A PS3 breakdown of all components.

[Johnny Awesome]

Why are we discussing this garbage article?

 

[Fox5]

If there was a bigger gap between the platforms we could say more. They took very different approaches, but in the end they are on the same process. MS rushed to get the X360 out -- with some shortages -- and Sony delayed their launch. So it seems they are farther apart in release than they are in many ways.

So PS3 is coming out a year later with roughly equivalent hardware?
Anyhow, what really matters more is the developers on each platform, and right now it looks like PS3 has a lot more triple A devs than xbox 360 does, so until that changes I say ps3 will have the better looking games.

 

[karlotta]

...Who cares if Xenos can do vertex texturing, coherant memory reads, hardware tesselation, and so forth if no uses it (for whatever reason). ...

Ima nerd i care. But duh , yeah its about SW.

 

[Pozer]

So PS3 is coming out a year later with roughly equivalent hardware?
Anyhow, what really matters more is the developers on each platform, and right now it looks like PS3 has a lot more triple A devs than xbox 360 does, so until that changes I say ps3 will have the better looking games.

AAA developing in past generations do not guarantee AAA work in future generations. Each generation new Rockstars, SCEAs, Bungies, Epics are born and some Segas, Ataris, and Konami's don't carry the same prestige they once did. Looking back in 5 years you'll be surprised where the AAA games actually came from.

 

[Rolf N]

Why are we discussing this garbage article?

Because it's such a large heap of garbage that it just has to be significant

I like how the guy believes that a "control" thread that dispatches work to concurrent threads on other cores actually needs to copy through all the payload data. This is my choice quote:

Having the PPE fill a task manager role may also means that all SPEs report or send its data back to the PPE. This has a negative impact on achievable bandwidth as the EIB doesn’t perform as well when massive amounts of data are all goin to a single destination element inside the Cell.

Wtf.

edit: The good thing about this is that there's a lot of correct information there. You just have to filter away all the gibberish.

 

[Fox5]

AAA developing in past generations do not guarantee AAA work in future generations. Each generation new Rockstars, SCEAs, Bungies, Epics are born and some Segas, Ataris, and Konami's don't carry the same prestige they once did. Looking back in 5 years you'll be surprised where the AAA games actually came from.

I was referring more graphically than best sellers or good reviews. But yeah, you're right, the industry can change pretty fast, but the devs that used to put out good graphics seem to still put out good graphics. Sony has most of them, while Microsoft has Bungie, Epic, and Rare, and Bungie has barely shown anything, Epic's effort looks fantastic, and Rare's efforts have looked really good though a bit rushed. (I don't think rare's games sold well though, which is why they are kind of overlooked but they do look fantastic imo)

 

[EbonySeraphim]

Greetings

Hello, I am the author of that post on the PS3Forums. A few quick things while I have time before leaving for work:

The most criticism of the post is based on my bias. I am biased. I even though I have done a lot more Java and C# work recently, I still have a soft spot for C/C++ and low level libraries. That is probably why a comment said that I was ignoring the development side(cost?) of things. Software development has to re-work solutions if computer hardware is evolving. If software designers are unwilling to look at potential massive hardware improvements due to the costs at the time, then the industry will evolve at the speed of a crawl.

My bias further comes in play when I see that one of my goals (which I wasn't concious of until reading early comments) was to show that the PS3 brings benefits to the table, and that it is possible to develop for. It's pro-Sony due to my ignorance of the Xenos, which I have discovered far more about since writing it and being badgered with questions. It's actually quite disturbing how I can argue against "myself" from two weeks ago and I didn't get that kind of information from people with opposing views. I am reviewing the graphics comparision section between PS3 and XBox360. Although, the processor is biased towards Sony because it is trying to shine a brighter light on the Cell and development for it. Pretty much due to my goals above, you could say I'm supporting the Cell, and thus supporting Sony.

Although I am biased, invalidating the entire post's content is pretty stupid. Even textbooks are biased. If you are well technically grounded (more so than me) See through the bias and try to pick out what isn't biased. I did get in an arguement who has probably been programming since longer than I've been alive and he seemed to lack something which I hope doesn't plague the development world too much - just because a problem was solved a certain way N times, people tend to lock themselves in that solution without reconsidering what formed it in the first place. It's not good when the premise for heading towards a solution changes, yet people completely miss it because they are stuck in their old ways. Basically, I'm sure people here and everywhere have far more technical knowledge and experience than me. Just make sure you aren't shutting yourself out in this way.

And yes, there are nasty sections if you "zoom in" and take a look at some focused quotes. Additionaly, I attempted to present a more easy to understand post which inevitably skipped certain nitty gritty facts. If something seems entirely wrong, at least make a slight effort to see what it's trying to say before labeling it "garbage."

I'm out of time.

 

[Geo]

Thanks for stopping by, EbonySeraphim.

Gentlemen, responses/questioning may be pointed, but all rotten fruit, sharp objects, and blunt instruments must be checked at the door. . . :smile:

 

[Black Dragon37]

I'm wondering why he compared the PS3 to Major Nelson's comparison of the Xbox 360 and the PS3?

 

[Shifty Geezer]

Hello, I am the author of that post on the PS3Forums. A few quick things while I have time before leaving for work:...Although, the processor is biased towards Sony because it is trying to shine a brighter light on the Cell and development for it. Pretty much due to my goals above, you could say I'm supporting the Cell, and thus supporting Sony.

This is the strength and weakness of the work. If you had stuck to talking about PS3's strengths, it could have been a fine article on PS3's potential. However, your comparisons are 'broken' as a work to compare different hardwares, both as hardware platforms and as gaming rigs, which doesn't do the effort involved in it's creation justice and leads people to discredit it out of hand.

I'm right with you on the software front for Cell - ideas need to adapt and evolve. Because of this, Cell has a lot of potential. You cannot categorically go from this reasoning to decide PS3 will have the better games though. It could be responsible for worse games overall. As an analogy, consider a TVR Speed6 versus a Lotus Elise on an average race course. With bucket loads more power, the TVR is capable of vastly outperforming the Elise, given a driver with the talent and expertise to handle it. For most drivers, the Elise will give a faster time and beat the TVR in similarly average driver's hands. Raw power does not equate to better performance. It's useable, attainable power that matters. There is an argument that most devs won't be able to adapt to Cell and use it properly. They will stick to prior knowledge, not explore, because it costs too much or even haven't the education (most uni courses don't provide any insight into hardware bottlenecks and solutions appropriate for closed platform development), and thus the available power goes mostly to waste. PS2 shows this - some of it's resources have gone untouched a lot of the time because devs weren't in a position to use it. From a software comparison, it's quite possible XB360 will get more useful work done because it's more conventional. However, XeCPU shares some of Cell's limits, and I think the paradigm shifts needed to make the most of these parallel, float heavy architectures will favour Cell over XB360. The question is, how much are devs on average going to invest in developing these new techs?

For the GPU discussion, you did overlook some important info and made some painfully false observations/comparisons, like US not being the way to go because ATi haven't used it in their latest part. Unified shaders are more than just consolidating similar functions from two specialist devices into one. It provides considerable flexibility with the potential for much improved efficiency, even if the individual pipes underperform relative to fixed function units in classical GPUs (a matter we have no metrics on).

You would do well to read through some of the (epic) discussions on this forum to see some of the comparative strengths of the XB360, especially the GPU, and read up a few of the reservations about Cell. It might well not change your overall views, but it'll at least give you a fair grounding to those views. Then when you harrang the masses, you can do so without the chunks of wrongness that discredit your work and make people overlook the points that are right.

 

[scooby_dooby]

Hello, I am the author of that post on the PS3Forums. A few quick things while I have time before leaving for work:

The most criticism of the post is based on my bias. I am biased. I even though I have done a lot more Java and C# work recently, I still have a soft spot for C/C++ and low level libraries. That is probably why a comment said that I was ignoring the development side(cost?) of things. Software development has to re-work solutions if computer hardware is evolving. If software designers are unwilling to look at potential massive hardware improvements due to the costs at the time, then the industry will evolve at the speed of a crawl.

My bias further comes in play when I see that one of my goals (which I wasn't concious of until reading early comments) was to show that the PS3 brings benefits to the table, and that it is possible to develop for. It's pro-Sony due to my ignorance of the Xenos, which I have discovered far more about since writing it and being badgered with questions. It's actually quite disturbing how I can argue against "myself" from two weeks ago and I didn't get that kind of information from people with opposing views. I am reviewing the graphics comparision section between PS3 and XBox360. Although, the processor is biased towards Sony because it is trying to shine a brighter light on the Cell and development for it. Pretty much due to my goals above, you could say I'm supporting the Cell, and thus supporting Sony.

Although I am biased, invalidating the entire post's content is pretty stupid. Even textbooks are biased. If you are well technically grounded (more so than me) See through the bias and try to pick out what isn't biased. I did get in an arguement who has probably been programming since longer than I've been alive and he seemed to lack something which I hope doesn't plague the development world too much - just because a problem was solved a certain way N times, people tend to lock themselves in that solution without reconsidering what formed it in the first place. It's not good when the premise for heading towards a solution changes, yet people completely miss it because they are stuck in their old ways. Basically, I'm sure people here and everywhere have far more technical knowledge and experience than me. Just make sure you aren't shutting yourself out in this way.

And yes, there are nasty sections if you "zoom in" and take a look at some focused quotes. Additionaly, I attempted to present a more easy to understand post which inevitably skipped certain nitty gritty facts. If something seems entirely wrong, at least make a slight effort to see what it's trying to say before labeling it "garbage."

I'm out of time.

It was a good write-up. I would be really interested if you would do as much research in Xenos, and give us a similar breakdown of it's potential.

Really the 2 most interesting things about these consoles is PS3's CELL, and 360's Xenos, we need could really use some more detailed info on the latter

 

[Acert93]

Really the 2 most interesting things about these consoles is PS3's CELL, and 360's Xenos, we need could really use some more detailed info on the latter

Already been done by someone who understands GPUs. And it is less agenda driven and less "X is better than Y" focused but instead a discussion of the technology and reasons for those decisions allowing the reader to draw their own conclusions about whether those are the right decisions for the market.

You can find a reason to poo-poo on any hardware. Like I pointed out yesterday, when the conclusion precedes the analysis you will arrive at things such as US are not the right direction because ATI's R5x0 desktop processors did not incorperate such. We have had plenting of such heavy leaning discussions on these forums and little comes out of them other than platform advocacy. I believe that is why the mod team actually banned such discussions for the longest time.

Each hardware has different wins and draws, what is "better" will ultimately depend on a lot of factors outside the hardware like dev skill and resources, game design, and so forth. As we have seen from the comments from id Software, Valve Software, Crytek, Starbreeze, etc they all have different opinions based on where they are coming from and where they are going in regards to technology and game design.

I think the best we can come up with is reserved, "X is app. _% better than Y in task_A" with adjunct summary of various workflow scenarios. e.g. RSX has a peak 4.4Gpixel fillrate, yet noting the impact and limitations of bandwidth and MSAA put on this theoretical peak.

Anything beyond that, especially with the tendancy [even declared goal] to favor a specific platform will break down into chronic arguments that have little to do with the technical merits but instead angling to put one's platform of choice ahead of the competition, which really defeats the purpose of a technical forum like B3D.

Which is why a site like Xboxyde/Playsyde have segregated forums so fans of each platform can talk about such things in fanspeak all they want because on fan forums it is to be expected.

 

[Jesus2006]

I think the best we can come up with is reserved, "X is app. _% better than Y in task_A" with adjunct summary of various workflow scenarios. e.g. RSX has a peak 4.4Gpixel fillrate, yet noting the impact and limitations of bandwidth and MSAA put on this theoretical peak.

Not that i disagree, but isn't it after all a little bit more black and white as one migh suggest? I mean Xenos has several advantages over the RSX, features which the PS3 GPU does not have like the EDRAM and the Unified Shaders, but this is not only positive, and this should be enlightened as well.

RSX has no hardware support to accelerate MSAA like Xenos has with its EDRAM, but is it worse because of this?

Looking at the current Xbox 360 games, i can only come to one conclusion - no.
Still there are hardly any games using MSAA at all, possibly because the cost of higher geometry / Z-only pass when tiling is too high for "real" games in contrast to theory?

On the other hand there are already games confirmed for PS3 that have 4xAA from the beginning and it doesnt seem to be much of a problem at all, despite the PS3 has no such accelerator, maybe because of the work balancing between CELL-RSX?

So it comes to my mind that lots of the theoretical features of Xenos are neglectible when it comes to real games, because of the limitations coming along with these feauters.

I might be wrong though

 

[scooby_dooby]

I guess that would be similar to assuming the extra processing power provided by the CELL SPU's are not useful because their impact is not evident in the first generation of games.

 

[Jesus2006]

I guess that would be similar to assuming the extra processing power provided by the CELL SPU's are not useful because their impact is not evident in the first generation of games.

Sorry i don't understand what you mean there

 

[Bobbler]

Sorry i don't understand what you mean there

I think he means: Just because we aren't seeing as much AA use from the edram as the theory would suggest possible, doesn't mean that there is something actually wrong with the theory (or hardware) -- it's more likely that it's still early and there aren't a lot of engines from scratch made to deal with tiling effectively (and because of that likely have a higher penalty than normal when it gets used). Even considering that most games still have 2x AA, and I would imagine 4x will become prevalent in the coming years.

 

[Proelite]

Not that i disagree, but isn't it after all a little bit more black and white as one migh suggest? I mean Xenos has several advantages over the RSX, features which the PS3 GPU does not have like the EDRAM and the Unified Shaders, but this is not only positive, and this should be enlightened as well.

RSX has no hardware support to accelerate MSAA like Xenos has with its EDRAM, but is it worse because of this?

Looking at the current Xbox 360 games, i can only come to one conclusion - no.
Still there are hardly any games using MSAA at all, possibly because the cost of higher geometry / Z-only pass when tiling is too high for "real" games in contrast to theory?

On the other hand there are already games confirmed for PS3 that have 4xAA from the beginning and it doesnt seem to be much of a problem at all, despite the PS3 has no such accelerator, maybe because of the work balancing between CELL-RSX?

So it comes to my mind that lots of the theoretical features of Xenos are neglectible when it comes to real games, because of the limitations coming along with these feauters.

I might be wrong though

You are. A lot of games on Xbox 360 are taking advantage of the EDRAM, such as GRAW, Oblivion, Motogp, Fight Night, DOA4, and many others. Most future releases such as Madden 07, NCAA 07, Dead Rising, Halo 3, and Lost Planet have AA. The ones without AA are either launch titles, or games with engines that doesn't support AA, such as UE3. The EDRAM is clearly been put to good use by the developers. Contrarily, only Warhawk and Resistance currently have AA on the Playstation 3. So as of yet, we don't know to what extent will PS3 games will feature AA.

 

[Jesus2006]

You are. A lot of games on Xbox 360 are taking advantage of the EDRAM, such as GRAW, Oblivion, Motogp, Fight Night, DOA4, and many others. Most future releases such as Madden 07, NCAA 07, Dead Rising, Halo 3, and Lost Planet have AA. The ones without AA are either launch titles, or games with engines that doesn't support AA, such as UE3. The EDRAM is clearly been put to good use by the developers. Contrarily, only Warhawk and Resistance currently have AA on the Playstation 3. So as of yet, we don't know to what extent will PS3 games will feature AA.

Afaik all PS3 games have (or are confirmed to have) AA , WarHawk is 4xAA (@60fps), HS is 4xAA, Resistance i don't know but i guess at least 2xAA.

MotoGP does not have AA i think at least not the demo which i downloaded. The only 360 game i know to have 4xAA is DoA but this is "just" a fighting game with little geometry and not much on the scene.

 

[scooby_dooby]

Sorry i don't understand what you mean there

Meaning when you have a new architecture that developers are not used to. You have to give them time to exploit that architecture before you can judge it's effectiveness. 8 months in is not nearly enough time to proclaim: " the theoretical features of Xenos are neglectible when it comes to real games"

Afaik all PS3 games have (or are confirmed to have) AA

Well this sort of speaks for itself. You're pretty misinformed.

 

[ShootMyMonkey]

Where RSX has 22 GB/s for textures, models and framebuffer, Xenos has some 21 GB/s from RAM + 256 GB/s for the same purposes. Xenos will never be bandwidth starved for framebuffer effects, just like PS2.

Well, the first part of the sentence neglects that RSX can use the other memory pool for textures, but framebuffer access is still pretty much the suck. The end of that sentence is totally false. It's actually quite easy to bandwidth-starve Xenos on framebuffer effects because the 256 GB/sec is localized to the eDRAM die itself. data moved from the functional part of the GPU to the eDRAM is only going to move at 32 GB/sec, which is still a lot better than RSX has, but not a level that can't be totally swallowed up (and yeah, there are simple test cases for this) given that it's only 2/3 of what the PS2 has while you're working at a much higher resolution, potentially with MSAA and so on. Xenos would be a dog if it were given the task of all the framebuffer blends and accumulations which are practically a given for a PS2 renderer.

Well, there are still plenty of bottlenecks associated with Xenos and MSAA and blending, but they're largely software problems rather than hardware ones (sort of)...

Basically, the main point is that Xenos is a little more generic in what its good at than RSX.