Function gemm_strided_batched_ex 

pub unsafe fn gemm_strided_batched_ex(
    handle: cublasHandle_t,
    transa: cublasOperation_t,
    transb: cublasOperation_t,
    m: c_int,
    n: c_int,
    k: c_int,
    alpha: *const c_void,
    a: CUdeviceptr,
    a_type: cudaDataType,
    lda: c_int,
    stride_a: c_longlong,
    b: CUdeviceptr,
    b_type: cudaDataType,
    ldb: c_int,
    stride_b: c_longlong,
    beta: *const c_void,
    c: CUdeviceptr,
    c_type: cudaDataType,
    ldc: c_int,
    stride_c: c_longlong,
    batch_count: c_int,
    compute_type: cublasComputeType_t,
    algo: cublasGemmAlgo_t,
) -> Result<(), GpuError>

cublasGemmStridedBatchedEx — typed-erased strided-batched gemm.

Safety

Like gemm_ex, plus stride_* describes the byte stride between consecutive batch entries inside a single allocation.