Skip to content

Commit

Permalink
[HIPIFY][ROCm#1086][hipBLAS][tests] Added tests on __half functions
Browse files Browse the repository at this point in the history
  • Loading branch information
@emankov
emankov committed Oct 23, 2023
1 parent 6920a21 commit a85ddee
Show file tree
Hide file tree
Showing 2 changed files with 98 additions and 6 deletions.
52 changes: 49 additions & 3 deletions tests/unit_tests/synthetic/libraries/cublas2hipblas.cu
View file
Original file line number Diff line number Diff line change
Expand Up @@ -214,6 +214,49 @@ int main() {
float** fCarray = 0;
float** fTauarray = 0;

// CHECK: __half* ha = 0;
__half* ha = 0;
// CHECK: __half* hA = 0;
__half* hA = 0;
// CHECK: __half* hb = 0;
__half* hb = 0;
// CHECK: __half* hB = 0;
__half* hB = 0;
// CHECK: __half* hc = 0;
__half* hc = 0;
// CHECK: __half* hC = 0;
__half* hC = 0;

// CHECK: __half** hAarray = 0;
__half** hAarray = 0;
// CHECK: const __half** const hAarray_const = const_cast<const __half**>(hAarray);
const __half** const hAarray_const = const_cast<const __half**>(hAarray);
// CHECK: __half** hBarray = 0;
__half** hBarray = 0;
// CHECK: const __half** const hBarray_const = const_cast<const __half**>(hBarray);
const __half** const hBarray_const = const_cast<const __half**>(hBarray);
// CHECK: __half** hCarray = 0;
__half** hCarray = 0;
// CHECK: const __half** const hCarray_const = const_cast<const __half**>(hCarray);
const __half** const hCarray_const = const_cast<const __half**>(hCarray);
// CHECK: __half** hxarray = 0;
__half** hxarray = 0;
// CHECK: const __half** const hxarray_const = const_cast<const __half**>(hxarray_const);
const __half** const hxarray_const = const_cast<const __half**>(hxarray_const);
// CHECK: __half** hyarray = 0;
__half** hyarray = 0;

// CHECK: hip_bfloat16** bf16Aarray = 0;
__nv_bfloat16** bf16Aarray = 0;
// CHECK: const hip_bfloat16** const bf16Aarray_const = const_cast<const hip_bfloat16**>(bf16Aarray);
const __nv_bfloat16** const bf16Aarray_const = const_cast<const __nv_bfloat16**>(bf16Aarray);
// CHECK: hip_bfloat16** bf16xarray = 0;
__nv_bfloat16** bf16xarray = 0;
// CHECK: const hip_bfloat16** const bf16xarray_const = const_cast<const hip_bfloat16**>(bf16xarray_const);
const __nv_bfloat16** const bf16xarray_const = const_cast<const __nv_bfloat16**>(bf16xarray_const);
// CHECK: hip_bfloat16** bf16yarray = 0;
__nv_bfloat16** bf16yarray = 0;

double da = 0;
double dA = 0;
double db = 0;
Expand Down Expand Up @@ -1065,9 +1108,10 @@ int main() {
// CHECK: blasStatus = hipblasZgemm_v2(blasHandle, transa, transb, m, n, k, &dcomplexa, &dcomplexA, lda, &dcomplexB, ldb, &dcomplexb, &dcomplexC, ldc);
blasStatus = cublasZgemm_v2(blasHandle, transa, transb, m, n, k, &dcomplexa, &dcomplexA, lda, &dcomplexB, ldb, &dcomplexb, &dcomplexC, ldc);

// TODO: __half -> hipblasHalf
// CUDA: CUBLASAPI cublasStatus_t CUBLASWINAPI cublasHgemm(cublasHandle_t handle, cublasOperation_t transa, cublasOperation_t transb, int m, int n, int k, const __half* alpha, const __half* A, int lda, const __half* B, int ldb, const __half* beta, __half* C, int ldc);
// HIP: HIPBLAS_EXPORT hipblasStatus_t hipblasHgemm(hipblasHandle_t handle, hipblasOperation_t transA, hipblasOperation_t transB, int m, int n, int k, const hipblasHalf* alpha, const hipblasHalf* AP, int lda, const hipblasHalf* BP, int ldb, const hipblasHalf* beta, hipblasHalf* CP, int ldc);
// CHECK: blasStatus = hipblasHgemm(blasHandle, transa, transb, m, n, k, ha, hA, lda, hB, ldb, hb, hC, ldc);
blasStatus = cublasHgemm(blasHandle, transa, transb, m, n, k, ha, hA, lda, hB, ldb, hb, hC, ldc);

// CUDA: CUBLASAPI cublasStatus_t CUBLASWINAPI cublasSgemmBatched(cublasHandle_t handle, cublasOperation_t transa, cublasOperation_t transb, int m, int n, int k, const float* alpha, const float* const Aarray[], int lda, const float* const Barray[], int ldb, const float* beta, float* const Carray[], int ldc, int batchCount);
// HIP: HIPBLAS_EXPORT hipblasStatus_t hipblasSgemmBatched(hipblasHandle_t handle, hipblasOperation_t transA, hipblasOperation_t transB, int m, int n, int k, const float* alpha, const float* const AP[], int lda, const float* const BP[], int ldb, const float* beta, float* const CP[], int ldc, int batchCount);
Expand All @@ -1079,9 +1123,10 @@ int main() {
// CHECK: blasStatus = hipblasDgemmBatched(blasHandle, transa, transb, m, n, k, &da, dAarray_const, lda, dBarray_const, ldb, &db, dCarray, ldc, batchCount);
blasStatus = cublasDgemmBatched(blasHandle, transa, transb, m, n, k, &da, dAarray_const, lda, dBarray_const, ldb, &db, dCarray, ldc, batchCount);

// TODO: __half -> hipblasHalf
// CUDA: CUBLASAPI cublasStatus_t CUBLASWINAPI cublasHgemmBatched(cublasHandle_t handle, cublasOperation_t transa, cublasOperation_t transb, int m, int n, int k, const __half* alpha, const __half* const Aarray[], int lda, const __half* const Barray[], int ldb, const __half* beta, __half* const Carray[], int ldc, int batchCount);
// HIP: HIPBLAS_EXPORT hipblasStatus_t hipblasHgemmBatched(hipblasHandle_t handle, hipblasOperation_t transA, hipblasOperation_t transB, int m, int n, int k, const hipblasHalf* alpha, const hipblasHalf* const AP[], int lda, const hipblasHalf* const BP[], int ldb, const hipblasHalf* beta, hipblasHalf* const CP[], int ldc, int batchCount);
// CHECK: blasStatus = hipblasHgemmBatched(blasHandle, transa, transb, m, n, k, ha, hAarray_const, lda, hBarray_const, ldb, hb, hCarray, ldc, batchCount);
blasStatus = cublasHgemmBatched(blasHandle, transa, transb, m, n, k, ha, hAarray_const, lda, hBarray_const, ldb, hb, hCarray, ldc, batchCount);

// CUDA: CUBLASAPI cublasStatus_t CUBLASWINAPI cublasCgemmBatched(cublasHandle_t handle, cublasOperation_t transa, cublasOperation_t transb, int m, int n, int k, const cuComplex* alpha, const cuComplex* const Aarray[], int lda, const cuComplex* const Barray[], int ldb, const cuComplex* beta, cuComplex* const Carray[], int ldc, int batchCount);
// HIP: HIPBLAS_EXPORT hipblasStatus_t hipblasCgemmBatched_v2(hipblasHandle_t handle, hipblasOperation_t transA, hipblasOperation_t transB, int m, int n, int k, const hipComplex* alpha, const hipComplex* const AP[], int lda, const hipComplex* const BP[], int ldb, const hipComplex* beta, hipComplex* const CP[], int ldc, int batchCount);
Expand Down Expand Up @@ -1516,9 +1561,10 @@ int main() {
// CHECK: blasStatus = hipblasZgemmStridedBatched_v2(blasHandle, transa, transb, m, n, k, &dcomplexa, &dcomplexA, lda, strideA, &dcomplexB, ldb, strideB, &dcomplexb, &dcomplexC, ldc, strideC, batchCount);
blasStatus = cublasZgemmStridedBatched(blasHandle, transa, transb, m, n, k, &dcomplexa, &dcomplexA, lda, strideA, &dcomplexB, ldb, strideB, &dcomplexb, &dcomplexC, ldc, strideC, batchCount);

// TODO: __half -> hipblasHalf
// CUDA: CUBLASAPI cublasStatus_t CUBLASWINAPI cublasHgemmStridedBatched(cublasHandle_t handle, cublasOperation_t transa, cublasOperation_t transb, int m, int n, int k, const __half* alpha, const __half* A, int lda, long long int strideA, const __half* B, int ldb, long long int strideB, const __half* beta, __half* C, int ldc, long long int strideC, int batchCount);
// HIP: HIPBLAS_EXPORT hipblasStatus_t hipblasHgemmStridedBatched(hipblasHandle_t handle, hipblasOperation_t transA, hipblasOperation_t transB, int m, int n, int k, const hipblasHalf* alpha, const hipblasHalf* AP, int lda, long long strideA, const hipblasHalf* BP, int ldb, long long strideB, const hipblasHalf* beta, hipblasHalf* CP, int ldc, long long strideC, int batchCount);
// CHECK: blasStatus = hipblasHgemmStridedBatched(blasHandle, transa, transb, m, n, k, ha, hA, lda, strideA, hB, ldb, strideB, hb, hC, ldc, strideC, batchCount);
blasStatus = cublasHgemmStridedBatched(blasHandle, transa, transb, m, n, k, ha, hA, lda, strideA, hB, ldb, strideB, hb, hC, ldc, strideC, batchCount);

void* aptr = nullptr;
void* Aptr = nullptr;
Expand Down
52 changes: 49 additions & 3 deletions tests/unit_tests/synthetic/libraries/cublas2hipblas_v2.cu
View file
Original file line number Diff line number Diff line change
Expand Up @@ -225,6 +225,49 @@ int main() {
float** fCarray = 0;
float** fTauarray = 0;

// CHECK: __half* ha = 0;
__half* ha = 0;
// CHECK: __half* hA = 0;
__half* hA = 0;
// CHECK: __half* hb = 0;
__half* hb = 0;
// CHECK: __half* hB = 0;
__half* hB = 0;
// CHECK: __half* hc = 0;
__half* hc = 0;
// CHECK: __half* hC = 0;
__half* hC = 0;

// CHECK: __half** hAarray = 0;
__half** hAarray = 0;
// CHECK: const __half** const hAarray_const = const_cast<const __half**>(hAarray);
const __half** const hAarray_const = const_cast<const __half**>(hAarray);
// CHECK: __half** hBarray = 0;
__half** hBarray = 0;
// CHECK: const __half** const hBarray_const = const_cast<const __half**>(hBarray);
const __half** const hBarray_const = const_cast<const __half**>(hBarray);
// CHECK: __half** hCarray = 0;
__half** hCarray = 0;
// CHECK: const __half** const hCarray_const = const_cast<const __half**>(hCarray);
const __half** const hCarray_const = const_cast<const __half**>(hCarray);
// CHECK: __half** hxarray = 0;
__half** hxarray = 0;
// CHECK: const __half** const hxarray_const = const_cast<const __half**>(hxarray_const);
const __half** const hxarray_const = const_cast<const __half**>(hxarray_const);
// CHECK: __half** hyarray = 0;
__half** hyarray = 0;

// CHECK: hip_bfloat16** bf16Aarray = 0;
__nv_bfloat16** bf16Aarray = 0;
// CHECK: const hip_bfloat16** const bf16Aarray_const = const_cast<const hip_bfloat16**>(bf16Aarray);
const __nv_bfloat16** const bf16Aarray_const = const_cast<const __nv_bfloat16**>(bf16Aarray);
// CHECK: hip_bfloat16** bf16xarray = 0;
__nv_bfloat16** bf16xarray = 0;
// CHECK: const hip_bfloat16** const bf16xarray_const = const_cast<const hip_bfloat16**>(bf16xarray_const);
const __nv_bfloat16** const bf16xarray_const = const_cast<const __nv_bfloat16**>(bf16xarray_const);
// CHECK: hip_bfloat16** bf16yarray = 0;
__nv_bfloat16** bf16yarray = 0;

double da = 0;
double dA = 0;
double db = 0;
Expand Down Expand Up @@ -1196,9 +1239,10 @@ int main() {
blasStatus = cublasZgemm(blasHandle, transa, transb, m, n, k, &dcomplexa, &dcomplexA, lda, &dcomplexB, ldb, &dcomplexb, &dcomplexC, ldc);
blasStatus = cublasZgemm_v2(blasHandle, transa, transb, m, n, k, &dcomplexa, &dcomplexA, lda, &dcomplexB, ldb, &dcomplexb, &dcomplexC, ldc);

// TODO: __half -> hipblasHalf
// CUDA: CUBLASAPI cublasStatus_t CUBLASWINAPI cublasHgemm(cublasHandle_t handle, cublasOperation_t transa, cublasOperation_t transb, int m, int n, int k, const __half* alpha, const __half* A, int lda, const __half* B, int ldb, const __half* beta, __half* C, int ldc);
// HIP: HIPBLAS_EXPORT hipblasStatus_t hipblasHgemm(hipblasHandle_t handle, hipblasOperation_t transA, hipblasOperation_t transB, int m, int n, int k, const hipblasHalf* alpha, const hipblasHalf* AP, int lda, const hipblasHalf* BP, int ldb, const hipblasHalf* beta, hipblasHalf* CP, int ldc);
// CHECK: blasStatus = hipblasHgemm(blasHandle, transa, transb, m, n, k, ha, hA, lda, hB, ldb, hb, hC, ldc);
blasStatus = cublasHgemm(blasHandle, transa, transb, m, n, k, ha, hA, lda, hB, ldb, hb, hC, ldc);

// CUDA: CUBLASAPI cublasStatus_t CUBLASWINAPI cublasSgemmBatched(cublasHandle_t handle, cublasOperation_t transa, cublasOperation_t transb, int m, int n, int k, const float* alpha, const float* const Aarray[], int lda, const float* const Barray[], int ldb, const float* beta, float* const Carray[], int ldc, int batchCount);
// HIP: HIPBLAS_EXPORT hipblasStatus_t hipblasSgemmBatched(hipblasHandle_t handle, hipblasOperation_t transA, hipblasOperation_t transB, int m, int n, int k, const float* alpha, const float* const AP[], int lda, const float* const BP[], int ldb, const float* beta, float* const CP[], int ldc, int batchCount);
Expand All @@ -1210,9 +1254,10 @@ int main() {
// CHECK: blasStatus = hipblasDgemmBatched(blasHandle, transa, transb, m, n, k, &da, dAarray_const, lda, dBarray_const, ldb, &db, dCarray, ldc, batchCount);
blasStatus = cublasDgemmBatched(blasHandle, transa, transb, m, n, k, &da, dAarray_const, lda, dBarray_const, ldb, &db, dCarray, ldc, batchCount);

// TODO: __half -> hipblasHalf
// CUDA: CUBLASAPI cublasStatus_t CUBLASWINAPI cublasHgemmBatched(cublasHandle_t handle, cublasOperation_t transa, cublasOperation_t transb, int m, int n, int k, const __half* alpha, const __half* const Aarray[], int lda, const __half* const Barray[], int ldb, const __half* beta, __half* const Carray[], int ldc, int batchCount);
// HIP: HIPBLAS_EXPORT hipblasStatus_t hipblasHgemmBatched(hipblasHandle_t handle, hipblasOperation_t transA, hipblasOperation_t transB, int m, int n, int k, const hipblasHalf* alpha, const hipblasHalf* const AP[], int lda, const hipblasHalf* const BP[], int ldb, const hipblasHalf* beta, hipblasHalf* const CP[], int ldc, int batchCount);
// CHECK: blasStatus = hipblasHgemmBatched(blasHandle, transa, transb, m, n, k, ha, hAarray_const, lda, hBarray_const, ldb, hb, hCarray, ldc, batchCount);
blasStatus = cublasHgemmBatched(blasHandle, transa, transb, m, n, k, ha, hAarray_const, lda, hBarray_const, ldb, hb, hCarray, ldc, batchCount);

// CUDA: CUBLASAPI cublasStatus_t CUBLASWINAPI cublasCgemmBatched(cublasHandle_t handle, cublasOperation_t transa, cublasOperation_t transb, int m, int n, int k, const cuComplex* alpha, const cuComplex* const Aarray[], int lda, const cuComplex* const Barray[], int ldb, const cuComplex* beta, cuComplex* const Carray[], int ldc, int batchCount);
// HIP: HIPBLAS_EXPORT hipblasStatus_t hipblasCgemmBatched_v2(hipblasHandle_t handle, hipblasOperation_t transA, hipblasOperation_t transB, int m, int n, int k, const hipComplex* alpha, const hipComplex* const AP[], int lda, const hipComplex* const BP[], int ldb, const hipComplex* beta, hipComplex* const CP[], int ldc, int batchCount);
Expand Down Expand Up @@ -1671,9 +1716,10 @@ int main() {
// CHECK: blasStatus = hipblasZgemmStridedBatched_v2(blasHandle, transa, transb, m, n, k, &dcomplexa, &dcomplexA, lda, strideA, &dcomplexB, ldb, strideB, &dcomplexb, &dcomplexC, ldc, strideC, batchCount);
blasStatus = cublasZgemmStridedBatched(blasHandle, transa, transb, m, n, k, &dcomplexa, &dcomplexA, lda, strideA, &dcomplexB, ldb, strideB, &dcomplexb, &dcomplexC, ldc, strideC, batchCount);

// TODO: __half -> hipblasHalf
// CUDA: CUBLASAPI cublasStatus_t CUBLASWINAPI cublasHgemmStridedBatched(cublasHandle_t handle, cublasOperation_t transa, cublasOperation_t transb, int m, int n, int k, const __half* alpha, const __half* A, int lda, long long int strideA, const __half* B, int ldb, long long int strideB, const __half* beta, __half* C, int ldc, long long int strideC, int batchCount);
// HIP: HIPBLAS_EXPORT hipblasStatus_t hipblasHgemmStridedBatched(hipblasHandle_t handle, hipblasOperation_t transA, hipblasOperation_t transB, int m, int n, int k, const hipblasHalf* alpha, const hipblasHalf* AP, int lda, long long strideA, const hipblasHalf* BP, int ldb, long long strideB, const hipblasHalf* beta, hipblasHalf* CP, int ldc, long long strideC, int batchCount);
// CHECK: blasStatus = hipblasHgemmStridedBatched(blasHandle, transa, transb, m, n, k, ha, hA, lda, strideA, hB, ldb, strideB, hb, hC, ldc, strideC, batchCount);
blasStatus = cublasHgemmStridedBatched(blasHandle, transa, transb, m, n, k, ha, hA, lda, strideA, hB, ldb, strideB, hb, hC, ldc, strideC, batchCount);

void* aptr = nullptr;
void* Aptr = nullptr;
Expand Down

0 comments on commit a85ddee

Please sign in to comment.