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 committed 2018-10-17 14:32:25 +05:30
1 parent c9e4ed1149
commit 52e320f396
78 files changed
+246 -264
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

tests/src/runtimeApi/stream/hipAPIStreamDisable.cpp
+4 -4
View File
@@ -28,7 +28,7 @@ THE SOFTWARE.
const int NN = 1 << 21;
__global__ void kernel(hipLaunchParm lp, float* x, float* y, int n) {
__global__ void kernel(float* x, float* y, int n) {
int tid = threadIdx.x;
if (tid < 1) {
for (int i = 0; i < n; i++) {
@@ -38,7 +38,7 @@ __global__ void kernel(hipLaunchParm lp, float* x, float* y, int n) {
}
}
__global__ void nKernel(hipLaunchParm lp, float* y) {
__global__ void nKernel(float* y) {
int tid = threadIdx.x;
y[tid] = y[tid] + 1.0f;
}
@@ -55,8 +55,8 @@ int main() {
for (int i = 0; i < num_streams; i++) {
HIPCHECK(hipStreamCreate(&streams[i]));
HIPCHECK(hipMalloc(&data[i], NN * sizeof(float)));
hipLaunchKernel(HIP_KERNEL_NAME(kernel), dim3(1), dim3(1), 0, streams[i], data[i], xd, N);
hipLaunchKernel(HIP_KERNEL_NAME(nKernel), dim3(1), dim3(1), 0, 0, yd);
hipLaunchKernelGGL(HIP_KERNEL_NAME(kernel), dim3(1), dim3(1), 0, streams[i], data[i], xd, N);
hipLaunchKernelGGL(HIP_KERNEL_NAME(nKernel), dim3(1), dim3(1), 0, 0, yd);
}
HIPCHECK(hipMemcpy(&x, xd, sizeof(float), hipMemcpyDeviceToHost));
tests/src/runtimeApi/stream/hipAPIStreamEnable.cpp
+4 -4
View File
@@ -30,7 +30,7 @@ THE SOFTWARE.
const int NN = 1 << 21;
__global__ void kernel(hipLaunchParm lp, float* x, float* y, int n) {
__global__ void kernel(float* x, float* y, int n) {
int tid = threadIdx.x;
if (tid < 1) {
for (int i = 0; i < n; i++) {
@@ -40,7 +40,7 @@ __global__ void kernel(hipLaunchParm lp, float* x, float* y, int n) {
}
}
__global__ void nKernel(hipLaunchParm lp, float* y) {
__global__ void nKernel(float* y) {
int tid = threadIdx.x;
y[tid] = y[tid] + 1.0f;
}
@@ -57,8 +57,8 @@ int main() {
for (int i = 0; i < num_streams; i++) {
HIPCHECK(hipStreamCreate(&streams[i]));
HIPCHECK(hipMalloc(&data[i], NN * sizeof(float)));
hipLaunchKernel(HIP_KERNEL_NAME(kernel), dim3(1), dim3(1), 0, streams[i], data[i], xd, N);
hipLaunchKernel(HIP_KERNEL_NAME(nKernel), dim3(1), dim3(1), 0, 0, yd);
hipLaunchKernelGGL(HIP_KERNEL_NAME(kernel), dim3(1), dim3(1), 0, streams[i], data[i], xd, N);
hipLaunchKernelGGL(HIP_KERNEL_NAME(nKernel), dim3(1), dim3(1), 0, 0, yd);
}
HIPCHECK(hipMemcpy(&x, xd, sizeof(float), hipMemcpyDeviceToHost));
tests/src/runtimeApi/stream/hipNullStream.cpp
+4 -4
View File
@@ -33,7 +33,7 @@ int p_db = 0;
template <typename T>
__global__ void vectorADDRepeat(hipLaunchParm lp, const T* A_d, const T* B_d, T* C_d, size_t NELEM,
__global__ void vectorADDRepeat(const T* A_d, const T* B_d, T* C_d, size_t NELEM,
int repeat) {
size_t offset = (blockIdx.x * blockDim.x + threadIdx.x);
size_t stride = blockDim.x * gridDim.x;
@@ -113,7 +113,7 @@ void Streamer<T>::enqueAsync() {
printf("testing: %s numElements=%zu size=%6.2fMB\n", __func__, _numElements,
_numElements * sizeof(T) / 1024.0 / 1024.0);
unsigned blocks = HipTest::setNumBlocks(blocksPerCU, threadsPerBlock, _numElements);
hipLaunchKernel(vectorADDRepeat, dim3(blocks), dim3(threadsPerBlock), 0, _stream,
hipLaunchKernelGGL(vectorADDRepeat, dim3(blocks), dim3(threadsPerBlock), 0, _stream,
static_cast<const T*>(_A_d), static_cast<const T*>(_B_d), _C_d, _numElements,
p_repeat);
}
@@ -206,7 +206,7 @@ int main(int argc, char* argv[]) {
// Dispatch to NULL stream, should wait for prior async activity to complete before
// beginning:
hipLaunchKernel(vectorADDRepeat, dim3(blocks), dim3(threadsPerBlock), 0,
hipLaunchKernelGGL(vectorADDRepeat, dim3(blocks), dim3(threadsPerBlock), 0,
0 /*nullstream*/, static_cast<const int*>(lastStreamer->_C_d),
static_cast<const int*>(lastStreamer->_C_d), nullStreamer->_C_d,
numElements, 1 /*repeat*/);
@@ -242,7 +242,7 @@ int main(int argc, char* argv[]) {
// Dispatch to NULL stream, should wait for prior async activity to complete before
// beginning:
hipLaunchKernel(vectorADDRepeat, dim3(blocks), dim3(threadsPerBlock), 0,
hipLaunchKernelGGL(vectorADDRepeat, dim3(blocks), dim3(threadsPerBlock), 0,
0 /*nullstream*/, static_cast<const int*>(lastStreamer->_C_d),
static_cast<const int*>(lastStreamer->_C_d), nullStreamer->_C_d,
numElements, 1 /*repeat*/);
tests/src/runtimeApi/stream/hipStream.h
+1 -1
View File
@@ -72,7 +72,7 @@ void D2H(T* Dst, T* Src, size_t size) {
}
template <typename T>
__global__ void Inc(hipLaunchParm lp, T* In) {
__global__ void Inc(T* In) {
int tx = threadIdx.x + blockIdx.x * blockDim.x;
In[tx] = In[tx] + 1;
}
tests/src/runtimeApi/stream/hipStreamL5.cpp
+20 -20
View File
@@ -76,7 +76,7 @@ void test12345() {
H2HAsync(Bh, Ah, size, stream);
H2DAsync(Ad, Bh, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
D2DAsync(Bd, Ad, size, stream);
D2HAsync(Ch, Bd, size, stream);
HIPCHECK(hipDeviceSynchronize());
@@ -111,7 +111,7 @@ void test13452() {
H2D(Ad, Dh, size);
H2HAsync(Bh, Ah, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
D2DAsync(Bd, Ad, size, stream);
D2HAsync(Ch, Bd, size, stream);
H2DAsync(Cd, Ch, size, stream);
@@ -152,7 +152,7 @@ void test14523() {
D2DAsync(Bd, Ad, size, stream);
D2HAsync(Ch, Bd, size, stream);
H2DAsync(Cd, Ch, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Cd);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Cd);
HIPCHECK(hipDeviceSynchronize());
@@ -190,7 +190,7 @@ void test15234() {
H2HAsync(Bh, Ah, size, stream);
D2HAsync(Ch, Ad, size, stream);
H2DAsync(Bd, Ch, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Bd);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Bd);
D2DAsync(Cd, Bd, size, stream);
D2H(Eh, Cd, size);
@@ -217,7 +217,7 @@ void test23451() {
setArray(Ah, N, T(1));
H2DAsync(Ad, Ah, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
D2DAsync(Bd, Ad, size, stream);
D2HAsync(Bh, Bd, size, stream);
H2HAsync(Ch, Bh, size, stream);
@@ -254,7 +254,7 @@ void test24513() {
D2DAsync(Bd, Ad, size, stream);
D2HAsync(Bh, Bd, size, stream);
H2HAsync(Ch, Bh, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Cd);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Cd);
HIPCHECK(hipDeviceSynchronize());
D2H(Eh, Cd, size);
@@ -291,7 +291,7 @@ void test25134() {
H2DAsync(Ad, Ah, size, stream);
D2HAsync(Bh, Ad, size, stream);
H2HAsync(Ch, Bh, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Bd);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Bd);
D2DAsync(Cd, Bd, size, stream);
D2H(Eh, Cd, size);
@@ -324,7 +324,7 @@ void test21345() {
H2DAsync(Ad, Ah, size, stream);
H2HAsync(Ch, Bh, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
D2DAsync(Bd, Ad, size, stream);
D2HAsync(Dh, Bd, size, stream);
@@ -358,7 +358,7 @@ void test34512() {
H2D(Ad, Ah, size);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
D2DAsync(Bd, Ad, size, stream);
D2HAsync(Bh, Bd, size, stream);
H2HAsync(Ch, Bh, size, stream);
@@ -393,7 +393,7 @@ void test35124() {
H2D(Ad, Dh, size);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
D2HAsync(Ah, Ad, size, stream);
H2HAsync(Bh, Ah, size, stream);
H2DAsync(Bd, Bh, size, stream);
@@ -430,7 +430,7 @@ void test31245() {
H2D(Ad, Dh, size);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
H2HAsync(Bh, Ah, size, stream);
H2DAsync(Bd, Bh, size, stream);
D2DAsync(Cd, Bd, size, stream);
@@ -469,7 +469,7 @@ void test32451() {
setArray(Eh, N, T(2));
H2D(Ad, Eh, size);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Ad);
H2DAsync(Bd, Ah, size, stream);
D2DAsync(Cd, Bd, size, stream);
D2HAsync(Bh, Cd, size, stream);
@@ -507,7 +507,7 @@ void test45123() {
D2HAsync(Ah, Bd, size, stream);
H2HAsync(Bh, Ah, size, stream);
H2DAsync(Cd, Bh, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Cd);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Cd);
D2H(Ch, Cd, size);
HIPCHECK(hipDeviceSynchronize());
@@ -539,7 +539,7 @@ void test41235() {
D2DAsync(Bd, Ad, size, stream);
D2HAsync(Ah, Bd, size, stream);
H2DAsync(Cd, Ah, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Cd);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Cd);
D2HAsync(Bh, Cd, size, stream);
HIPCHECK(hipDeviceSynchronize());
@@ -574,7 +574,7 @@ void test42351() {
D2DAsync(Bd, Ad, size, stream);
H2DAsync(Cd, Ah, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Cd);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Cd);
D2HAsync(Bh, Cd, size, stream);
H2HAsync(Ch, Bh, size, stream);
@@ -609,7 +609,7 @@ void test43512() {
H2D(Ad, Dh, size);
D2DAsync(Bd, Ad, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Bd);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Bd);
D2HAsync(Ah, Bd, size, stream);
H2HAsync(Bh, Ah, size, stream);
H2DAsync(Cd, Bh, size, stream);
@@ -645,7 +645,7 @@ void test51234() {
D2HAsync(Ah, Ad, size, stream);
H2HAsync(Bh, Ah, size, stream);
H2DAsync(Bd, Bh, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Bd);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Bd);
D2DAsync(Cd, Bd, size, stream);
D2H(Ch, Cd, size);
@@ -681,7 +681,7 @@ void test52341() {
D2HAsync(Ah, Ad, size, stream);
H2DAsync(Bd, Ah, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Bd);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Bd);
D2DAsync(Cd, Bd, size, stream);
H2HAsync(Ch, Bh, size, stream);
@@ -723,7 +723,7 @@ void test53412() {
H2D(Bd, Eh, size);
D2HAsync(Ah, Ad, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Bd);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Bd);
D2DAsync(Cd, Bd, size, stream);
H2HAsync(Ch, Bh, size, stream);
H2DAsync(Dd, Ch, size, stream);
@@ -770,7 +770,7 @@ void test54123() {
D2DAsync(Cd, Bd, size, stream);
H2HAsync(Ch, Bh, size, stream);
H2DAsync(Dd, Ch, size, stream);
hipLaunchKernel(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Dd);
hipLaunchKernelGGL(HIP_KERNEL_NAME(Inc), dim3(N / 512), dim3(512), 0, stream, Dd);
D2H(Fh, Cd, size);
D2H(Gh, Dd, size);