Questions and Answers :
GPU applications :
libfftw library redundant?
Message board moderation
Author | Message |
---|---|
S@NL - John van Gorsel Send message Joined: 5 Jul 99 Posts: 193 Credit: 139,673,078 RAC: 0 |
When adding Cuda to app_info.xml (running optimized applications) you need to declare a number of files, including libfftw3f-3-1-1a_upx.dll. Since there is already a Cuda specific FFT library (cufft.dll) I am wondering whether this file is still required. I haven't tried it myself, but someone told me he is successfully running Cuda without this file. Seti@Netherlands website |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14654 Credit: 200,643,578 RAC: 874 |
libfftw3f-3-1-1a_upx.dll is required if you are running the stock (Berkeley) CUDA application, because that version has the ability to fall back to CPU processing if it comes across a problem with the CUDA card. It's possible that some of Raistmer's applications no longer have this dependency, though I don't think he's done a combination version yet which uses full optimisation (and hence not doesn't need libfftw3f-3-1-1a_upx.dll) when CPU fallback is invoked. Probably best to keep the reference in app_info.xml unless you're absolutely certain that the version you're using doesn't need it: it doesn't add a huge amount extra to the complexity. |
S@NL - John van Gorsel Send message Joined: 5 Jul 99 Posts: 193 Credit: 139,673,078 RAC: 0 |
Thanks Richard! I'm working on a detailed instruction (in Dutch) on how to install and run optimized applications. My experience is that it's best to check all details to prepare for questions :-) |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.