Anandtech dashes cpus from ps3 and xbox360

[Carl B]

Is it really worth further damaging your credibility for 5 cents a hit, Anand?

This shit is bananas - B-A-N-A-N-A-S!

MAN but that song's been on my mind lately!

 

[Titanio]

Also, for those that think Anand are on the case this time around, look at HS runnin ATM with only a single thread and n hundred enemies on screen with their own AIs. Forgive me for being naive, but that looks a little bit better than 2x the performance of XB...

or kameo with 1000 trolls or 99 nights with 2000+ on screen characters!

Kind of a side note but..

Though these examples may prove a point relative to Xbox performance, I'm not sure if they prove as much wrt to expectation of next-gen CPU performance. Doing something like that is possibly more limited by the GPU than CPU - individually those characters aren't going to be very intelligent, and in terms of physical modelling, with such a busy scene you could probably introduce wider tolerances of less-than-perfect collision detection/response to compensate for the numbers. Note also that, wrt the trolls in Kameo at least, they're not very geometrically complex, probably to help with speeding up collision detection and so forth. You could implement a scheme to scale CPU work also depending on a troll's distance from the camera/main character - those far away may not be up to very much at all, or have much if any physically modelling, while those closer may be modelled more fully to maintain the player's illusion. In other words, 1000 trolls in a scene like that isn't necessarily the equivalent of 1000 main AI characters in even last gen games, or certainly many current PC games.

 

[V3]

Not really of a joke, its a very efficient CPU in term of transistor usage. Its not a luxury CPU like Athlon 64. So is suitable for console.

Next generation consoles don't usually have a CPU only twice as fast as its predecessor do they?

Well if you haven't notice, they sort of hit a wall in regard to CPU.

That's why we are getting multicore in the first place.

 

[Tacitblue]

Maybe he was trying to outdo Jon "Hannibal" Stokes Arstechnica article. They both were very odd to say the least. And he obviously didn't bother to visit IBM's site to check out the DD2 core revision for the Cell. The PPE in it is some 40% larger than the prototype.

 

[Carl B]

That's going to be cheap cheap cheap to produce!

Look at the Cell diegraph - 1MB of L2 alone takes up nearly 70mm2. Unless the XeCores are sitting in vacuum, you're not gonna be producing the cpu at that size.

Hey I'm just quoting them as saying it's half the size of a current P4; to which I gave the size. Plus, since a P4 has 2 megs of cache as it is, it's a great candidate for splitting down the middle.

BUT, if you're looking to dispute the article's claim go right ahead - because my point in raising it in the first place was that it sounded ridiculous to me.

 

[Shifty Geezer]

True. the added RAM helps too, so it's not a processor only thing. I could point out the Rubber Ducks and fluid dynamics demo for Cell's performance, and ask for an equivalent showing on PC. Though Anand did (reluctantly) admit SPE's might see some use for physics. Though it'll be 4-5 years before we see that kind of physics from the consoles, apparently, while PC's wil be then in a kiffy...

I think the overall impression is the author spoke with his PC developing chums, who are used to writing PC code (which can be really crappy and still perform well) and when they saw they had to actually learn to write DECENT code they balked. Kinda like anyone can cook a good dinner if it's a microwave meal (assuming there is such a things as a GOOD microwave meal!) but to be a chef needs training and talent. Any mug can write PC code but to write code that takes advantage of the new processors will need smarts and talent. Without info on who these developers are or what their experiences with the platforms are, I'm gonna file them away under 'extraordinarily unlikely to have any merit'. Plus the idea that both companies can be hoodwinked into producing crap CPU's...I return to my very first post this thread. It's just so darned laughable.

 

[wireframe]

I normally enjoy Anandtech's articles, but this one has something smelly going on. Perhaps I am not reading it right, Anand is not expressing his ideas correctly, or he has really come to what seem like absolutely absurd conclusions.

For one, the Cell's total area is dominated by the array of SPEs (8, with one disabled for a total of 7 for the PS3 design). In fact, about 2/3rds of the entire die is consumed by the SPEs and their related logic. Now, are you going to tell me that these SPEs are mostly useless? Sony could save a fortune by cutting them out or at least limiting the design to 2 SPEs (or even four) if that was so. They would be crazy to produce the Cell as it stands if Anand's claims are true. Now, I am sure he will be quick to point out that they were not his claims, but claims of anonymous developers. However, this doesn't make it less absurd.

Something is wrong here... Maybe I would start believing "total faliure" if it was a discussion of one design, but when the article manages to make two designs sound like something you would only want running your microwave, the mind boggles.

 

[Acert93]

Is this guy saying that a single core is twice the performance of XBox's CPU? Or is he saying the entire CPU is twice the performance? If its the second option then he surely has to be talking crap.

Though I am starting to get the impression that this CPU isn't going to be quite the power house people are expecting.

The entire CPU. And is roughly about right for real world performance.

Let me get this right. For real world performance, you are guestimating that a single PPC core (like the ones in the XeCPU) with 333K cache and running at 3.2GHz is ~2/3 as powerful as a Celeron 733MHz with 128K cache *in the real world* ?

Is that correct?

In order, deeply pipeline, what are you expecting ? This thing is not P4 that's for sure.

But we are NOT talking about a P4, now are we. You are talking about a Celeron with a ~1/3 the L2 cache and 25% of the frequency clobbering a PPC core by 50% in realworld performance.

So where are you getting your reference that in realworld performance XeCPU should be ~2x as fast as a Celeron 733MHz? What data are you relying on to arrive at this conclusion?

I would be interested to know how you arrive at this conclusion.

 

[dcforest]

As you read AnandTech's article, you have to keep in mind that AnandTech primarily supports the PC industry and is looking at console gaming to see how it might impact PC gaming. Not surprisingly they conclude that the next generation consoles are nice machines, but won't be able to compete with high-end PC gaming. (see "Final Words" in the article).

For example, the XBOX 360 when it ships in Nov/Dec for an estimated $300-400 is going to be an excellent gaming value and no PC at the same price range will be able to touch it. But the high-end gaming PC's that ship in Nov/Dec will substantially outperform the XBOX 360 in terms of CPU performance, GPU performance, memory and storage capacity. Over time the gap between the next-gen consoles and high-end gaming PC's will only widen. Now the high-end gaming PC will cost thousands of dollars, so it's not a fair fight, but expect the PC platform to deliver the best visuals, game physics, AI, etc. now and in the future.

 

[therealskywolf]

Is this guy saying that a single core is twice the performance of XBox's CPU? Or is he saying the entire CPU is twice the performance? If its the second option then he surely has to be talking crap.

Though I am starting to get the impression that this CPU isn't going to be quite the power house people are expecting.

The entire CPU. And is roughly about right for real world performance.

Let me get this right. For real world performance, you are guestimating that a single PPC core (like the ones in the XeCPU) with 333K cache and running at 3.2GHz is ~2/3 as powerful as a Celeron 733MHz with 128K cache *in the real world* ?

Is that correct?

In order, deeply pipeline, what are you expecting ? This thing is not P4 that's for sure.

But we are NOT talking about a P4, now are we. You are talking about a Celeron with a ~1/3 the L2 cache and 25% of the frequency clobbering a PPC core by 50% in realworld performance.

So where are you getting your reference that in realworld performance XeCPU should be ~2x as fast as a Celeron 733MHz? What data are you relying on to arrive at this conclusion?

I would be interested to know how you arrive at this conclusion.

Pc Small time Devs putting OoO code, 1 thread, running on it. Un optimized basically.....it could be possible.

 

[Megadrive1988]

slightly off topic ramble: maybe we will see the emergence of a new class of PROCESSOR that does EVERYTHING: general purpose code, programmable floating point / streaming code, vertex/ geometry processing, rasterizing, etc. and does each task fairly well. better than Cell and Xenon at general purpose code, plus all graphics processing. and then we get parallal PROCESSORS in systems. 2 in laptops and set-top boxes, 4 in a console, 4 in a personal computer, 4-8 in an arcade machine, and 1 in handhelds. or whatever. The graphics portion of this PROCESSOR that I am dreaming of gets the benefit of much higher CPU-like core clockspeeds. plus the move away from scan-conversion rendering, to something more like raytracing. I am just throwing a few thoughts of my own out there, I am not fron a technology background obviously but I do enjoy watching technology advance and grow. in the coming 6-7 years, technology beyond Xbox360 and PS3/Cell might emerge, that will attempt to address all of the major problems of low-cost high-performance CPUs and GPUs that are intended for consumer computers and consumer electronics / consoles, and do at least a decent job of it. nothing will be perfect or completely ideal, but I expect some major improvements 8)

 

[Titanio]

For one, the Cell's total area is dominated by the array of SPEs (8, with one disabled for a total of 7 for the PS3 design). In fact, about 2/3rds of the entire die is consumed by the SPEs and their related logic. Now, are you going to tell me that these SPEs are mostly useless?

He explicitly says that, but one of the points alone he makes completely contradicts that (physics acceleration). Maybe he's trying to put a negative spin on things relative to PCs as suggested, and thus has to try and "drag down" PS3/cell too, or maybe he just weighs every CPU task equally (which is just wrong).

 

[Teasy]

Well if you haven't notice, they sort of hit a wall in regard to CPU.

That's why we are getting multicore in the first place.

But that only makes sense if your using multiple cores to actually surpass what can be achieved with a single core CPU.

 

[Acert93]

Pc Small time Devs putting OoO code, 1 thread, running on it. Un optimized basically.....it could be possible.

But we are not talking about 1 thread or whatever.

We are talking about realworld performance, i.e. what you would expect to be able to get out of a processor. Not worse case scenarios. In the scenario you are painting you are talking about the equivalent of a PC dev trying to write PC code for a PS2.

Yet that has nothing to do with real world performance and everything to do with stupidity.

I am trying to isolate where V3 is deriving the conclusion that the Xbox CPU is 50% than a single 360 core in realworld performance (not worse case scenario).

 

[russo121]

"...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."

Why this is not going to happen with consoles? Consoles paired with the right developers should.... What is near future to him? 1 year? 2? 5?

"...it will take PC developers to bring about real change..." - This one kills me

 

[Readykilowatt]

"...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.

Interesting.

It's not hard to forgive the hardware publishers for a little bit of hyperbole at E3, the annual trade show of the video game industry. It is, after all, their moment in the sun. But now that the crowds have gone home and the booth babes have changed back into street clothes, it's time to recognize that a fair number of the promises made last week will quietly fade away.

Need proof? Look no further than the introductory days of the PlayStation 2 or Xbox.

The Xbox was supposed to have resolution that went beyond HDTV and have a graphics chip three times beyond that of the PC. Ultimately, only a handful of games have offered 1080i resolution (the current standard for high-def). Most titles offering advanced graphics stick with 480p resolution, which is lower than high definition. And PCs had nVidia's (Research) GeForce 3 (which featured a graphics chip comparable to that found in the Xbox) months before the console launched.

http://money.cnn.com/2005/05/26/commentary/game_over/column_gaming/index.htm

Hmmm?

Revolution to be two-to-three times more powerful than GameCube?

Things that make you go Hmmm?

 

[therealskywolf]

Pc Small time Devs putting OoO code, 1 thread, running on it. Un optimized basically.....it could be possible.

But we are not talking about 1 thread or whatever.

We are talking about realworld performance, i.e. what you would expect to be able to get out of a processor. Not worse case scenarios. In the scenario you are painting you are talking about the equivalent of a PC dev trying to write PC code for a PS2.

Yet that has nothing to do with real world performance and everything to do with stupidity.

I am trying to isolate where V3 is deriving the conclusion that the Xbox CPU is 50% than a single 360 core in realworld performance (not worse case scenario).

6x times the processing power, well were we expecting much more? 6x times is alot.

 

[X-AleX]

As you read AnandTech's article, you have to keep in mind that AnandTech primarily supports the PC industry and is looking at console gaming to see how it might impact PC gaming. Not surprisingly they conclude that the next generation consoles are nice machines, but won't be able to compete with high-end PC gaming. (see "Final Words" in the article).

For example, the XBOX 360 when it ships in Nov/Dec for an estimated $300-400 is going to be an excellent gaming value and no PC at the same price range will be able to touch it. But the high-end gaming PC's that ship in Nov/Dec will substantially outperform the XBOX 360 in terms of CPU performance, GPU performance, memory and storage capacity. Over time the gap between the next-gen consoles and high-end gaming PC's will only widen. Now the high-end gaming PC will cost thousands of dollars, so it's not a fair fight, but expect the PC platform to deliver the best visuals, game physics, AI, etc. now and in the future.

This is totally false.
X360 will have the GPU performance edge for about an year, thus delivering the best visuals for that time. Then PC will gain back that crown.

 

[Teasy]

For example, the XBOX 360 when it ships in Nov/Dec for an estimated $300-400 is going to be an excellent gaming value and no PC at the same price range will be able to touch it. But the high-end gaming PC's that ship in Nov/Dec will substantially outperform the XBOX 360 in terms of CPU performance, GPU performance, memory and storage capacity. Over time the gap between the next-gen consoles and high-end gaming PC's will only widen. Now the high-end gaming PC will cost thousands of dollars, so it's not a fair fight, but expect the PC platform to deliver the best visuals, game physics, AI, etc. now and in the future.

Having more raw power and actually using that power are two different things. High end PC's released at the same time as 360 will be more powerful but PC games will not use that power. As usual in the first year or two of release consoles will have the best looking games.

 

[Acert93]

6x times the processing power, well were we expecting much more? 6x times is alot.

50% faster. I guess I skipped over that word in the very last sentance. If you reread the entire post/echange it is pretty clear though. He is saying the Xbox CPU is 50% faster than a 360 core, and the 360 CPU (3 cores) is 2x as fast as the Xbox CPU. NOT 6x.