cuda : add cublasGemmStridedBatchedEx for non-broadcasted cases

ggerganov · Oct 24, 2023 · 3d297c1 · 3d297c1
1 parent d415669
commit 3d297c1
Showing 1 changed file with 18 additions and 5 deletions.
diff --git a/ggml-cuda.cu b/ggml-cuda.cu
@@ -30,6 +30,7 @@
 #define cublasCreate hipblasCreate
 #define cublasGemmEx hipblasGemmEx
 #define cublasGemmBatchedEx hipblasGemmBatchedEx
+#define cublasGemmStridedBatchedEx hipblasGemmStridedBatchedEx
 #define cublasHandle_t hipblasHandle_t
 #define cublasSetMathMode(handle, mode) CUBLAS_STATUS_SUCCESS
 #define cublasSetStream hipblasSetStream
@@ -7125,17 +7126,29 @@ static void ggml_cuda_mul_mat_mat_batched_cublas(const ggml_tensor * src0, const
                 CUBLAS_CHECK(
                         cublasGemmEx(g_cublas_handles[id], CUBLAS_OP_T, CUBLAS_OP_N,
                             ne01, ne11, ne10,
-                            &alpha_f16, (char *) src0_as_f16 + i02*src0->nb[2]   + i03*src0->nb[3]  , CUDA_R_16F, nb01/sizeof(half),
-                                        (char *) src1_as_f16 + i12*src1->nb[2]/2 + i13*src1->nb[3]/2, CUDA_R_16F, nb11/sizeof(float),
-                            &beta_f16,  (char *)     dst_f16 + i12* dst->nb[2]/2 + i13* dst->nb[3]/2, CUDA_R_16F, ne01,
+                            &alpha_f16, (const char *) src0_as_f16 + i02*src0->nb[2]   + i03*src0->nb[3]  , CUDA_R_16F, nb01/sizeof(half),
+                                        (const char *) src1_as_f16 + i12*src1->nb[2]/2 + i13*src1->nb[3]/2, CUDA_R_16F, nb11/sizeof(float),
+                            &beta_f16,  (      char *)     dst_f16 + i12* dst->nb[2]/2 + i13* dst->nb[3]/2, CUDA_R_16F, ne01,
                             CUBLAS_COMPUTE_16F,
                             CUBLAS_GEMM_DEFAULT_TENSOR_OP));
             }
         }
     }
 #else
-    // use cublasGemmBatchedEx
-    {
+    if (r2 == 1 && r3 == 1 && src0->nb[2]*src0->ne[2] == src0->nb[3] && src1->nb[2]*src1->ne[2] == src1->nb[3]) {
+        // there is no broadcast and src0, src1 are contiguous across dims 2, 3
+        // use cublasGemmStridedBatchedEx
+        CUBLAS_CHECK(
+        cublasGemmStridedBatchedEx(g_cublas_handles[id], CUBLAS_OP_T, CUBLAS_OP_N,
+                ne01, ne11, ne10,
+                &alpha_f16, (const char *) src0_as_f16, CUDA_R_16F, nb01/sizeof(half),  src0->nb[2]/sizeof(half),  // strideA
+                            (const char *) src1_as_f16, CUDA_R_16F, nb11/sizeof(float), src1->nb[2]/sizeof(float), // strideB
+                &beta_f16,  (      char *)     dst_f16, CUDA_R_16F, ne01,                dst->nb[2]/sizeof(float), // strideC
+                ne12*ne13,
+                CUBLAS_COMPUTE_16F,
+                CUBLAS_GEMM_DEFAULT_TENSOR_OP));
+    } else {
+        // use cublasGemmBatchedEx
         const int ne23 = ne12*ne13;
 
         // TODO: avoid this alloc