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 3bf8e9ad8e
commit e1fe095471
78 changed files with 246 additions and 264 deletions
Download Patch File Download Diff File Expand all files Collapse all files
hipamd/tests/src/kernel/hipLanguageExtensions.cpp
+5 -9
View File
@@ -31,7 +31,7 @@ THE SOFTWARE.
#include <test_common.h>
#ifdef __HCC__
#include <amp.h>
#include <hc.hpp>
#endif
// cudaA
@@ -53,7 +53,7 @@ __device__ __forceinline__ int sum1_forceinline(int a) { return a + 1; };
__device__ __host__ float PlusOne(float x) { return x + 1.0; }
__global__ void MyKernel(const hipLaunchParm lp, const float* a, const float* b, float* c,
__global__ void MyKernel(const float* a, const float* b, float* c,
unsigned N) {
// KERNELBEGIN;
@@ -71,12 +71,12 @@ void callMyKernel() {
const unsigned blockSize = 256;
unsigned N = blockSize;
hipLaunchKernel(MyKernel, dim3(N / blockSize), dim3(blockSize), 0, 0, a, b, c, N);
hipLaunchKernelGGL(MyKernel, dim3(N / blockSize), dim3(blockSize), 0, 0, a, b, c, N);
}
template <typename T>
__global__ void vectorADD(const hipLaunchParm lp, T __restrict__* A_d, T* B_d, T* C_d, size_t N) {
__global__ void vectorADD(T __restrict__* A_d, T* B_d, T* C_d, size_t N) {
// KERNELBEGIN;
#ifdef NOT_YET
int a = __shfl_up(x, 1);
@@ -93,11 +93,7 @@ __global__ void vectorADD(const hipLaunchParm lp, T __restrict__* A_d, T* B_d, T
int b = threadIdx.x;
int c;
// TODO - move to HIP atomics when ready.
concurrency ::atomic_fetch_add(&c, b);
// Concurrency::atomic_add_unsigned (&x, a);
// concurrency ::atomic_add_ (x, a);
atomicAdd(&c, b);
#endif
__syncthreads();