Merge branch 'privatestaging' of https://github.com/AMDComputeLibraries/HIP-privatestaging into privatestaging

[ROCm/hip commit: 1d2f267dc3]
This commit is contained in:
Ben Sander
2016-05-03 11:16:09 -05:00
melakukan ff19d93fff
7 mengubah file dengan 51 tambahan dan 32 penghapusan
@@ -30,7 +30,7 @@ THE SOFTWARE.
if (error != hipSuccess) { \
fprintf(stderr, "error: '%s'(%d) at %s:%d\n", hipGetErrorString(error), error,__FILE__, __LINE__); \
exit(EXIT_FAILURE);\
}\
}\
}
void __global__
@@ -43,59 +43,60 @@ bit_extract_kernel(hipLaunchParm lp, uint32_t *C_d, const uint32_t *A_d, size_t
#ifdef __HIP_PLATFORM_HCC__
C_d[i] = hc::__bitextract_u32(A_d[i], 8, 4);
#else /* defined __HIP_PLATFORM_NVCC__ or other path */
C_d[i] = ((A_d[i] & 0xf00) >> 8);
C_d[i] = ((A_d[i] & 0xf00) >> 8);
#endif
}
}
}
int main(int argc, char *argv[])
{
uint32_t *A_d, *C_d;
uint32_t *A_h, *C_h;
size_t N = 1000000;
size_t Nbytes = N * sizeof(uint32_t);
uint32_t *A_d, *C_d;
uint32_t *A_h, *C_h;
size_t N = 1000000;
size_t Nbytes = N * sizeof(uint32_t);
int deviceId;
CHECK (hipGetDevice(&deviceId));
hipDeviceProp_t props;
CHECK(hipGetDeviceProperties(&props, deviceId));
printf ("info: running on device #%d %s\n", deviceId, props.name);
hipDeviceProp_t props;
CHECK(hipGetDeviceProperties(&props, deviceId));
printf ("info: running on device #%d %s\n", deviceId, props.name);
printf ("info: allocate host mem (%6.2f MB)\n", 2*Nbytes/1024.0/1024.0);
A_h = (uint32_t*)malloc(Nbytes);
CHECK(A_h == 0 ? hipErrorMemoryAllocation : hipSuccess );
C_h = (uint32_t*)malloc(Nbytes);
CHECK(C_h == 0 ? hipErrorMemoryAllocation : hipSuccess );
printf ("info: allocate host mem (%6.2f MB)\n", 2*Nbytes/1024.0/1024.0);
A_h = (uint32_t*)malloc(Nbytes);
CHECK(A_h == 0 ? hipErrorMemoryAllocation : hipSuccess );
C_h = (uint32_t*)malloc(Nbytes);
CHECK(C_h == 0 ? hipErrorMemoryAllocation : hipSuccess );
for (size_t i=0; i<N; i++)
{
A_h[i] = i;
}
{
A_h[i] = i;
}
printf ("info: allocate device mem (%6.2f MB)\n", 2*Nbytes/1024.0/1024.0);
CHECK(hipMalloc(&A_d, Nbytes));
CHECK(hipMalloc(&C_d, Nbytes));
printf ("info: allocate device mem (%6.2f MB)\n", 2*Nbytes/1024.0/1024.0);
CHECK(hipMalloc(&A_d, Nbytes));
CHECK(hipMalloc(&C_d, Nbytes));
printf ("info: copy Host2Device\n");
printf ("info: copy Host2Device\n");
CHECK ( hipMemcpy(A_d, A_h, Nbytes, hipMemcpyHostToDevice));
printf ("info: launch 'bit_extract_kernel' \n");
const unsigned blocks = 512;
const unsigned threadsPerBlock = 256;
hipLaunchKernel(bit_extract_kernel, dim3(blocks), dim3(threadsPerBlock), 0, 0, C_d, A_d, N);
printf ("info: launch 'bit_extract_kernel' \n");
const unsigned blocks = 512;
const unsigned threadsPerBlock = 256;
hipLaunchKernel(bit_extract_kernel, dim3(blocks), dim3(threadsPerBlock), 0, 0, C_d, A_d, N);
printf ("info: copy Device2Host\n");
printf ("info: copy Device2Host\n");
CHECK ( hipMemcpy(C_h, C_d, Nbytes, hipMemcpyDeviceToHost));
printf ("info: check result\n");
printf ("info: check result\n");
for (size_t i=0; i<N; i++) {
unsigned Agold = ((A_h[i] & 0xf00) >> 8);
if (C_h[i] != Agold) {
if (C_h[i] != Agold) {
fprintf (stderr, "mismatch detected.\n");
printf ("%zu: %08x =? %08x (Ain=%08x)\n", i, C_h[i], Agold, A_h[i]);
CHECK(hipErrorUnknown);
}
}
CHECK(hipErrorUnknown);
}
}
printf ("PASSED!\n");
}
@@ -16,6 +16,7 @@
int main(int argc, char *argv[])
{
int sizeElements = 1000000;
bool pass = true;
// Allocate auto-managed host/device views of data:
concurrency::array_view<float> A(sizeElements);
@@ -43,6 +44,8 @@ int main(int argc, char *argv[])
// Because C is an array_view, the HCC runtime will copy C back to host at first access here:
if (C[i] != ref) {
printf ("error:%d computed=%6.2f, reference=%6.2f\n", i, C[i], ref);
pass = false;
}
};
if (pass) printf ("PASSED!\n");
}
@@ -11,6 +11,7 @@ int main(int argc, char *argv[])
{
int sizeElements = 1000000;
size_t sizeBytes = sizeElements * sizeof(float);
bool pass = true;
// Allocate host memory
float *A_h = (float*)malloc(sizeBytes);
@@ -54,6 +55,8 @@ int main(int argc, char *argv[])
float ref= 1.618f * i + 3.142f * i;
if (C_h[i] != ref) {
printf ("error:%d computed=%6.2f, reference=%6.2f\n", i, C_h[i], ref);
pass = false;
}
};
if (pass) printf ("PASSED!\n");
}
@@ -11,6 +11,7 @@ int main(int argc, char *argv[])
{
int sizeElements = 1000000;
size_t sizeBytes = sizeElements * sizeof(float);
bool pass = true;
// Allocate host memory
float *A_h = (float*)malloc(sizeBytes);
@@ -48,6 +49,8 @@ int main(int argc, char *argv[])
float ref= 1.618f * i + 3.142f * i;
if (C_h[i] != ref) {
printf ("error:%d computed=%6.2f, reference=%6.2f\n", i, C_h[i], ref);
pass = false;
}
};
if (pass) printf ("PASSED!\n");
}
@@ -3,6 +3,7 @@
int main(int argc, char *argv[])
{
int size = 1000000;
bool pass = true;
// Allocate auto-managed host/device views of data:
hc::array_view<float> A(size);
@@ -28,6 +29,8 @@ int main(int argc, char *argv[])
float ref= 1.618f * i + 3.142f * i;
if (C[i] != ref) {
printf ("error:%d computed=%6.2f, reference=%6.2f\n", i, C[i], ref);
pass = false;
}
};
if (pass) printf ("PASSED!\n");
}
@@ -16,6 +16,7 @@
int main(int argc, char *argv[])
{
int sizeElements = 1000000;
bool pass = true;
// Allocate auto-managed host/device views of data:
hc::array_view<float> A(sizeElements);
@@ -43,6 +44,8 @@ int main(int argc, char *argv[])
// Because C is an array_view, the HCC runtime will copy C back to host at first access here:
if (C[i] != ref) {
printf ("error:%d computed=%6.2f, reference=%6.2f\n", i, C[i], ref);
pass = false;
}
};
if (pass) printf ("PASSED!\n");
}
@@ -14,6 +14,7 @@ int main(int argc, char *argv[])
{
int sizeElements = 1000000;
size_t sizeBytes = sizeElements * sizeof(float);
bool pass = true;
// Allocate host memory
float *A_h = (float*)malloc(sizeBytes);
@@ -46,6 +47,8 @@ int main(int argc, char *argv[])
float ref= 1.618f * i + 3.142f * i;
if (C_h[i] != ref) {
printf ("error:%d computed=%6.2f, reference=%6.2f\n", i, C_h[i], ref);
pass = false;
}
};
if (pass) printf ("PASSED!\n");
}