AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Match Occupancy APIs syntax with CUDA (#1625)

Browse Source 
* Match Occupancy APIs syntax with CUDA and fix tests using these APIs
This commit is contained in:
satyanveshd
2020-01-30 02:35:53 +05:30
committed by GitHub
parent c7280bec4d
commit acb2ea8618
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
samples/2_Cookbook/13_occupancy/occupancy.cpp
+2 -2
View File
@@ -86,8 +86,8 @@ void launchKernel(float* C, float* A, float* B, bool manual){
printf("kernel Execution time = %6.3fms\n", eventMs);
//Calculate Occupancy
uint32_t numBlock = 0;
HIP_CHECK(hipOccupancyMaxActiveBlocksPerMultiprocessor(&numBlock, multiply, blockSize, 0));
int numBlock = 0;
HIP_CHECK(hipOccupancyMaxActiveBlocksPerMultiprocessor(&numBlock, multiply,(int)blockSize, 0));
if(devProp.maxThreadsPerMultiProcessor){
std::cout << "Theoretical Occupancy is " << (double)numBlock* blockSize/devProp.maxThreadsPerMultiProcessor * 100 << "%" << std::endl;