AnandTech article up (X360 vs PS3)

[scooby_dooby]

There's very few universal constants in the world. Anand being a tool is one of thoses.

I mean...

The most important thing to keep in mind is that the revolution in physics engines and collision detection isn’t going to happen over night. The first games for both consoles will, for all intents and purposes, be single threaded titles.More adventurous developers may even split up execution into two concurrent threads

I'm being harsh, though, the article is not that bad...

You're a little harsh.

He based that on what Microsoft told him, as well as the fact his own two eyes aren't seeing anything more impressive than current single-threaded physics or AI.

Here's what MS told him:

"The majority of developers are doing things no differently than they have been on the PC. A single thread is used for all game code, physics and AI and in some cases, developers have split out physics into a separate thread, but for the most part you can expect all first generation and even some second generation titles to debut as basically single threaded games. The move to two hardware execution threads may in fact only be an attempt to bring performance up to par with what can be done on mid-range or high-end PCs today, since a single thread running on Xenon isn’t going to be very competitive performance wise, especially executing code that is particularly well suited to OoO desktop processors. "

"

 

[Carl B]

He based that on what Microsoft told him, as well as the fact his own two eyes aren't seeing anything more impressive than current single-threaded physics or AI.

Like I said I'm a big Anand fan, but his 'own eyes' aren't actually working with either of the hardware architectures. Instead it seems he's calling up companies and asking for their take on things. It's almost the same as how Hannibal came at his Xenon review from his own heavy-PC background; that just doesn't work in this brave new world of next-gen console architectures as well as it used to.

Devs are obviously going to be making use of multi-threaded code on these consoles, it goes without saying.

 

[scooby_dooby]

"The most important thing to keep in mind is that the revolution in physics engines and collision detection isn’t going to happen over night. The first games for both consoles will, for all intents and purposes, be single threaded titles. More adventurous developers may even split up execution into two concurrent threads, but for the most part don’t expect to see a dramatic change in the quality and reality of the physics simulation of the first titles, especially when compared to titles like Doom 3 and Half Life 2.

However, a change is coming and by the end of next year multi-threaded game engines should be commonplace on both consoles and PCs, which will hopefully lead to much more entertaining experiences. The approach to that change will be different according to the platform; without a doubt, developers will have their work cut out for them. "

The 1st games probably will be mostly single-threaded won't they?

If study's have shown switching to multi-threading to increase dev time 2-3 times over, and the XBOX needs games in les than 5 months, while the PS3 needs games in less than 1 year, isn't it safe to say that the FIRST games for each console WILL be for the most part single threaded designs??

Especially when you consider Tim Sweeney's comments alluding to the limitations of CELL, and the extra work required to optimize code for the SPE's.

I mean, I'm sure they will use multiple threads when and where they can, but I don't think we'll see anything really built multi-threaded from the ground up for a while.

I thought it was a great article, kinda cuts through the CELL hype, and Xenos hype, and brought both consoles back down to earth.

 

[pc999]

The dual core dev kits are tp dev learn who to use it, so why will we see single-thread code, if dual I understando but single :? .

 

[Carl B]

@Scooby:

Maybe maybe not, but I could imagine most PS3 launch titles using the SPE's in at least one form or another; Sony really seems to be pushing the notion and almost all the dev kit stuff we're hearing lately is on how they're trying to make it easier to thread code onto the SPE's. Not saying they'll play a big role at first, but only in cross-platform titles would I expect them not to be used at all.

 

[scooby_dooby]

I don't disagree, I just think Anand's comment were prety bang on.

"For all intents and purposes" he says they will be single threaded, so to me that means they may use multiple threads as possible, but it will still be basically a single-threaded design with a few exceptions when possible.

Then in the 3rd or 4th gen you'll start to see games designed from the ground-up to be multi-threaded, and that's when we'll really see all that power.

What it won't be, is a truly multi-threaded design approach, there's just no time.

I so think to some extent the Dev's will be forced to use multiple threads, since many of the games today are CPU limited, and the single PPE's of both these machines are weak, I'd imagine they would have to multi-thread just to match the CPU performance of current PC's.

I dunno, anyways, I'm not really disagreeing with you.

 

[deathstar121]

BLah, blah, blah, all this techno crap is pointless, this is how I see it games of this generation sucked, I have been a gamer since atari days I loved the 8bit era, 16bit, ps1, saturn, and n64 era, but this generation was boring games weren't even innovative I only liked 6 games from this generation, 6? yeah you heard right 6, I found the rest to be border line average, some had graphics and no gameplay, some had good gameplay but poor execution, but mostly all of them fail to impress me, the games that sold well this gen didn't even impress me, for example halo 2 it sold 6.5 million units but the game was crap.

My point is the same formula that worked for game companies will work for them again, only devs that will use PS3 to the max are the sony only devs, but beyond that all multi-platform games will follow the same formula that they follow now, with all this power of the next gen system there's alot of room for innovation but it won't happen because companies need to make money not sit there and waste $30-$40 million a game just to satisfy there fan base.

I see on alot of forums that I can't wait for PS3 or X360 they gonna rock, sure they gonna rock, graphics will always fool even the hardcore of gamers, but after the shock and awe is over, he's gonna realized that they game he bought sucks, I don't see next gen system bringing alot of innovation, but we will see some great AI and graphics, but at the end of the day I will probably be playing the same crap like everyone else just that I won't like it.

PS: not 6 games 7, when zelda is release that will be my 7th.

 

[scooby_dooby]

BLah, blah, blah, all this techno crap is pointless, this is how I see it games of this generation sucked, I have been a gamer since atari days I loved the 8bit era, 16bit, ps1, saturn, and n64 era, but this generation was boring games weren't even innovative I only liked 6 games from this generation, 6? yeah you heard right 6, I found the rest to be border line average, some had graphics and no gameplay, some had good gameplay but poor execution, but mostly all of them fail to impress me, the games that sold well this gen didn't even impress me, for example halo 2 it sold 6.5 million units but the game was crap.

My point is the same formula that worked for game companies will work for them again, only devs that will use PS3 to the max are the sony only devs, but beyond that all multi-platform games will follow the same formula that they follow now, with all this power of the next gen system there's alot of room for innovation but it won't happen because companies need to make money not sit there and waste $30-$40 million a game just to satisfy there fan base.

I see on alot of forums that I can't wait for PS3 or X360 they gonna rock, sure they gonna rock, graphics will always fool even the hardcore of gamers, but after the shock and awe is over, he's gonna realized that they game he bought sucks, I don't see next gen system bringing alot of innovation, but we will see some great AI and graphics, but at the end of the day I will probably be playing the same crap like everyone else just that I won't like it.

PS: not 6 games 7, when zelda is release that will be my 7th.

it's just because you were young so you have fond memories of the older games, simple as that.

Go back an play some Atari games or NES games, they are not nearly as fun or as good as you remember. And 64? I had a 64 and there were VERY VERY few good games for it. Zelda, Mario 64, Starfox........what else? Turok?

Halo 2 was a game with great story(short but good), great action, great surround sound, excellent graphics, and very strong AI(play it on legendary and tell me the AI isn't sick). It didn't "suck", it was an excellent game by any measure.

I have all the ROMS from older games, and as much as I have fond memories, when I go back and play the games really are not that great, most are extremely boring and very repetitive, it's just that as a kid you remember them in a different light.

 

[Carl B]

PS: not 6 games 7, when zelda is release that will be my 7th.

Now - how can you pre-enjoy a game that hasn't even be released? I think if you believe in what you're writing, you should at least wait until the game comes out before you add it to your list of games you've 'enjoyed' this gen.

 

[Titanio]

Especially when you consider Tim Sweeney's comments alluding to the limitations of CELL, and the extra work required to optimize code for the SPE's.

I actually found his comments to be encouraging and actually weighed more favourably toward Sony's approach than anything else. He doesn't give much credit to the time required for certain things and if the SPEs can "accelerate" the things he's talking about alone, particularly physics, I think the design choices made will have been justified.

 

[gurgi]

BLah, blah, blah, all this techno crap is pointless

Then no offense, but you might not enjoy this community very much. Not that gameplay is irrelevant here, but b3d is very much a tech site.

 

[BlueTsunami]

Devs don't have enough time to make FULL GAMES by the time the consoles are released, let alone games that are made to take advantage of the muli-thread setup these CPU's are capable of. I see that more as a give-in then something surprising...

Launch games=Very Pretty, minus the gameplay...

 

[scooby_dooby]

I actually found his comments to be encouraging and actually weighed more favourably toward Sony's approach than anything else. He doesn't give much credit to the time required for certain things and if the SPEs can "accelerate" the things he's talking about alone, particularly physics, I think the design choices made will have been justified.

I found it really pointed out some potential weaknesses in the CELL design.

Firstly, it's only good for certain things(physics, particle systems and sound), but not good for others(game logic, AI, and scripting).

And some of the things it can be good at, like collission detection, need a redesigned approach because of the lack of branch prediction.

Couple that with the fact the single main PPE is fairly weak, and I think that's a big potential weakness there, you have a weak PPE with SPEs's that are quite limited in what they can and can not help with.

He does state that the potential is there for extremely high performance, but raises the very real question of how many Dev's will have the time or skill to use it.

The potential will not be nearly as easy to exploit as the X360.

"Architecting a game engine around Cell and optimizing for SPE acceleration will take more effort than developing for the Xbox 360 or PC, but it can be done. The question then becomes, will developers do it?

The end result is that multi-threaded game development takes between 2 and 3 times longer than conventional single-threaded game development, to add additional time in order to restructure elements of your engine to get better performance on the PS3 isn’t going to make the transition any easier on developers. "

 

[Titanio]

I found it really pointed out some potential weaknesses in the CELL design.

Firstly, it's only good for certain things(physics, particle systems and sound), but not good for others(game logic, AI, and scripting).

And some of the things it can be good at, like collission detection, need a redesigned approach because of the lack of branch prediction.

Couple that with the fact the single main PPE is fairly weak, and I think that's a big potential weakness there, you have a weak PPE with SPEs's that are quite limited in what they can and can not help with.

I think saying "can not" is a bit strong given the creativity of (some?) game developers

But to address the rest, the number of things a chip is good/better at is less relevant than what those things actually are. Don't look at a checklist of tasks and mark them under Xenon or Cell and then count them up, you need to look at the proportion of processing those tasks take. Not all tasks are created equally (take as much time and/or are as important, though the latter is more a subjective point).

 

[ERP]

Stupid observation of the weak time.

Better physics and Ai and all that.....
Do we really see it much on PC's verus consoles?
I mean a 3+GHz Pentium is quite a lot faster than a 733 MHz Celeron, but is the gameplay experience that much better?

I guess my point is all this comparing to PC technology is moot, the single PPE is significantly faster than the PS2 or Xbox processor, so even if the games were completly singlethreaded (and decoupling graphics from logic is relatively simple) you should see a pretty significant jump.

 

[pegisys]

The first games for both consoles will, for all intents and purposes, be single threaded titles

I can see this being true for just about all cross platform games through out the whole life of the consoles, and when I say this I'm thinking of EA mostly, with the systems they can have the same game logic for both consoles run off one core with only changes to support the graphics, and it will look and play the same on both consoles.

If someone like EA took the time to make engines fine tuned for both consoles for all there games, they wouldn't make any money, because they make there money off of volume, only exclusives will take full advantage of the hardware, and I don't see that changing

 

[mckmas8808]

Couple that with the fact the single main PPE is fairly weak, and I think that's a big potential weakness there, you have a weak PPE with SPEs's that are quite limited in what they can and can not help with.

So Scooby what I get from you is that the CELL processor is pretty much crap. I mean you named almost everything that the CELL is and said it's either weak or limited.

So let me ask you what should have Sony went with? Would going with a more MS apporach helped? What if they put in a 3.2 Ghz P4 chip with the 512 memory (kinda like a super beefed up Xbox1) would that have been better?

 

[cobragt]

Did anandtech know the ps2 could do 1080i? Well it did with gt4. Point, anandtech isn't a dev, they dont have the right to say what's going to be easy to develop for and what's not. When they get the ps3's hardware tell me

 

[_phil_]

And some of the things it can be good at, like collission detection, need a redesigned approach because of the lack of branch prediction.

On what authority do you assume that ?
Novodex has the fastest collision detection on the market ,and is multithreaded. mark rein said Novedex told him that basically ,the cell matches their PPU.
You should not try to be technical if you haven't the knowlege ,imo...

 

[Titanio]

Better physics and Ai and all that.....
Do we really see it much on PC's verus consoles?
I mean a 3+GHz Pentium is quite a lot faster than a 733 MHz Celeron, but is the gameplay experience that much better?

I guess my point is all this comparing to PC technology is moot, the single PPE is significantly faster than the PS2 or Xbox processor, so even if the games were completly singlethreaded (and decoupling graphics from logic is relatively simple) you should see a pretty significant jump.

I agree with you, though I guess technical potential and competency (of a chip, cpu) doesn't always map to the realisation of such, particularly on an open platform like the PC. I do think it's not helpful approaching this from a PC mindset though, but people will start with what they're familiar with.