Key insight into PS3 design

[cbarcus]

While browsing a recent thread, I chanced upon a couple of posts that really gets to the heart of the matter concerning the design of the PS3. For those who may have missed these posts buried deep in "PS3 is unrefined," I've dug them up for you:

Entropy's post:
http://www.beyond3d.com/forum/showpost.php?p=662246&postcount=634

and followup:
http://www.beyond3d.com/forum/showpost.php?p=662890&postcount=649

 

[nondescript]

Very good posts, thanks for linking them.

I believe the key difference is that Xbox360 developers are predominantly PC developers, so Microsoft chose a hardware architecture that would be easily accessible to them.

By contrast, the asymmetrical nature of the the PS3, and the large data bandwidth available between each element is reminiscent of the EE's large bandwidth to RDRAM, and the bandwidth available to the MIPS, VU0, VU1, and IOP. For programmers who have already learned the tackle the PS2, perhaps the PS3 might actually be easier to develop in some respects.

Because the Xbox360 and PS3 have different developer pools (some overlap, but limited), perhaps the architectures were chosen to fit their competencies.

 

[Edge]

From that Forbes article:

"The good news: Some designers say creating games for Cell is far less complicated than writing for PlayStation 2. "Anyone who worked on the PlayStation 2 is jumping for joy," says Jeremy Gordon, chief executive of Secret Level, a gamemaker in San Francisco that is remaking a classic 1980s Sega videogame for the new Sony box."

 

[trinibwoy]

Thanks Entropy (and cbarcus for the links). It's so cool reading something from somebody who knows what they're talking about.

It does look like Cell is a much more accessible and streamlined approach than the EE and with Nvidia's help, programming RSX should be a breeze if the APi is OpenGL based. It will certainly be very interesting to see what devs are doing with Cell at the end of the PS3's life-cycle.

 

[expletive]

From that Forbes article:

I think developers need to clarify this somewhat. I mean they could use just the PPE and sure, with the nvidia chip that should be straightforward. What i would like to know is how easy or difficult it is to use the Cell in the way Sony is claiming it will be, which is, using multiple SPEs constantly to have a somputational edge...

 

[scooby_dooby]

Of course it's easier than PS2. PS3 has a conventional PC-based GPU instead of using the EE+GS, how could that not be easier?

 

[scificube]

Well...Sony is also trying much harder to make a good SDK for developers all around and from the sound of things the state of affairs is much better than what they did with the PS2.

 

[liverkick]

Of course it's easier than PS2. PS3 has a conventional PC-based GPU instead of using the EE+GS, how could that not be easier?

Ill give the benefit of the doubt that he was talking about the entire system including Cell, because it was an article based on Cell afterall.

And what scifi said, SDK seems to be night and day compared to PS2.

Some of you guys try way too hard.

 

[3dcgi]

Very good posts, thanks for linking them.

I believe the key difference is that Xbox360 developers are predominantly PC developers, so Microsoft chose a hardware architecture that would be easily accessible to them.

By contrast, the asymmetrical nature of the the PS3, and the large data bandwidth available between each element is reminiscent of the EE's large bandwidth to RDRAM, and the bandwidth available to the MIPS, VU0, VU1, and IOP. For programmers who have already learned the tackle the PS2, perhaps the PS3 might actually be easier to develop in some respects.

Because the Xbox360 and PS3 have different developer pools (some overlap, but limited), perhaps the architectures were chosen to fit their competencies.

Sony was only able to create/use Cell because they started working on PS3 very early. Microsoft didn't have as much time and had no option to attempt to create a revolutionary design. I don't think the developer pools had much to do with it.

 

[patsu]

Not quite fair to highlight only the "time" element. It's mostly about will and execution. No one told the Playstation team to start early. I think Sony, Toshiba and IBM also:

* Set a lofty performance goal up front and try very hard to enforce and achieve it despite very different company and ethnic culture

* Dream up and stick to their technical vision. If IBM were to do it alone, they would have chosen a similar approach to Xenon (says some IBM guy). Input from Sony and Toshiba helped to shape Cell the way it is today. I think a lot of tough, no-compromise (as opposed to "balanced") choices went into the design.

* Somehow have faith in developers to exploit the theoretical power in Cell (These guys are optimistic :|). The CPU is just one element, Sony also invested in software projects and standards to complete the picture. Granted they probably still need to catch up in this area.

* Invested in expensive plants as planned despite corporate-wide losses and board room shuffle

* Attempt to establish Cell as a platform. Push for strategic cell adoption early. Kutaragi met up with Jobs to anchor Cell adoption, Toshiba will produce TVs based on Cell afterall, IBM just open sourced Cell simulator and SDK, etc.

Beyond the CPU...

* Engage in next-gen format war, and still have the appetite to want HDMI 2.0, DLNA, ... Are they mad ? With so many different pieces converge into 1 box, something's gotta give.

(Rant: And the forum people still want a multi-core, next-gen GPU to go with PS3)

Looking at the amount of sh*t they went through, I would like to see these guys succeed.