function definition of cublasDgemm is different from documentation in CUDA 8.0

Accelerated Computing GPU-Accelerated Libraries
1

According to the documentation the definition of cublasDgemm is

cublasStatus_t cublasDgemm(cublasHandle_t handle,
cublasOperation_t transa, cublasOperation_t transb,
int m, int n, int k,
const double *alpha,
const double *A, int lda,
const double *B, int ldb,
const double *beta,
double *C, int ldc)

However, when I use the function it is expected in the form

void_stdcall cublasDgemm(char transa, char transb,
int m, int n, int k,
double alpha,
const double *A, int lda,
const double *B, int ldb,
double beta,
double *C, int ldc)

Was the documentation not updated as the function was modified? I’m not sure what the character flags for transa and transb are supposed to be for no transpose.

2

You’re confused between the legacy CUBLAS API and the modern CUBLAS API.

This is covered in the CUBLAS documentation:

[url]http://docs.nvidia.com/cuda/cublas/index.html#appendix-a-using-the-cublas-legacy-api[/url]

The first calling parameter in the modern CUBLAS API is the handle. You’ll note that no handle exists in the legacy API call.

3

Thanks! I see now that the modern cublas library is cublas_v2.h and the modern cublasdgemm is cublasdgemm_v2