Seti for the PS3 and for Windows X64

[Sir Ulli]

first the PS3 Version

http://setiathome.berkeley.edu/forum_thread.php?id=39910

it is Alpha testing at this time...

and than the Windows X86-64 Version

http://boinc.berkeley.edu/download_all.php

Greetings from Germany
Sir Ulli

 

[theheyes]

Interesting.

Some initial thoughts in regards to the ps3 version: the ps3 needs to be running linux, so its not just a matter of leaving the box to idle to start working. This means the casual cruncher isn't going to be too keen on booting into linux to start the process.

"So what?" you say. Rebooting is trivial to someone who is enthusiastic about distributed computing, right?

The poster linked to says the Cell doesnt compare well with the C2Ds in terms of performance. Anyone who is "that" serious about it isn't going to bother with something that inefficient. Unless it seriously improves I cant see it catching on.

That said, it is early software and the guy does state hes looking for people to test. Who knows, it may turn into a monster when the software gets optimized. Kudos to the guy for even getting this far programming for the Cell.

Just my 2 pennies.

 

[BillytheImpaler]

The PS3 client runs on Linux. That's not really a good thing. PS3 Linux runs on top of Sony's hypervisor. This hypervisor locks the OS out of the GPU and the Cell's Synergistic Processing Units, the source of its immense floating point power. The Folding@Home client is authorized by Sony and has direct access to the hardware (as really all software should, don't get me started on Sony's anti-consumer and anti-customer practices). This is why FAH on PS3 performs calculations orders of magnitude faster than on a C2D. Since this client cannot do that it as is limited to the single PPC core.

Expect performance to be a wee bit faster than a fast single core PowerMac G5.

The author says he has gotten the client to use the SPUs but I don't see how that's possible.

 

[rich99million]

It's great to see Borkley trying to get themselves into the PS3's powahhh - just a shame that Sony don't seem to be as welcoming to them as they were with Stanford.

Maybe the open-source BOINC/Seti side of things is a bit too risky for them compared to the nailed down tightly F@H clients

Also I really don't see how running anything in Linux on the PS3 is going to give results worthy of such a powerful machine - I guess every little helps but an order of magnitude faster would be nice

 

[oceaness]

So you'd be better off running folding cause you'd get 10x the work done in half the time

 

[Pumpkinstew]

There are vague rumblings of PS3 and X-Box 360 clients on a lot of the Boinc projects, almost everything is bedroom tinkering though.

Needs Sony or MS involvement to properly optimise everything.

You never know though, Rosetta have been getting pretty cosy with the Gates Foundation recently.