I am trying to get the profiling data for cuFFT library calls for example plan and exec. I am using nvprof (command line profiling tool), with option of "--print-api-trace". It prints the time for all the apis except the cuFFT apis. Is there a any flag i need to change to get the cuFFT profiling data ? Or I need to use the events and measure myself ??
Asked
Active
Viewed 298 times
2 Answers
5
According to the nvprof documentation, api-trace-mode:
API-trace mode shows the timeline of all CUDA runtime and driver API calls
cuFFT is neither the CUDA runtime API nor the CUDA driver API. It is a library of routines for FFT, whose documentation is here.
You can still use either nvprof, the command line profiler, or the visual profiler, to gather data about how cuFFT uses the GPU, of course.

Ashwin Nanjappa
- 76,204
- 83
- 211
- 292

Robert Crovella
- 143,785
- 11
- 213
- 257
-2
Got it working.. Instead of using the nvprof i used the CUDA_PROFILE environment variable.

Sagar Masuti
- 1,271
- 2
- 11
- 30
-
2Specifying CUDA_PROFILE not only fails to resolve your question but it enables a completely different tool. None of the NVIDIA profiling tools support tracing the cuFFT CPU API calls. All of the tools can be enabled to trace the CUDA kernel launches initiated by cuFFT. – Greg Smith Apr 15 '13 at 05:25
-
thanks all for the clarification.. but could u please share where i can get the what CUDA_PROFILE does? and how it does ? – Sagar Masuti Apr 15 '13 at 06:35
-
1(CUDA command line profiler)[http://docs.nvidia.com/cuda/profiler-users-guide/index.html#compute-command-line-profiler-overview]. The CUDA command line profiler is built into the CUDA driver. nvprof, a new command line profiler, was released in CUDA 5.0 and provides significantly more features than the previous CUDA command line profiler. – Greg Smith Apr 15 '13 at 13:22