Einstein support OpenCL ?

Elphidieus
Elphidieus
Joined: 20 Feb 05
Posts: 245
Credit: 20,603,702
RAC: 0

Regardless CUDA or OpenCL...

Regardless CUDA or OpenCL... Will there be anything on the Mac side...? I'm sick all these developments meant for Windows and some for Linux, but none for Mac... My Pro's dying for more.

Jord
Joined: 26 Jan 05
Posts: 2,952
Credit: 5,878,109
RAC: 7,006

Not unless Nvidia releases a

Message 90693 in response to message 90692

Not unless Nvidia releases a 64bit driver for the Macintosh. CUDA capability is built in in the BOINC 6.4 and 6.6 clients on the Mac, so as soon as a CUDA capable driver is released by Nvidia detections will light up. And then you have to press the projects for apps for the Mac to use this new found territory.

Elphidieus
Elphidieus
Joined: 20 Feb 05
Posts: 245
Credit: 20,603,702
RAC: 0

RE: Not unless Nvidia

Message 90694 in response to message 90693

Quote:
Not unless Nvidia releases a 64bit driver for the Macintosh. CUDA capability is built in in the BOINC 6.4 and 6.6 clients on the Mac, so as soon as a CUDA capable driver is released by Nvidia detections will light up. And then you have to press the projects for apps for the Mac to use this new found territory.

Just curious... Even if nVidia releases the 64-bit driver for Mac, are we still gonna be limited by the fact that Einstein's gonna run either CPU or GPU clients (just like SETI), but not both...?

Richard Haselgrove
Richard Haselgrove
Joined: 10 Dec 05
Posts: 2,143
Credit: 2,923,376,221
RAC: 933,007

RE: Just curious... Even if

Message 90695 in response to message 90694

Quote:
Just curious... Even if nVidia releases the 64-bit driver for Mac, are we still gonna be limited by the fact that Einstein's gonna run either CPU or GPU clients (just like SETI), but not both...?


That's a function of BOINC, not the individual projects. If BOINC supports mixed-mode work allocation, then I see no reason why projects would go to the effort of blocking it again.

And the good news is that BOINC v6.6 does support simultaneous running of both CPU and GPU applications - first for stock applications, and since v6.6.14 for optimised applications as well. The SETI restriction you mention has finally been resolved.

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4,305
Credit: 248,801,925
RAC: 33,277

RE: Regardless CUDA or

Message 90696 in response to message 90692

Quote:
Regardless CUDA or OpenCL... Will there be anything on the Mac side...? I'm sick all these developments meant for Windows and some for Linux, but none for Mac... My Pro's dying for more.


Since the developer previews of Snow Leopard are the only useful OpenCL development environment right now I would expect at least the first OpenCL Einstein@hone App to be issued for MacOS ;-)

BM

BM

Jord
Joined: 26 Jan 05
Posts: 2,952
Credit: 5,878,109
RAC: 7,006

RE: Since the developer

Message 90697 in response to message 90696

Quote:
Since the developer previews of Snow Leopard are the only useful OpenCL development environment right now I would expect at least the first OpenCL Einstein@home App to be issued for MacOS ;-)


There are CUDA drivers for the Mac available for OSX 10.5.2 and above...

Elphidieus
Elphidieus
Joined: 20 Feb 05
Posts: 245
Credit: 20,603,702
RAC: 0

RE: Since the developer

Message 90698 in response to message 90696

Quote:


Since the developer previews of Snow Leopard are the only useful OpenCL development environment right now I would expect at least the first OpenCL Einstein@hone App to be issued for MacOS ;-)

BM

Which means to say by the time Snow Leopard comes in, say 6 months from now, my tower will most likely be out of serviceable graphics that utilises whatever updated drivers corresponding to their most updated cards available in the market of that time....

Oh boy...! :x

hoarfrost
hoarfrost
Joined: 9 Feb 05
Posts: 207
Credit: 103,053,349
RAC: 9

What type of graphics card

What type of graphics card will be necessary for Einstein@Home GPU-application?

DanNeely
DanNeely
Joined: 4 Sep 05
Posts: 1,364
Credit: 3,562,358,667
RAC: 0

On the nVidia side a minimum

On the nVidia side a minimum of a GF 8000 series card. Anything earlier won't support cuda/openCL. Beyond that: the cuda support in the original 8800 cards is less capable than in newer ones and might not run. There's a >30x performance spread between a GTX285 and the lowest end 8xxx/9xxx cards so some of the slowest cards might not be able to complete a WU in time.

On the ATI side, you need a 2xxx series card minimum, but I'm not aware of the technical details about API versioning/performace thresholds. Supposedly an ATO friendly app is farther down the queue than the nVidia one in any event.

Paul D. Buck
Paul D. Buck
Joined: 17 Jan 05
Posts: 754
Credit: 5,385,205
RAC: 0

If they do OpenCL, then the

If they do OpenCL, then the application with be "friendly" to both types of cards. Though the last comment from the project suggested that the CUDA specific version was nearer to completion than other development work.

BUt, as experience on GPU Grid shows, higher end cards are sometimes the only way to get the work done. That said, the run time of the CURRENT tasks on Einstein takes about 6-20 hours on typical systems ... with a 9800GT card or above I would expect this to drop to about about 30 minutes to an hour on the aforementioned 9800GT...

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.