Fixed the CUDA interface: replaced nullptr with 0
parent
e71c037304
commit
ad197da08d
|
@ -494,7 +494,7 @@ StatusCode Gemm(const Layout layout, const Transpose a_transpose, const Transpos
|
||||||
const T beta,
|
const T beta,
|
||||||
CUdeviceptr c_buffer, const size_t c_offset, const size_t c_ld,
|
CUdeviceptr c_buffer, const size_t c_offset, const size_t c_ld,
|
||||||
const CUcontext context, const CUdevice device,
|
const CUcontext context, const CUdevice device,
|
||||||
CUdeviceptr temp_buffer = nullptr);
|
CUdeviceptr temp_buffer = 0);
|
||||||
|
|
||||||
// Symmetric matrix-matrix multiplication: SSYMM/DSYMM/CSYMM/ZSYMM/HSYMM
|
// Symmetric matrix-matrix multiplication: SSYMM/DSYMM/CSYMM/ZSYMM/HSYMM
|
||||||
template <typename T>
|
template <typename T>
|
||||||
|
|
|
@ -1732,8 +1732,8 @@ StatusCode Gemm(const Layout layout, const Transpose a_transpose, const Transpos
|
||||||
const auto device_cpp = Device(device);
|
const auto device_cpp = Device(device);
|
||||||
auto queue_cpp = Queue(context_cpp, device_cpp);
|
auto queue_cpp = Queue(context_cpp, device_cpp);
|
||||||
auto routine = Xgemm<T>(queue_cpp, nullptr);
|
auto routine = Xgemm<T>(queue_cpp, nullptr);
|
||||||
const auto temp_buffer_provided = temp_buffer != nullptr;
|
const auto temp_buffer_provided = temp_buffer != 0;
|
||||||
auto temp_buffer_cpp = temp_buffer_provided ? Buffer<T>(temp_buffer) : Buffer<T>(nullptr);
|
auto temp_buffer_cpp = temp_buffer_provided ? Buffer<T>(temp_buffer) : Buffer<T>(0);
|
||||||
routine.DoGemm(layout, a_transpose, b_transpose,
|
routine.DoGemm(layout, a_transpose, b_transpose,
|
||||||
m, n, k,
|
m, n, k,
|
||||||
alpha,
|
alpha,
|
||||||
|
|
|
@ -159,7 +159,7 @@ class Xgemm: public Routine {
|
||||||
const Buffer<T> &b_buffer, const size_t b_offset, const size_t b_ld,
|
const Buffer<T> &b_buffer, const size_t b_offset, const size_t b_ld,
|
||||||
const T beta,
|
const T beta,
|
||||||
const Buffer<T> &c_buffer, const size_t c_offset, const size_t c_ld,
|
const Buffer<T> &c_buffer, const size_t c_offset, const size_t c_ld,
|
||||||
const Buffer<T> &temp_buffer = Buffer<T>(nullptr), const bool temp_buffer_provided = false);
|
const Buffer<T> &temp_buffer = Buffer<T>(0), const bool temp_buffer_provided = false);
|
||||||
|
|
||||||
// Indirect version of GEMM (with pre and post-processing kernels)
|
// Indirect version of GEMM (with pre and post-processing kernels)
|
||||||
void GemmIndirect(const size_t m, const size_t n, const size_t k,
|
void GemmIndirect(const size_t m, const size_t n, const size_t k,
|
||||||
|
|
Loading…
Reference in New Issue