profiling

Guangxuan-Xiao · Jun 21, 2021 · 86e40cd · 86e40cd
1 parent 3ce6344
commit 86e40cd
Show file tree

Hide file tree

Showing 3 changed files with 4 additions and 2 deletions.
diff --git a/prof/arxiv-256.nvprof b/prof/arxiv-256.nvprof
diff --git a/script/prof.sh b/script/prof.sh
@@ -1,3 +1,3 @@
 source /home/spack/spack/share/spack/setup-env.sh
 spack load cuda
-srun nvprof -o ../prof/${1}-${2}.nvprof ~/PA4_build/test/unit_tests --dataset $1 --datadir ~/PA4/data/ --len $2
+srun nvprof -f -o ../prof/${1}-${2}.nvprof ~/PA4_build/test/unit_tests --dataset $1 --datadir ~/PA4/data/ --len $2
diff --git a/src/spmm_opt.cu b/src/spmm_opt.cu
@@ -1,7 +1,7 @@
 #include "spmm_opt.h"
 #include <stdio.h>
 #include <cuda.h>
-
+#include <cuda_profiler_api.h>
 const int BLOCK_X = 4;
 const int BLOCK_Y = 32;
 const int NUM_THREADS = BLOCK_X * BLOCK_Y;
@@ -95,5 +95,7 @@ void SpMMOpt::run(float *vin, float *vout)
     // printf("num_v = %d, feat_in = %d\n", num_v, feat_in);
     // printf("Grid = <%d, %d, %d>\n", grid.x, grid.y, grid.z);
     // printf("Block = <%d, %d, %d>\n", block.x, block.y, block.z);
+    cudaProfilerStart();
     spmm_kernel_merge<<<grid, block>>>(d_ptr, d_idx, d_val, vin, vout, num_v, feat_in);
+    cudaProfilerStop();
 }