Skip to content

Add check with cublas call gemmex #74306

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Jul 31, 2025
Merged

Add check with cublas call gemmex #74306

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b31bd33
add check with cublas call gemmex
ggggxm Jul 30, 2025
ad42cac
fix typo
ggggxm Jul 30, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
19 changes: 19 additions & 0 deletions paddle/phi/kernels/funcs/blas/blas_impl.cu.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1260,6 +1260,14 @@ struct CUBlas<phi::dtype::complex<double>> {
}
};

inline void CheckGEMMNSize(int64_t N) {
constexpr int64_t kMaxN = 1073741823;
if (N > kMaxN) {
PADDLE_THROW(common::errors::Unimplemented(
"cublas GEMM does not support N > %ld. Got N = %ld. ", kMaxN, N));
}
}

template <>
template <typename T>
void Blas<phi::GPUContext>::GEMM(CBLAS_TRANSPOSE transA,
Expand Down Expand Up @@ -1307,6 +1315,7 @@ void Blas<phi::GPUContext>::GEMM(CBLAS_TRANSPOSE transA,
"GEMM_EX_64 is not supported on cuda < 12.3"));
#endif
} else {
CheckGEMMNSize(N);
CUBlas<T>::GEMM_EX(&cuda_ctx,
cuTransB,
cuTransA,
Expand Down Expand Up @@ -1418,6 +1427,7 @@ inline void Blas<phi::GPUContext>::GEMM(CBLAS_TRANSPOSE transA,
"GEMM_EX_64 is not supported on cuda < 12.3"));
#endif // CUDA_VERSION >= 12030
} else {
CheckGEMMNSize(N);
CUBlas<phi::dtype::float16>::GEMM_EX(&cuda_ctx,
cuTransB,
cuTransA,
Expand Down Expand Up @@ -1514,6 +1524,7 @@ void Blas<phi::GPUContext>::GEMM(CBLAS_TRANSPOSE transA,
"GEMM_EX_64 is not supported on cuda < 12.3"));
#endif
} else {
CheckGEMMNSize(N);
CUBlas<T>::GEMM_EX(&cuda_ctx,
cuTransB,
cuTransA,
Expand Down Expand Up @@ -1627,6 +1638,7 @@ inline void Blas<phi::GPUContext>::GEMM(CBLAS_TRANSPOSE transA,
#endif // CUDA_VERSION >= 12030
} else {
#if CUDA_VERSION >= 8000
CheckGEMMNSize(N);
CUBlas<phi::dtype::float16>::GEMM_EX(&cuda_ctx,
cuTransB,
cuTransA,
Expand Down Expand Up @@ -1736,6 +1748,7 @@ inline void Blas<phi::GPUContext>::GEMM(CBLAS_TRANSPOSE transA,
"cublasGemmEx_64 is not supported on cuda < 12.3"));
#endif // CUDA_VERSION >= 12030
} else {
CheckGEMMNSize(N);
dev_ctx_.TensorCoreCublasCallIfAvailable([&](cublasHandle_t handle) {
PADDLE_ENFORCE_GPU_SUCCESS(
phi::dynload::cublasGemmEx(handle,
Expand Down Expand Up @@ -1836,6 +1849,7 @@ inline void Blas<phi::GPUContext>::GEMM(CBLAS_TRANSPOSE transA,
"cublasGemmEx_64 is not supported on cuda < 12.3"));
#endif // CUDA_VERSION >= 12030
} else {
CheckGEMMNSize(N);
dev_ctx_.TensorCoreCublasCallIfAvailable([&](cublasHandle_t handle) {
PADDLE_ENFORCE_GPU_SUCCESS(
phi::dynload::cublasGemmEx(handle,
Expand Down Expand Up @@ -1931,6 +1945,7 @@ inline void Blas<phi::GPUContext>::GEMM(CBLAS_TRANSPOSE transA,
#endif // CUDA_VERSION >= 12030
} else {
#if CUDA_VERSION >= 8000
CheckGEMMNSize(N);
CUBlas<phi::dtype::complex<float>>::GEMM_EX(&cuda_ctx,
cuTransB,
cuTransA,
Expand Down Expand Up @@ -2040,6 +2055,7 @@ inline void Blas<phi::GPUContext>::GEMM(CBLAS_TRANSPOSE transA,
#endif // CUDA_VERSION >= 12030
} else {
#if CUDA_VERSION >= 8000
CheckGEMMNSize(N);
CUBlas<phi::dtype::complex<double>>::GEMM_EX(&cuda_ctx,
cuTransB,
cuTransA,
Expand Down Expand Up @@ -2101,6 +2117,7 @@ void Blas<phi::GPUContext>::GEMM(bool transA,
cublasOperation_t cuTransB = transB ? CUBLAS_OP_T : CUBLAS_OP_N;

#if CUDA_VERSION >= 8000
CheckGEMMNSize(N);
if (FLAGS_enable_cublas_tensor_op_math && std::is_same<T, float>::value) {
auto &cuda_ctx = const_cast<phi::GPUContext &>(dev_ctx_);
CUBlas<T>::GEMM_EX(&cuda_ctx,
Expand Down Expand Up @@ -2173,6 +2190,7 @@ inline void Blas<phi::GPUContext>::GEMM(bool transA,
if (use_tensor_op_math) {
algo = CUBLAS_GEMM_DFALT_TENSOR_OP;
}
CheckGEMMNSize(N);
dev_ctx_.TensorCoreCublasCallIfAvailable([&](cublasHandle_t handle) {
PADDLE_ENFORCE_GPU_SUCCESS(phi::dynload::cublasGemmEx(handle,
cuTransB,
Expand Down Expand Up @@ -2234,6 +2252,7 @@ inline void Blas<phi::GPUContext>::GEMM(bool transA,
algo = CUBLAS_GEMM_DFALT_TENSOR_OP;
}

CheckGEMMNSize(N);
dev_ctx_.TensorCoreCublasCallIfAvailable([&](cublasHandle_t handle) {
PADDLE_ENFORCE_GPU_SUCCESS(phi::dynload::cublasGemmEx(handle,
cuTransB,
Expand Down
Loading