Starbreeze take on the Ps3 vs Xbox 360 (the Darkness Int)

[assen]

If Cell gets only 70% real world performance on a certain task, it is reasonable to assume similar utilization rates on the Xenon.

I define "real world" as "the performance I, or the other real-world programmers on my team, can achieve within a reasonable real world time frame (read: several months), attempting to port our real world codebase for the PS3", not as "the performance a rocket scientist in IBM's basement can achieve fine-tuning a tech demo for the architecture". It's very hard to believe that the "real world" utilization (under the first definition) of Cell would be anywhere near that of Xenon. I can see how the utilization rate according second definition of "real world" will probably be approached by some developers somewhere in 2008-2009, but as for my first encounters with the devkits (which have yet to happen), I fear Cell much more than I fear Xenon.

 

[Tap In]

I.... Can you freaking Microsoft and Sony fanbois just STOP. ......

I know E3 is around the corner and everyone is on their platform specific blow horn, but this is just absurd. The great contributions by ERP, Mintmaster, nAo, etc in the last couple weeks have been completely flooded with a sea of misinformation and rampant fanboism.

agreed and thank you

there was some real informative discussion here prior to all the system posturing

 

[dukmahsik]

I define "real world" as "the performance I, or the other real-world programmers on my team, can achieve within a reasonable real world time frame (read: several months), attempting to port our real world codebase for the PS3", not as "the performance a rocket scientist in IBM's basement can achieve fine-tuning a tech demo for the architecture". It's very hard to believe that the "real world" utilization (under the first definition) of Cell would be anywhere near that of Xenon. I can see how the utilization rate according second definition of "real world" will probably be approached by some developers somewhere in 2008-2009, but as for my first encounters with the devkits (which have yet to happen), I fear Cell much more than I fear Xenon.

very interesting points there assen... it would surely echo the other comments made by devs on cell and xenon.

 

[nonamer]

I define "real world" as "the performance I, or the other real-world programmers on my team, can achieve within a reasonable real world time frame (read: several months), attempting to port our real world codebase for the PS3", not as "the performance a rocket scientist in IBM's basement can achieve fine-tuning a tech demo for the architecture". It's very hard to believe that the "real world" utilization (under the first definition) of Cell would be anywhere near that of Xenon. I can see how the utilization rate according second definition of "real world" will probably be approached by some developers somewhere in 2008-2009, but as for my first encounters with the devkits (which have yet to happen), I fear Cell much more than I fear Xenon.

Which company do you work for and what games you are planning to make? I wish to hear more.

 

[Inane_Dork]

IGN: What is it like to program and design the game for Xbox 360 versus the PlayStation 3? We're not looking to start a blood war here.

Doh.

 

[AlNom]

No offence, but I'm going to insult the living daylights out of you, so don't take offence....

 

[rabidrabbit]

Is it so that the real advantage of Cell is that it scales well with applications, that it really is more of a multitasking monster than a monster in running just one application? (edit: well, and an advantage in physics processing too, but I won't be touching that here, now)

I mean for example an OS can be run in just one SP..whatever of seven available without affecting that much the work of the other SP...thingies.
The CPU of xbox360 will use the resources one of the three cores for OS. Is it that running multiple tasks simultaneously on xbox360 CPU is much less effective than on Cell?

Please note the ?'s, which means I'm asking questions, not stating "facts" or "general assumptions".

 

[assen]

Which company do you work for and what games you are planning to make? I wish to hear more.

We're a typical PC developer, only started serious development with an eye for consoles in the last several months. "Real" console developers would probably have an easier time with PS3.

I can't disclose info on any games in development.

 

[Phil]

I define "real world" as "the performance I, or the other real-world programmers on my team, can achieve within a reasonable real world time frame (read: several months), attempting to port our real world codebase for the PS3", not as "the performance a rocket scientist in IBM's basement can achieve fine-tuning a tech demo for the architecture". It's very hard to believe that the "real world" utilization (under the first definition) of Cell would be anywhere near that of Xenon. I can see how the utilization rate according second definition of "real world" will probably be approached by some developers somewhere in 2008-2009, but as for my first encounters with the devkits (which have yet to happen), I fear Cell much more than I fear Xenon.

All valid points there assen. But using your above described definition of "real world performance" (which btw; I agree with), we also come to the conclusion that it's a very subjective and relative term. What you define as "real world performance" is relative to the time/money/effort/skill you invest into your program and what *YOU* as a team are able to extract. This may or may not be in line in what other developers achieve with it - which is precisely the problem when comparing "real world performance" --> Everyone has there own definition of it, and that is based on various factors (time/money/effort/skill).

Are we going to compare platforms based on what some EA developer (who's primary concern is to be able to make their title work reasonably well on all platforms they support) or are we going to go to the extend to speculate on what a platform exclusive developer is going to achieve on it? The results will be quite different, not unlike the differences we are wittnessing this generation in titles. At least with potential specs, you are given a rough estimate on what the system can handle given certain conditions.

Anyway, I find these platform comparasions pathetic anyway. At the end of the day, each platform will have its strengths and it's going to be solely up to the developers to find out how much they can extract from it. Lets just treat them as potential advantages and all be happy, shall we?

 

[PiNkY]

Without derailing the thread, AnandTech's look at the new AGEIA PhysX PPU might have some relevance in this discussion

link

 

[Farid]

Without derailing the thread, AnandTech's look at the new AGEIA PhysX PPU might have some relevance in this discussion

link

That hardly has anything to do with the comment made by the Starbreeze developer.

Not that the actual discussion occurring in this thread is of a great quality or anything, though.

 

[pc999]

Sugestion:

Put in the forum rules a few links that people must read before post so we never see again things like the 115Gflops from XeCPU and etc...

 

[London Geezer]

Sugestion:

Put in the forum rules a few links that people must read before post so we never see again things like the 115Gflops from XeCPU and etc...

Mmm that would require to actually click on the link and read through the articles before posting. That would have a rate of success of -19.

 

[Shifty Geezer]

Given the number of FAQ rules/guidelines not adhered to, I don't think any one of these...ill judged posters bothers to read them anyway.

 

[rabidrabbit]

Can't the basic forum rules be put on the "registration page" fhere the new members give their username password and such, or is it a forum software thing.

Then the rules would at least be there for everyone that is going to register for posting.
Better still, if there was a questionnaire of some rules before you could click "Submit"

Like (from the B3D Console Forum User Manual [/i]http://www.beyond3d.com/forum/showthread.php?t=22121[/url]
Q: What are opinions like:
A) John Reynold's hairdo
B) *ssholes

That would force the new members to read at least some of it.

The new posters might not know the fact that the User Manual is really quite entertaining read (I didn't until I just checked what's in there )

 

[London Geezer]

The new posters might not know the fact that the User Manual is really quite entertaining read (I didn't until I just checked what's in there )

Totally, Vysez did a tremendous job in turning a usally boring manual into a very funny thread

 

[pc999]

Mmm that would require to actually click on the link and read through the articles before posting. That would have a rate of success of -19.

That happens with rules too, but at least we could ban them right at the begining if they didnt present previos knowledge and/or will to learn (a post like "please explain me the rules links" would be valid) ence this forum wouldnt apreciate their posts.

 

[Guilty Bystander]

Weren't there people on this forum saying the Xenon only had a theoretical max of 77-80GFLOPS instead of 115,2GFLOPS?

Even if the Xenon would manage to achieve 70GFLOPS in realtime it's almost half that of the Cell's realtime performance.

On a sidenote I was always under the assumption the Ageia PPU could do about 60-70GFLOPS in realtime while it can only do 50GFLOPS theoretical.
Shouldn't both the Xenon and especially the manage to mimic the Ageia PPU no problems with Cell using about 2-3 SPE's?

 

[London Geezer]

Good lord will you give it up?

Your one-man war fought with theoretical FLOP figures got old so long ago, now it's really pathetic to keep seeing your silly posts on every thread.

 

[pc999]

On a sidenote I was always under the assumption the Ageia PPU could do about 60-70GFLOPS in realtime while it can only do 50GFLOPS theoretical.

Meybe we need even more rules...