summaryrefslogtreecommitdiff
path: root/llama.cpp/ggml/src/ggml-cuda/fattn-tile.cu
diff options
context:
space:
mode:
authorMitja Felicijan <mitja.felicijan@gmail.com>2026-02-12 20:57:17 +0100
committerMitja Felicijan <mitja.felicijan@gmail.com>2026-02-12 20:57:17 +0100
commitb333b06772c89d96aacb5490d6a219fba7c09cc6 (patch)
tree211df60083a5946baa2ed61d33d8121b7e251b06 /llama.cpp/ggml/src/ggml-cuda/fattn-tile.cu
downloadllmnpc-b333b06772c89d96aacb5490d6a219fba7c09cc6.tar.gz
Engage!
Diffstat (limited to 'llama.cpp/ggml/src/ggml-cuda/fattn-tile.cu')
-rw-r--r--llama.cpp/ggml/src/ggml-cuda/fattn-tile.cu49
1 files changed, 49 insertions, 0 deletions
diff --git a/llama.cpp/ggml/src/ggml-cuda/fattn-tile.cu b/llama.cpp/ggml/src/ggml-cuda/fattn-tile.cu
new file mode 100644
index 0000000..3fcb09b
--- /dev/null
+++ b/llama.cpp/ggml/src/ggml-cuda/fattn-tile.cu
@@ -0,0 +1,49 @@
+#include "common.cuh"
+#include "fattn-tile.cuh"
+#include "fattn-wmma-f16.cuh"
+
+void ggml_cuda_flash_attn_ext_tile(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+ const ggml_tensor * K = dst->src[1];
+ const ggml_tensor * V = dst->src[2];
+ switch (K->ne[0]) {
+ case 40: {
+ GGML_ASSERT(V->ne[0] == K->ne[0]);
+ ggml_cuda_flash_attn_ext_tile_case< 40, 40>(ctx, dst);
+ } break;
+ case 64: {
+ GGML_ASSERT(V->ne[0] == K->ne[0]);
+ ggml_cuda_flash_attn_ext_tile_case< 64, 64>(ctx, dst);
+ } break;
+ case 72: {
+ GGML_ASSERT(V->ne[0] == K->ne[0]);
+ ggml_cuda_flash_attn_ext_tile_case< 72, 72>(ctx, dst);
+ } break;
+ case 80: {
+ GGML_ASSERT(V->ne[0] == K->ne[0]);
+ ggml_cuda_flash_attn_ext_tile_case< 80, 80>(ctx, dst);
+ } break;
+ case 96: {
+ GGML_ASSERT(V->ne[0] == K->ne[0]);
+ ggml_cuda_flash_attn_ext_tile_case< 96, 96>(ctx, dst);
+ } break;
+ case 112: {
+ GGML_ASSERT(V->ne[0] == K->ne[0]);
+ ggml_cuda_flash_attn_ext_tile_case<112, 112>(ctx, dst);
+ } break;
+ case 128: {
+ GGML_ASSERT(V->ne[0] == K->ne[0]);
+ ggml_cuda_flash_attn_ext_tile_case<128, 128>(ctx, dst);
+ } break;
+ case 256: {
+ GGML_ASSERT(V->ne[0] == K->ne[0]);
+ ggml_cuda_flash_attn_ext_tile_case<256, 256>(ctx, dst);
+ } break;
+ case 576: {
+ GGML_ASSERT(V->ne[0] == 512);
+ ggml_cuda_flash_attn_ext_tile_case<576, 512>(ctx, dst);
+ } break;
+ default: {
+ GGML_ABORT("Unsupported head size");
+ } break;
+ }
+}