is there anyway to avoid this serialization behavior in cuda profiling?

StackOverflow https://stackoverflow.com/questions/14470609

  •  17-01-2022
  •  | 
  •  

سؤال

According to CUDA streams not overlapping , "the profiler will serialize streaming to get accurate timing data". Now the question is, is there anyway to avoid this serialization behavior in cuda profiling (say nvvp)? I am using Fermin M2090 and cuda-4.0.

هل كانت مفيدة؟

المحلول

The Visual Profiler 5.0 (including nvprof and CUPTI) and Nsight Visual Studio Edition 2.0 and greater (>2 years old) support concurrent kernel trace for Fermi and Kepler devices.

مرخصة بموجب: CC-BY-SA مع الإسناد
لا تنتمي إلى StackOverflow
scroll top