article reports.
 
Pug said:
gohepcat,the PPE's in 360 have nothing to do with G5

J Allard disagrees:

G: It's announced that the CPU is a symmetric 3-core CPU. Is this core an equivalent to PowerPC970/G5 or simplified one?

A: Simpler and advanced. Basically we adopted the same CPU core as PowerPC G5. It's based on PowerPC G5, but we removed unimportant features from it. For example instead of having L2 cache for each core we adopted L2 cache shared by 3 cores.

http://www.beyond3d.com/forum/viewtopic.php?t=23595
 
The only thing that comes close to being worse than PC programmers who refuse to learn to program on anything but the x86 architecture through the Windows API is a tech-site writer who refuses to understand anything beyond the x86 architecture.
 
dorio said:
Well, maybe this will unite sony and ms fanboys in a show of love.

can we unite and drive off the nintendo kids once and for all?
 
gofreak said:
He's not discussing his own experience, but those of developers he's talked to.

I like Anand's site, but anonymous developers? What kind of developers? I mean they could be writing code for McDonald's automated french-fryers for all we know.
 
HokieJoe said:
I like Anand's site, but anonymous developers? What kind of developers? I mean they could be writing code for McDonald's automated french-fryers for all we know.

I disagree with the tone of the article and his spin on some of the points as much as the next, but I think it's safe to trust he was talking to game developers on both systems and at least as far as the actual points on performance (the specific things like 2x the performance of that, or 1/3 of the speed of that) actually came from developers. I wouldn't be in a hurry to write it off just because the comments are anonymous - obviously NDAs are still in effect. I'd have preferred word-for-word quotes from said developers, however.
 
monkeymagic said:
How did the EE and XBox CPU compare with PC CPUs at the time?

I'm guessing not very favourably and yet PS2 is outputting games like GTA:SA and Gran Turismo 4 5 years later and XBox's 'Celeron' 733 is running Doom III, Half Life 2, and Conker.

.


yeah, Xbox is pretty impressive for a celeron 733, 64(!) MB ram and geforce 3. Almost pulling off HL2 and Doom II.

now consider what that means for the next gen consoles....drool.
 
gofreak said:
I disagree with the tone of the article and his spin on some of the points as much as the next, but I think it's safe to trust he was talking to game developers on both systems and at least as far as the actual points on performance (the specific things like 2x the performance of that, or 1/3 of the speed of that) actually came from developers. I wouldn't be in a hurry to write it off just because the comments are anonymous - obviously NDAs are still in effect. I'd have preferred word-for-word quotes from said developers, however.

more like he spoke to some pc-centric developers who were like "oh boo hoo, now we have to write multthreaded code, or we're doomed".
 
mrklaw said:
more like he spoke to some pc-centric developers who were like "oh boo hoo, now we have to write multthreaded code, or we're doomed".

Well yeah, I think it's definitely a case of waa-waa-ing over the gulf in design between these chips and PC chips too.

Unfortunately waa-waa-ing isn't innocuous..if some developers can't be arsed working with these chips instead of against them, their games will suffer, technically. Bad for them, of course, but not exactly great for us either if a lot of devs take the same attitude.
 
mrklaw said:
more like he spoke to some pc-centric developers who were like "oh boo hoo, now we have to write multthreaded code, or we're doomed".


I guess the crying will continue given that AMD and Intel have hinged their futures on multi-core designs as well. :) Maybe not the same scale, but multi-core nonetheless.

Holmdahl said that MS has incorporated Open-MP into their development kit. I imagine that most developers aren't familiar with it. I wonder how hard it will be for developers to transition?
 
"With much more powerful CPUs and, in the near future, more powerful GPUs, the PC paired with the right developers should be able to bring about that revolution in game physics and graphics we've been hoping for. Consoles will help accelerate the transition to multithreaded gaming, but it looks like it will take PC developers to bring about real change in things like game physics, AI and other non-visual elements of gaming."

I like how the 'geniuses' in the PC development world will be the only ones with the brains and sheer motherfuckin' hutzpah to harness a multi-threaded architecture. Those console devs like Kojima, Bungie, Naughty Dog, PolyD, Ubisoft et al will be playing in the kiddie pool where they belong. *insert rolling eyes gif here*

I dunno, Im no techy but this "analysis" seems utterly obtuse to me, and not just a little skewed either.
 
the PS3 and XCPU processors are more simplified processors. He doesn't seem to realise that single-threaded code on PS3 and XCPU will only be tapping well... the majority of PS3s power (not even using a single SPE) and using a single core of Xbox 360 without hyperthreading.
That's what he's been told by some developers though, and that's why it doesn't sound very good... I've been actually predicting this myself (developers mostly ignoring SPEs and two extra cores on Xenon, thus having "ideal" multiplatform situation), but hoping against it

Well, let's hope he's been talking with some crappy devs, I seriously don't understand the logic that even if SPEs can be used for physics acceleration that they are somehow waste of space. Physics alone on that level of acceleration is a huge thing, IMO.
 
Anand gets good sources and blah blah blah, but as gofreak said here, and I said in the last Anandtech thread, he runs a PC-centric site. He shows a basic lack of understanding of the console market. Single-threaded code shouldn't dominate the first-gen, b/c a number of console devs have experience with multi-core chips already...the EE. It's not exactly the same, but it gives them a leg up on lazy PC devs who are used to being spoonfed everything. Sure, cheap PC ports will be single-threaded, but we're not judging console performance by the bottom rung, we're gonna base it on the in-house teams and the Namcos and Squares who can push hardware from the first game. Well, maybe not Square, but you get my point. I expect more than a few multi-threaded games in the first gen. Heavenly Sword should be, and that's from a smallish dev with SCEE's publishing. I don't see why others can't do the same. PEACE.
 
I dunno, what Anand says agrees with what I've been hearing from devs. The next-gen will be a good leap in terms of getting consoles into the HD era. And yes, the GPU's really are the saving grace of the console.

Anyone who is a programmer knows that threading and synchronization is a bitch not only to implement but to debug. I would wager there are some devs from last gen that won't be able to handle working in a thread-heavy environment -- and so they won't even try. The result? Underwhelming performance.

I think MS/Sony would have been just as well off using a high-end P4 or AMD64 chip as Anand states. Ya, maybe MS saved some money...but what the hell is Sony doing? The cell is anything but amazing *with respect to console gaming.* Yes, Cell is impressive, but not in a console.
 
Well it seems Apple tends to agree with Anand, Jobs called the IBM PPC architecture a dead end and both the cell, and xenon chip is based on it. Jobs also put his money where his mouth was by dumping the powerPC, and swithing to x86 intel chips. (Sure there are other reasons like slow chip production, causing production delays, but the bottom line is the G5 was not progressing as fast as predicted. As jobs promised a G5 3 Ghz procosser over a year ago, which has yet to materialize.) It's also common knowledge that the PC's (x86) has been more powerful than Mac's. (PowerPC) This single fact should speaks volumes for Anands article.

Either way the CPU is not that important anymore. As someone pointed out the Cpu requirement for half life is a measly 1.2 Ghz pentium 4, and doom 3 is 1.5 Ghz. The biggest bottleneck is going to be the GPU, and both systems will have powerful GPU's so it's all good.
 
Pimpwerx said:
Anand gets good sources and blah blah blah, but as gofreak said here, and I said in the last Anandtech thread, he runs a PC-centric site. He shows a basic lack of understanding of the console market. Single-threaded code shouldn't dominate the first-gen, b/c a number of console devs have experience with multi-core chips already...the EE. It's not exactly the same, but it gives them a leg up on lazy PC devs who are used to being spoonfed everything. Sure, cheap PC ports will be single-threaded, but we're not judging console performance by the bottom rung, we're gonna base it on the in-house teams and the Namcos and Squares who can push hardware from the first game. Well, maybe not Square, but you get my point. I expect more than a few multi-threaded games in the first gen. Heavenly Sword should be, and that's from a smallish dev with SCEE's publishing. I don't see why others can't do the same. PEACE.



yeah, although it was bad news on the EE up until recently. Lots of devs not even using VU1, and pretty much only using VU0 in macro mode. Pretty close to running it as a single thread standard CPU.

Hopefully some of the better ones have gotten used to it, because while they might have got away with it on PS2 with only leaving out one vector unit, if they do that on PS3 the whole thing will grind to a halt.
 
fugimax said:
Ya, maybe MS saved some money...but what the hell is Sony doing? The cell is anything but amazing *with respect to console gaming.* Yes, Cell is impressive, but not in a console.

Care to elaborate?

As for the rest, performance is no longer going to come for "free". That's the same everywhere, not just in games - before we could rely on just simple increasing clockspeed to bring performance boosts, but no more. Everyone's moving horizontally now, not vertically.


Lancelet Pink said:
Either way the CPU is not that important anymore. As someone pointed out the Cpu requirement for half life is a measly 1.2 Ghz pentium 4, and doom 3 is 1.5 Ghz. The biggest bottleneck is going to be the GPU, and both systems will have powerful GPU's so it's all good.

If you want games to behave as realistically as they look, you can't bank on a relatively "weak" CPU providing that. And no, I don't think HL2 represents the absolute pinnacle of in-game behaviour and simulation.
 
Lancelet Pink said:
Either way the CPU is not that important anymore. As someone pointed out the Cpu requirement for half life is a measly 1.2 Ghz pentium 4, and doom 3 is 1.5 Ghz. The biggest bottleneck is going to be the GPU, and both systems will have powerful GPU's so it's all good.

oh please. GPUs are big news in PCs because they have such limited processing power, and shit buses to transfer data over. GPUs are the crutch that allow PCs to get away with having crap architectures.

CPUs will be *very* important next gen for games, and IMO if the PC CPU guys dont' buck up their ideas, it'll be a very long time before PC games catch up to the best offerings from X360 and PS3.

Maybe hardcore PC nutters will just install a physics card and a maths coprocessor card to help out on the FLOPs front? GPUs are good at drawing pictures. Next gen will be more than just pictures.
 
this is a little out of control. I've heard complaints, but mostly because developers dont have time to really develop for the platform, and are forced to use old practices in their current games.
 
Lancelet Pink said:
Well it seems Apple tends to agree with Anand, Jobs called the IBM PPC architecture a dead end and both the cell, and xenon chip is based on it. Jobs also put his money where his mouth was by dumping the powerPC, and swithing to x86 intel chips. (Sure there are other reasons like slow chip production, causing production delays, but the bottom line is the G5 was not progressing as fast as predicted. As jobs promised a G5 3 Ghz procosser over a year ago, which has yet to materialize.) It's also common knowledge that the PC's (x86) has been more powerful than Mac's. (PowerPC) This single fact should speaks volumes for Anands article.

Either way the CPU is not that important anymore. As someone pointed out the Cpu requirement for half life is a measly 1.2 Ghz pentium 4, and doom 3 is 1.5 Ghz. The biggest bottleneck is going to be the GPU, and both systems will have powerful GPU's so it's all good.

You're making the same mistake Anand is making...you're comparing a PC (or Mac) enviorment to a console. The difference in CPUS is that Intels and AMD's are meant to be very strong out-of-order chips and it seems as though his arguement is based around the idea of how well PC-centric code runs on these cpus.

From what I gather, games (which have to multi-threaded to get the most out of these consoles) are all designed around single-threaded code to run on PC's or Xbox's...heck most devs don't get the most out of the PS2 because they are not coding the EE properly.

I'm no techie, but this article really screams biased...or at the worst misinformed.
 
mrklaw said:
oh please. GPUs are big news in PCs because they have such limited processing power, and shit buses to transfer data over. GPUs are the crutch that allow PCs to get away with having crap architectures.

CPUs will be *very* important next gen for games, and IMO if the PC CPU guys dont' buck up their ideas, it'll be a very long time before PC games catch up to the best offerings from X360 and PS3.

Maybe hardcore PC nutters will just install a physics card and a maths coprocessor card to help out on the FLOPs front? GPUs are good at drawing pictures. Next gen will be more than just pictures.

First of all Creating games for PC's follows the rule of the lowest common denominator to capture as much potential buyers at it can. Where as the games for consoles are optimized for the system. Not to mention console games has always had the advantage of usually running low 640 x 480 resolutions. If a PC game is created solely for a single system configuration type it would more than equal any console game. And the GPU since it's inception has been more important than the CPU in terms of pushing games to the next level.






gofreak said:
If you want games to behave as realistically as they look, you can't bank on a relatively "weak" CPU providing that. And no, I don't think HL2 represents the absolute pinnacle of in-game behaviour and simulation.

Sure realistic AI, and physics are CPU dependant, but those are secondary to what the GPU brings to the table. Besides programming AI, is a beast and creating a realistic AI in a world like GTA's would be much more limited by the games budget, and time constraints than by the CPU. Either way both the cell and Xenon will be capable, who said HL2 was the pinnacle of in game behavioir or simulation?
 
Lancelet Pink said:
Sure realistic AI, and physics are CPU dependant, but those are secondary to what the GPU brings to the table.

Well that's a subjective point. In terms of sheer processing requirements, though, we're seeing dedicated physics chips appear in the PC space, which should give you an idea of future requirements..
 
Top Bottom