http://www.dailytech.com/article.aspx?newsid=3910
Not Einstein and not really BOINC yet, but it seems that Folding@home is able to put the Cell to good use, reaching over 100 GFlops per PS3. Very nice!
http://www.dailytech.com/article.aspx?newsid=3910
Not Einstein and not really BOINC yet, but it seems that Folding@home is able to put the Cell to good use, reaching over 100 GFlops per PS3. Very nice!
MrS
YES ! http://folding.stanford.edu/news.html
big parts of the stanfordapplication is directly from sony !
also they will use normal (coming soon) pci-e cards based on ati-chips with the same power like sony ps-3 and higher.
A game console is sold with bundled games, without a keyboard and other necessary HW. Then you need an OS (Linux?) able to run on that hardware, a client such as BOINC and an application compiled on that HW.IMHO it is unlikely that Stanford U will ever use a software developed at Berkeley U.
Tullio
[IMHO it is unlikely that Stanford U will ever use a software developed at Berkeley U.
We'll see about PS3 and graphics cards crunching becoming mainstream - it's definitely harder to do than normal x86 CPU-crunching, for the project staff as well as the participants.
Concerning the quote, take a look here, in the section "FAH ON BOINC": http://folding.stanford.edu/FAQ-highperformance.html
They may not be trying very hard yet, but they're trying ;)
[IMHO it is unlikely that Stanford U will ever use a software developed at Berkeley U.
We'll see about PS3 and graphics cards crunching becoming mainstream - it's definitely harder to do than normal x86 CPU-crunching, for the project staff as well as the participants.
Concerning the quote, take a look here, in the section "FAH ON BOINC": http://folding.stanford.edu/FAQ-highperformance.html
They may not be trying very hard yet, but they're trying ;)
MrS
See also here: CELL
Tullio
Edit: after visiting your link, I agree they are not trying very hard. Let's wait and see.
This last weekend Sony released a folding @ Home app for the PS3. These machines can crunch !!!!!
There are about 35000 PS3's signed up to folding at home and those 35000 machines are now doing more work than all computers on all BOINC projects!!!!!!!!!!!!!! Home run for Folding @ Home.
Bare in mind that this PS3 delivered work is only half math precision, and FAH apparently being the only Project to accept that (and only from its PS3 client, I assume their PC Clients are still doing full precision math for a good reason).
That kinda puts it into perspective IMHO.
[Warning type="regurgitated vendor speil"]
My employer's had engineers from Mercury Computing in flogging thier cell based workstations and commercial crunch boxes. They claim that most computationally intensive apps have critical sections small enough that the single 64bit FPU can be multiplexed to support all seven 32bit FPUs without being a bottleneck. They're willing to sell both APIs to abstract away the complexity without significant performance hits (and alledged futureproofing with successor systems regardless of hte architechture). They're also willing to sell thier engineers time to help you port your apps if you PHB doesn't think you have the skills needed to do a proper job yourself.
[/Warning]
Here's some news from "boinc projects" mail list, that some of you might find of interest
Quote:
David Anderson to Boinc
show details 12:47 pm (5 hours ago)
A group in Barcelona has ported the BOINC client
and an app (molecular dynamics) to the PS3 running Linux.
We'll offer the client for download on the BOINC web site,
and will add a Wiki page with info about porting apps to PS3.
Projects can consider porting their app to the PS3
(warning: this is harder than just getting it to compile).
Of course, there are only a few PS3s running Linux.
FYI, I have been negotiating with Sony for 8 months
about porting BOINC to the standard PS3
(which runs a proprietary OS, not Linux)
and having it included in the default software,
like they're currently doing for Folding@home.
These negotiations stalled, at least for the time being.
Maybe some activity in the PS3/Linux/BOINC area will
rekindle their interest.
-- David
I just checked the master download page, and don't see it there yet.
And at the Boinc main page, I see the following:
Quote:
News
May 31, 2007
Sony Playstation 3 owners: researchers at the Barcelona Biomedical Research Park have launched PS3GRID, a BOINC-based project whose application (molecular dynamics simulations) runs on the PS3's Cell processor, on Linux.
Just to let you guys know that another BOINC project is testing an initial PS3 optimized application. If you look at some of the machine's results, you can see that the speed of the app rivals the stock application running on a Pentium 4 clocked in the 3GHz+ range. This is great because the first app tested on the PS3 was the standard Linux PPC app, which gave very poor performance with times of a PII/P3. This proves that the speed will vary wildly depending on how well tuned the application is to the PS3.
http://www.dailytech.com/arti
)
http://www.dailytech.com/article.aspx?newsid=3910
Not Einstein and not really BOINC yet, but it seems that Folding@home is able to put the Cell to good use, reaching over 100 GFlops per PS3. Very nice!
MrS
Scanning for our furry friends since Jan 2002
RE: http://www.dailytech.co
)
YES ! http://folding.stanford.edu/news.html
big parts of the stanfordapplication is directly from sony !
also they will use normal (coming soon) pci-e cards based on ati-chips with the same power like sony ps-3 and higher.
i think many projects follow this great way.
RE: i think many projects
)
A game console is sold with bundled games, without a keyboard and other necessary HW. Then you need an OS (Linux?) able to run on that hardware, a client such as BOINC and an application compiled on that HW.IMHO it is unlikely that Stanford U will ever use a software developed at Berkeley U.
Tullio
RE: [IMHO it is unlikely
)
We'll see about PS3 and graphics cards crunching becoming mainstream - it's definitely harder to do than normal x86 CPU-crunching, for the project staff as well as the participants.
Concerning the quote, take a look here, in the section "FAH ON BOINC":
http://folding.stanford.edu/FAQ-highperformance.html
They may not be trying very hard yet, but they're trying ;)
MrS
Scanning for our furry friends since Jan 2002
RE: RE: [IMHO it is
)
See also here:
CELL
Tullio
Edit: after visiting your link, I agree they are not trying very hard. Let's wait and see.
This last weekend Sony
)
This last weekend Sony released a folding @ Home app for the PS3. These machines can crunch !!!!!
There are about 35000 PS3's signed up to folding at home and those 35000 machines are now doing more work than all computers on all BOINC projects!!!!!!!!!!!!!! Home run for Folding @ Home.
Bare in mind that this PS3
)
Bare in mind that this PS3 delivered work is only half math precision, and FAH apparently being the only Project to accept that (and only from its PS3 client, I assume their PC Clients are still doing full precision math for a good reason).
That kinda puts it into perspective IMHO.
[Warning type="regurgitated
)
[Warning type="regurgitated vendor speil"]
My employer's had engineers from Mercury Computing in flogging thier cell based workstations and commercial crunch boxes. They claim that most computationally intensive apps have critical sections small enough that the single 64bit FPU can be multiplexed to support all seven 32bit FPUs without being a bottleneck. They're willing to sell both APIs to abstract away the complexity without significant performance hits (and alledged futureproofing with successor systems regardless of hte architechture). They're also willing to sell thier engineers time to help you port your apps if you PHB doesn't think you have the skills needed to do a proper job yourself.
[/Warning]
Here's some news from "boinc
)
Here's some news from "boinc projects" mail list, that some of you might find of interest
I just checked the master download page, and don't see it there yet.
And at the Boinc main page, I see the following:
Just to let you guys know
)
Just to let you guys know that another BOINC project is testing an initial PS3 optimized application. If you look at some of the machine's results, you can see that the speed of the app rivals the stock application running on a Pentium 4 clocked in the 3GHz+ range. This is great because the first app tested on the PS3 was the standard Linux PPC app, which gave very poor performance with times of a PII/P3. This proves that the speed will vary wildly depending on how well tuned the application is to the PS3.