Blundering Sony Nukes PS3 Supercomputing and At What Cost? | Hacker News Reader

Blundering Sony Nukes PS3 Supercomputing and At What Cost?

hothardware.com

22 pointsMojoKid9 comments

Gamers, of course, are the ones who will pay for Sony's mistake long term. Don't expect the scientific/HPC community to come rushing as saviors when it comes time for Sony to babble about the PS4--and without the help of this august assemblage, maybe we'll get stuck with games featuring cutting-edge 1987 graphics fidelity. Leisure Suit Larry, here we come.

Open articleView on HN

"It would have required a minimum amount of effort to earmark a small run of pre-Slim hardware (or a certain batch of PS3 Slims) for HPC researchers."

Even a small amount of effort is more than most companies are willing to expend on a sales channel that loses money with every sale. Stupid Sony isn't going to lose as much money as last year. Keep this up and they may make a profit. What a blunder!

All sony had/has to do was license out the cell processor and the OtherOS technology, or increase the price on the units and limit the sales channels. Sony could have figured out a way to make money on this.

Of course, I'm surprised somebody else hasn't done the same thing by now.

The article forgot to mention one important point. The option to install another OS on PS3 only disappears if the firmware is updated to the latest version. And to my knowledge, I see no reason why these computing farms should update. They need no PSN access and updates bring no new benefits to the installed OS. New PS3 slim models are already missing that option out of the box.

This is true, but I believe the point isn't that pre-existing PS3 HPC clusters will be hampered, it's that no new ones will be created.

That hardly matters, since the PS3 is already obsolete for HPC. In 2007 it was pretty interesting but today you should be using GPUs.

Aw, man, I'm so conflicted about that.

I made the financial effort to refurbish the "fat" PS3 when it died, even though I'd already bought a Slim, just to keep that system around to learn Cell programming. It's still running the old firmware and, in fact, boots directly into Yellow Dog Linux by default.

Yet, I agree, it feels kind of pointless nowadays, with the GPUs becoming so much better all the time. Just recently the GTX 260 broke the $200 floor at newegg.com and I'll probably buy one pretty soon (maybe when they are around $150) - and those things are pretty powerful. The real motivator is actually to play Tomb Raider Underworld :-) but I'm pretty sure I'll at least try and do some CUDA with it.

So, doing an open source project on the Cell doesn't seem so attractive nowadays. Maybe just for bragging?

I've also heard rumors that there's a successor to the Cell in the works. I'm not sure how it will compare with the GPUs though.

Putting aside what I know about this specific topic, a hypothetical Cell 2 would remain the same for five years and thus might be interesting when it came out but would also be obsolete for most of its lifetime. There's just no way for console chips that are updated every five years to compete with CPUs and GPUs that are updated every year.

Slim PS3's never had OtherOS capability, so new PS3 supercomputers have been hampered for a while before 3.21.

The article makes it sound as if the firmware update is killing off the HPC clusters though. And if it is talking about how the new PS3s don't have OtherOS, I'm not sure why it would be to Sony's advantage to keep adding that in so they could lose hundreds of PS3s to non-game buying researchers.

I think the concern is that the firmware update cut off HPC capabilities from consumers who sought to use it as both a game device and an HPC device.

Reply on news.ycombinator.com