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

Replace hipLaunchKernel -> hipLaunchKernelGGL

Browse Source 
Change-Id: I4d99009e1199811d417becf1e1b934ec4d4e30be
This commit is contained in:
Maneesh Gupta
2018-10-17 12:01:44 +05:30
parent c9e4ed1149
commit 52e320f396
78 changed files with 246 additions and 264 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/src/Functional/device/hipFuncDeviceSynchronize.cpp
+2 -2
View File
@@ -33,7 +33,7 @@ THE SOFTWARE.
#define _SIZE sizeof(int) * 1024 * 1024
#define NUM_STREAMS 2
__global__ void Iter(hipLaunchParm lp, int* Ad, int num) {
__global__ void Iter(int* Ad, int num) {
int tx = threadIdx.x + blockIdx.x * blockDim.x;
// Kernel loop designed to execute very slowly... ... ... so we can test timing-related
// behavior below
@@ -58,7 +58,7 @@ int main() {
HIPCHECK(hipMemcpyAsync(Ad[i], A[i], _SIZE, hipMemcpyHostToDevice, stream[i]));
}
for (int i = 0; i < NUM_STREAMS; i++) {
hipLaunchKernel(HIP_KERNEL_NAME(Iter), dim3(1), dim3(1), 0, stream[i], Ad[i], 1 << 30);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Iter), dim3(1), dim3(1), 0, stream[i], Ad[i], 1 << 30);
}
for (int i = 0; i < NUM_STREAMS; i++) {
HIPCHECK(hipMemcpyAsync(A[i], Ad[i], _SIZE, hipMemcpyDeviceToHost, stream[i]));