backends : unsupport batched FA in CUDA and Vulkan

ggml-ci

Author: ggerganov

ggml/src/ggml-cuda/ggml-cuda.cu

@@ -3377,7 +3377,8 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
                 return false;
             }
             // TODO: support broadcast
-            // ref: https://github.com/ggml-org/llama.cpp/pull/14435
+            // note: this was initially implemented in https://github.com/ggml-org/llama.cpp/pull/14500, but
+            //       the interface of ggml_flash_attn_ext() changed in https://github.com/ggml-org/llama.cpp/pull/14505
             if (op->src[0]->ne[3] != 1) {
                 return false;
             }

ggml/src/ggml-vulkan/ggml-vulkan.cpp

@@ -10265,6 +10265,12 @@ static bool ggml_backend_vk_device_supports_op(ggml_backend_dev_t dev, const ggm
                 if (op->src[3] && op->src[3]->type != GGML_TYPE_F16) {
                     return false;
                 }
+                // TODO: support broadcast
+                // note: this was initially implemented in https://github.com/ggml-org/llama.cpp/pull/14449, but
+                //       the interface of ggml_flash_attn_ext() changed in https://github.com/ggml-org/llama.cpp/pull/14505
+                if (op->src[0]->ne[3] != 1) {
+                    return false;
+                }
                 // It's straightforward to support different K/V dequant, but would
                 // significantly increase the number of pipelines
                 if (op->src[1]->type != op->src[2]->type) {