Einstein: New Einstein@Home Radio Pulsar Search and NVIDIA GPU Code

[RSS]

Einstein@Home is beginning a new round of searching for radio pulsars in short-orbital-period binary systems. http://einstein.phys.uwm.edu/img/NV_DesignedFor_CUDA_H_2D.png (http://www.nvidia.com/cuda) This is accompanied by the release of a new application (called BRP3). The new application is particularly efficient on NVIDIA Graphics Processor Cards (http://www.nvidia.com/cuda) (up to a factor of 20 faster than the CPU-only application). In addition, when running on an NVIDIA GPU card, this new application makes very little use of the CPU (typically around 20% CPU use when the GPU is devoted to Einstein@Home). The NVIDIA GPU application is initially available for Windows and Linux only. We hope to have a Macintosh version available soon. Due to limitations in the NVIDIA drivers, the Linux version still makes heavy use of the CPU. This will be fixed in Spring 2011, when a new version of the NVIDIA Driver is released. Many thanks to NVIDIA technical support for their assistance! Because we have exhausted the backlog of data from Arecibo Observatory, this new application is being shipped with data from the Parkes Multibeam Pulsar Survey (from the Parkes Radio Telescope in Australia). In the next weeks we expect to also start using this new application on fresh Arecibo data taken with the latest 'Mock Spectrometer' back-end. Questions, problems or bug reports related to this new application and search should be reported in this news item thread as a 'Comment'. Bruce Allen Director, Einstein@Home

More... (http://einstein.phys.uwm.edu/forum_thread.php?id=8690)

[zombie67]

Awesome! It looks like they finally have a usable CUDA app.

[Teratoma]

Anyone know what cards will work? Is there a min compute capability requirement? I see driver version of 260 is needed but not much else.

[Fire$torm]

Anyone know what cards will work? Is there a min compute capability requirement? I see driver version of 260 is needed but not much else.

If the app is Single Precision than any nVidia card with Compute Capability of 1.1 or better will work. For desktop GPUs, that means most of the G-Force 8xxx and up and Quadro FX 370 and up. The mobile series starts at the 8200M. If the app requires double precision than compute capability of 1.3 or higher is required.

Edit: The list of BOINC capable cards (https://spreadsheets.google.com/ccc?key=0AqCMiOAyW0eBdHdWRDlkZ2xHeUxDQ0E1TVQtVkI3c kE&hl=en&authkey=COmlndUK#gid=0)

[DrPop]

Awesome! It looks like they finally have a usable CUDA app.

Which means that in theory, they could port it to OpenCL and give us an ATI app like PG did. Wouldn't be as efficient as writing an ATI app from scratch, but better than using the CPU.

[zombie67]

Takes 3000-3200 seconds per task on my GTX 580. However, GPU load is only about 50%. No idea about credits yet. All my returned tasks are still pending, waiting for wingmen.

[Beerdrinker]

Takes 3000-3200 seconds per task on my GTX 580. However, GPU load is only about 50%. No idea about credits yet. All my returned tasks are still pending, waiting for wingmen.

Keep us posted please....

Could be some usefull info!!

[zombie67]

Takes 3000-3200 seconds per task on my GTX 580. However, GPU load is only about 50%. No idea about credits yet. All my returned tasks are still pending, waiting for wingmen.

It looks like they pay a fixed 500 credits. So 500 / (3100/3600) = 580 c/h