musa: enable MUSA graphs

Signed-off-by: Xiaodong Ye <[email protected]>

Author: yeahdongcn

ggml/src/ggml-cuda/vendors/musa.h

@@ -13,7 +13,7 @@
 #define CUBLAS_OP_N MUBLAS_OP_N
 #define CUBLAS_OP_T MUBLAS_OP_T
 #define CUBLAS_STATUS_SUCCESS MUBLAS_STATUS_SUCCESS
-#define CUBLAS_TF32_TENSOR_OP_MATH MUBLAS_MATH_MODE_DEFAULT
+#define CUBLAS_TF32_TENSOR_OP_MATH MUBLAS_DEFAULT_MATH
 #define CUDA_R_16F  MUSA_R_16F
 #define CUDA_R_16BF MUSA_R_16BF
 #define CUDA_R_32F  MUSA_R_32F
@@ -29,7 +29,7 @@
 #define cublasSgemm mublasSgemm
 #define cublasStatus_t mublasStatus_t
 #define cublasOperation_t mublasOperation_t
-#define cublasGetStatusString mublasStatus_to_string
+#define cublasGetStatusString mublasGetStatusString
 #define cudaDataType_t musaDataType_t
 #define cudaDeviceCanAccessPeer musaDeviceCanAccessPeer
 #define cudaDeviceDisablePeerAccess musaDeviceDisablePeerAccess

ggml/src/ggml-musa/CMakeLists.txt

@@ -67,6 +67,10 @@ if (MUSAToolkit_FOUND)
     add_compile_definitions(GGML_USE_MUSA)
     add_compile_definitions(GGML_CUDA_PEER_MAX_BATCH_SIZE=${GGML_CUDA_PEER_MAX_BATCH_SIZE})
 
+    if (GGML_CUDA_GRAPHS)
+        add_compile_definitions(GGML_CUDA_USE_GRAPHS)
+    endif()
+
     if (GGML_CUDA_FORCE_MMQ)
         add_compile_definitions(GGML_CUDA_FORCE_MMQ)
     endif()