Match Occupancy APIs syntax with CUDA (#1625)

* Match Occupancy APIs syntax with CUDA and fix tests using these APIs


[ROCm/hip-tests commit: acb2ea8618]
This commit is contained in:
satyanveshd
2020-01-30 02:35:53 +05:30
committed by GitHub
parent 5a83813c43
commit 884f9f5b84
@@ -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;