- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I have a program a.out build with MKL. I exec different command:
1.
export MKL_ENABLE_INSTRUCTIONS=AVX2;export MKL_NUM_THREADS=1;export MKL_DYNAMIC=TRUE;export OMP_NUM_THREADS=1;amplxe-cl -c hpc-performance ./a.out
The FPU Utilization is 0.0%
2.
export MKL_ENABLE_INSTRUCTIONS=AVX2;export MKL_NUM_THREADS=2;export MKL_DYNAMIC=TRUE;export OMP_NUM_THREADS=2;amplxe-cl -c hpc-performance ./a.out
The FPU Utilization is 0.6%
FP Instruction Mix
% of Packed FP Instr.: 100.0%
% of 256-bit: 100.0%
3. export MKL_ENABLE_INSTRUCTIONS=AVX2;export MKL_NUM_THREADS=2;export MKL_DYNAMIC=TRUE;export OMP_NUM_THREADS=2;amplxe-cl -c hpc-performance taskset -c 1,2 ./a.out
The FPU Utilization is 0.0%
4. export MKL_ENABLE_INSTRUCTIONS=AVX2;export MKL_NUM_THREADS=2;export MKL_DYNAMIC=TRUE;export OMP_NUM_THREADS=2;amplxe-cl -c hpc-performance taskset -c 1,2,3,4,5,6 ./a.out
The FPU Utilization is 1.3%
FP Instruction Mix
% of Packed FP Instr.: 85.7%
% of 256-bit: 85.7%
% of Scalar FP Instr.: 14.3%
I wonder the behavior of MKL?
- Tags:
- Intel® Advanced Vector Extensions (Intel® AVX)
- Intel® Streaming SIMD Extensions
- Parallel Computing
Link Copied
0 Replies
Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page