How come Xenos dose not get as much publicity as Cell?

[Sound_Card]

Sound_card, by your logic, MS should have put two Xenoses in the 360, rather than having the Xenon with it. You need to realize that it's only good at particular tasks, and that's why it's accompanied by a more general processor.

No, don't get me wrong here, I fully understand the purpose and need for a central proscessing unit.

As you can see, this wen't the wrong course. I never wanted to jump out of the game industry. Bah, my fault.

So let me rephrase my qustion... How come in the game industry(console), Cell is potrayed by media as a super chip that is sopposed to give PS3 the best graphics and performace, while Xenos in many ways dose wonders for the 360? Bah, I just give up! I knew this was a little dumb...

 

[Sound_Card]

Its obvious by now sound_card didn't make a question with this topic. He made a demand.

then my bad.

 

[liolio]

MS, in regard to ubber sony's PR, should have push xenos as a dirext X10 compliant gpu even if it's false.
Show some directX 10 demo and how it will be possible to do this on the 360 vs directX 9 rsx.

That's all, they have been conservative in the presentation of their hardware, when sony 's marketting have done an astouding job, in may 2005 at E3 the whole mass media have reported the pS3 is two time more powerfull than the ms system.
It's all ms mistake, it's sad but they havent lied enought.
the mindshare could have been differentor more even in the mass market minds.

 

[22psi]

MS, in regard to ubber sony's PR, should have push xenos as a dirext X10 compliant gpu even if it's false.
Show some directX 10 demo and how it will be possible to do this on the 360 vs directX 9 rsx.

That's all, they have been conservative in the presentation of their hardware, when sony 's marketting have done an astouding job, in may 2005 at E3 the whole mass media have reported the pS3 is two time more powerfull than the ms system.
It's all ms mistake, it's sad but they havent lied enought.
the mindshare could have been differentor more even in the mass market minds.

Well to be fair that was 2005. Look where Sony's PR/Marketing machine has got them now. I can't remember the last time there was some good news.

 

[Rainbow Man]

If it's any consolation to you liolio both MS and Nintendo have "lied" (though like most people I think, I'd prefer to say "exaggerated" instead) plenty in the past. MS with the original xbox, and Nintendo with the N64.

I'm sure Nintendo would have exaggerated just as much as suual with the gamecube if they'd dtill been in the position to do so by the time that thing launched..

It's in the nature of PR departments to exaggerate. Everybody's guilty of it, not just sony. Look at nvidia for example. they callde the original geforce fx an 8-pipe chip when it only had 4, that's as big a lie as you can get basically.


Peace.

 

[liolio]

Well to be fair that was 2005. Look where Sony's PR/Marketing machine has got them now. I can't remember the last time there was some good news.

In fact is more the fact that sony can't produce enought units that put them in such a situation.
But nobody wants to have another BR conversation here.

What is really fun in MS attitude is that that they are likely to be the worse company in world when it comes to unfair commercial practices.
Maybe MS have been conservative because they were scared to be sued again...... lol

 

[Lord Darkblade]

Sound Card, part of the point is that a graphics card generally does just that, it makes things shiny and pretty. While that may make many people salivate it is a very small field, there is the potential to move into the GPGPU field however we don't get a lot of information on that either, and indeed Xenos may not be the most useful in that field if its EDram design is not a big benefit (iirc 70mm^2 is a fairly hefty investment in silicon).

Cell however is a CPU, it has applications in physics, maths and many other aspects of games, in addition to being potentially a graphics device. It has been pushed by its manufacturers into a wider industry and is actually very well suited for defense work (radar etc is a constant stream of data, the dsp like config would likely be good), similarly in CE devices 1:2, 1:3, 1:4 Cells could be very useful as open source software can be ported to them (thinking linux based here) and also Cell has shown itself to be a monster at decoding (2 full streams @ 40mbps at the last article I read, while Waternoose apparently suffers under 1 stream).

So we have marketing, application to the "real world" and also area of interest. Cell might not be better than Xensos in a particular test that you want to run, but it is of interest in potentially a lot more areas which makes it big. Also this "cross field" capability is very big because it means that there are a lot of devs with experience in coding for it, there is a lot of software out there for it and it can support excellent security through the hypervisor system (or is that Sony only?)

 

[liolio]

Sound Card, part of the point is that a graphics card generally does just that, it makes things shiny and pretty. While that may make many people salivate it is a very small field, there is the potential to move into the GPGPU field however we don't get a lot of information on that either, and indeed Xenos may not be the most useful in that field if its EDram design is not a big benefit (iirc 70mm^2 is a fairly hefty investment in silicon).

Cell however is a CPU, it has applications in physics, maths and many other aspects of games, in addition to being potentially a graphics device. It has been pushed by its manufacturers into a wider industry and is actually very well suited for defense work (radar etc is a constant stream of data, the dsp like config would likely be good), similarly in CE devices 1:2, 1:3, 1:4 Cells could be very useful as open source software can be ported to them (thinking linux based here) and also Cell has shown itself to be a monster at decoding (2 full streams @ 40mbps at the last article I read, while Waternoose apparently suffers under 1 stream).

So we have marketing, application to the "real world" and also area of interest. Cell might not be better than Xensos in a particular test that you want to run, but it is of interest in potentially a lot more areas which makes it big. Also this "cross field" capability is very big because it means that there are a lot of devs with experience in coding for it, there is a lot of software out there for it and it can support excellent security through the hypervisor system (or is that Sony only?)

I don't think this is really the point, the mass market can't care less about the facts you're speaking about.
In may 2005, sony has push in gamers's mind that ps3 =2X 360, nobody has feel like 360's gpu is revolutionnary games will look astounding. I don't think this thread or the mindshare of the market in regard to compared power of these plateform has something to do with the technical merit of a cpu or a gpu.
Most people are technically agnostic and could have bitten any bullshit PR in regard to the power of these systems.

 

[Albuquerque]

Is there actually a high-end PC with a higher flop figure than the CELL, what kind of configuration do you have in mind?

The theoretical performance and actual performance of Cell are quite different. FLOPs are an interesting figure, but are not representative of how fast real work would get done. Howabout some real-life examples?

Between a P5 2.4ghz non-hyperthreaded and a 7-SPE cell, which one would perform the following tasks faster?

Defrag your hard drive?

Use IE to surf the B3D forums?

Print a picture of your family on your 4-color inkjet printer?

Spellcheck your resume?

Sync up your Mp3 player?

 

[liolio]

There a thread somewhere on this very forum about the future of X86,sse4,etc...

for almost the transistor budget of a cell a slightly cut down core2 duo with 2 altivec like simd unit (one per core)could beat the cell on every flop heavy real bench, cell will shine on very very few specialised workload.
I don't even want to speak about xenon.

 

[Cheezdoodles]

Is there actually a high-end PC with a higher flop figure than the CELL, what kind of configuration do you have in mind?

Do you actually know what a FLOP is?

Do you actually think that FLOP numbers is the best way to measure CPU performance for gaming?

Do you actually think that a $1000 Intel Conroe Quad Core wouldnt rape the Cell cpu in 99% of the cases when you play games?

FLOPS are mostly interesting if your going to do scientific stuff on it, or if you wanna build good chess simulators, its good for certain aspects of games aswell, but its definately not the end of all arguments when it comes to CPU performance.

Infact, FLOPS numbers can also be deciving. The P4 has at the same clockspeed, higher flop counts than what a n AMD 64 on the same clockspeed has. If we base performance only on FLOPS, this would mean that the P4 should be better for gaming. Everybody knows thats not the case, mainly because the AMD 64 has much faster memory access which is needed in real world applications, but not when you measure teoretical bs numbers.

 

[London Geezer]

Do you actually think that a $1000 Intel Conroe Quad Core wouldnt rape the Cell cpu in 99% of the cases when you play games?

Heh, if it didn't, i'd start questioning Intel's pricing big time! Oh wait...

Infact, FLOPS numbers can also be deciving. The P4 has at the same clockspeed, higher flop counts than what a n AMD 64 on the same clockspeed has. If we base performance only on FLOPS, this would mean that the P4 should be better for gaming. Everybody knows thats not the case, mainly because the AMD 64 has much faster memory access which is needed in real world applications, but not when you measure teoretical bs numbers.

(Genuine question) How does Cell memory access and bandwidth compare to Conroe? I know that Cell has pretty high bandwidth inside the ship itself and also to memory thanks to FlexIO, but i know very little about the Intel chip...

 

[scooby_dooby]

Ya, sorry if it sounds dumb... but I can't help but think.... why?


It seems to me that Xenos dose more for 360 than Cell dose for PS3. Every dev out their can say how easy it is to code for 360, and most of this can be thanked by the USA in Xenos. I mean when it comes to it... what would you rather build your console from? Cell... or Xenos..? I have no doubt in my mind that most of you would pick Xenos including my self. But the problem is.... Cell is just in this spot light of glory among magazines and web sites, and I cant help but wonder why Xenos is not getting fair treatment?

It's a good question. Xenos is MS's saving grace, and is their true masterstroke in this generation. ATI allowed them to launch a year earlier, and have a system that matched one launched 12 months later.

It's almost unthinkable really, but it seems they pulled it off.

 

[scooby_dooby]

But ultimately, I think the reason Cell gets more publicity... is because it exists outside of the gaming space as well, whereas Xenos does not. As soon as third-party vendors start putting Xenos in servers and supercomputers and such, you'll see the publicity. And of course, we both know that's not happening. So...

Sort of, but I see it from a slightly different angle. CELL is a product that IBM and Sony are trying to sell to prospective clients, and therefore there is a huge hype train following CELL. Regular releases of benchmark showing it's dominance, regular interviews with engineers extolling it's virtues, lots of news articles singing praise.

It's all part of the marketing machine imo.

 

[Crossbar]

Do you actually know what a FLOP is?

Yes I do, thank you very much. Do you know the meaning of the word rape?

Do you actually think that FLOP numbers is the best way to measure CPU performance for gaming?

I think it´s one of the better simpler ones for game software, if you know a best way, please enlighten me.

Do you actually think that a $1000 Intel Conroe Quad Core wouldnt rape the Cell cpu in 99% of the cases when you play games?

Thanks, that was the answer to my question. Not a lot of those Quad Cores around, and even less casual people that know they even exist.

 

[Cheezdoodles]

I think it´s one of the better simpler ones for game software, if you know a best way, please enlighten me.

The best way, would be real benchmarks. FLOPS numbers alone mean nothing. Remember back in the day, when people measured FLOPS numbers on GPU's? The ATI X1900XTX is theoretically capable of 550GFLOPS. If FLOPS was any real indication of performance (in a real world scenario) it should be close to twice as fast as the G70\71's, its not.

Thanks, that was the answer to my question. Not a lot of those Quad Cores around, and even less casual people that know they even exist.

So what? You asked for a high end cpu that had more flops.. Well the Quad-Conroe doesnt have a higher flop count, but it would still rape it in 99% of the cases. Infact, im confident than a normal dual-cored Conroe extreme would rape it in 99% of the cases. It would however, loose in heavy vector processing and drawing 3d graphics. Then again, you would use your GPU to draw 3d graphics in a real world scenario.

The Cell, (in contrast to what you would like to belive) is NOT a supercomputer. It is however cheap compared to the performance.

 

[Fox5]

Not to mention all the CELLs that don't make the 3.2GHz cut within the power envelope.

This is the only way CELL is ever going to make economic sense in CE products. Otherwise it would be up against dedicated solutions from Broadcom and others with significantly lower cost and power usage.

Cheers

Cell is a stream processor, is it not? Stanford did some research not too long ago on Stream processors and found that their combination of a general purpose processor and essentially controlled DSPs could have about 2x to 3x the performance per power usage of just a dedicated DSP alone since it would allow algorithms not possible otherwise. See the Imagine stream processor.
As for lower cost, don't know about that.

they callde the original geforce fx an 8-pipe chip when it only had 4, that's as big a lie as you can get basically.

16 pipe GS.

for almost the transistor budget of a cell a slightly cut down core2 duo with 2 altivec like simd unit (one per core)could beat the cell on every flop heavy real bench, cell will shine on very very few specialised workload.

By that logic, shouldn't a dual core G5 already beat the Cell in just about every flop heavy real bench? (Stanford's research indicates a true stream processor blows away a traditional processor with SIMD in most flops heavy DSP type stuff, while losing out in more traditional applications)

 

[patsu]

No, don't get me wrong here, I fully understand the purpose and need for a central proscessing unit.

As you can see, this wen't the wrong course. I never wanted to jump out of the game industry. Bah, my fault.

So let me rephrase my qustion... How come in the game industry(console), Cell is potrayed by media as a super chip that is sopposed to give PS3 the best graphics and performace, while Xenos in many ways dose wonders for the 360? Bah, I just give up! I knew this was a little dumb...

You answered your own question...

Xenos does wonder mainly for the graphics pipeline of 360. That is expected of a GPU.

As a CPU, Cell does wonder for all of PS3's tasks (e.g., helping RSX out in graphics, solving physics equations, decoding hi-end audio, playing BR video efficiently, processing network packets, ... all in a small, efficient package). It will also pave the road for PS4 (likely to be Cell-like to preserve their current investments).

I would give both systems more time to see how far they can go once the tools, OS and experiences mature. Cell's value will grow over time if it establishes itself and outlives PS3 (easier backward compatibility in the future ?).

Sort of, but I see it from a slightly different angle. CELL is a product that IBM and Sony are trying to sell to prospective clients, and therefore there is a huge hype train following CELL. Regular releases of benchmark showing it's dominance, regular interviews with engineers extolling it's virtues, lots of news articles singing praise.

It's all part of the marketing machine imo.

Cell has its inherent X-factors and sex appeal, that's why people are talking about it (See below). It may not be fair to brush all off as hype (although there are exagerations). I also don't think IBM and Sony spend much dollars on hyping Cell. The audience and media took it and ran.

The theoretical performance and actual performance of Cell are quite different. FLOPs are an interesting figure, but are not representative of how fast real work would get done. Howabout some real-life examples?

Between a P5 2.4ghz non-hyperthreaded and a 7-SPE cell, which one would perform the following tasks faster?

Defrag your hard drive?

Use IE to surf the B3D forums?

Print a picture of your family on your 4-color inkjet printer?

Spellcheck your resume?

Sync up your Mp3 player?

Firstly, Cell is a brand new architecture (Part of the reason why people are talking so much about it ). If it takes root, these applications (and OS !) will be optimized for it. Today web surfing on PS3 is already as fast as my P4 3.0Ghz 1 Gb RAM laptop. Tomorrow who knows. There are at least 2 papers/articles suggesting how to program Cell better. We will see Cell variants in the future too.

People forget that they are comparing a 1.5 month old baby with a 10+ (probably longer) year old, established teen in the CPU world. There are indications the mainstream CPU makers are considering a Cell-like future product. But Cell, in its infancy, is here today. At the same time, impatient people are waiting to judge its success/failure. These generated much discussions.

Secondly, Cell includes design elements from specialized units into its "core". Some people end up comparing a (GP)GPU or DSP with it. They can't seem to place Cell into traditional "pigeon holes". People are also trying to understand how to harness this power. This is IMHO, the second reason why people talk about Cell.

Thirdly, Cell shrinks much power and CPU versatility into an everyday package _at home_. This is only possible with huge upfront investments and distribution of IBM, Toshiba and Sony combined. These are very real marketing factors and not mere hype. Enterprising people are just starting to find new applications for it. This is the third reason for Cell talks.

Cell has its justified technical merits based on what we are seeing today in BR playback and scientific computing. These are its beach-heads/landing points. All it needs now is time, support and em... some love to grow. I hope it succeed coz I'm a techie at heart. Bold move like this, if rewarded, encourages further innovation.

EDIT:
I forgot a fourth reason:

Cell adopts an aggressive design philosophy... sacrificing some "common" features for more gain in peak power. Since this is a technical board, it is only natural that people debate the relative merits of these trade offs. So there... Cell is a hot topic in its own right.

 

[Cheezdoodles]

Y
People forget that they are comparing a 1.5 month old baby with a 10+ (probably longer) year old, established teens in the CPU world. .

You forget that the Cell has been around longer than the PS3

but i agree that its very new compared to the traditional cpu's.

 

[patsu]

You forget that the Cell has been around longer than the PS3

but i agree that its very new compared to the traditional cpu's.

By golly you're right. Cell was available in a blade box about a year ago. But that's still for niche imaging/military/scientific computing communities. For the home market and everyday use, it's only 1.5 months ago.