Lumpiasty/llama.cpp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

metal : optimize Metal Tensor API usage for GGML_OP_MUL_MAT (#20962)

Browse Source 
* Optimize Metal Tensor API usage for matmul2d

Separates the Metal Tensor API (matmul2d) path in kernel_mul_mm into its own standalone kernel, gated by GGML_METAL_HAS_TENSOR.

The legacy simdgroup_matrix kernel is preserved under #else.

Previously both paths were interleaved via #ifdef blocks within a single kernel, forcing the tensor path to share the legacy kernel's data layout and threadgroup memory scheme. Splitting the kernel enabled memory and dispatch optimizations that weren't possible when the two paths shared code structure.

* cont : cleanup

* cont : cleanup

* cont : cleanup

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This commit is contained in:
Developer-Ecosystem-Engineering
2026-04-25 15:14:28 +03:00
committed by GitHub
co-authored by Georgi Gerganov
parent 9d34231bb8
commit d1649047a3
6 changed files with 189 additions and 109 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-metal/ggml-metal-ops.cpp
+6 -1
View File
@@ -2195,7 +2195,12 @@ int ggml_metal_op_mul_mat(ggml_metal_op_t ctx, int idx) {
const size_t smem = pipeline.smem;
ggml_metal_encoder_set_threadgroup_memory_size(enc, smem, 0);
ggml_metal_encoder_dispatch_threadgroups(enc, ((ne11 + 31)/32), ((ne01 + 63)/64), ne12*ne13, 128, 1, 1);
const int nr0 = pipeline.nr0;
const int nr1 = pipeline.nr1;
const int nsg = pipeline.nsg;
ggml_metal_encoder_dispatch_threadgroups(enc, ((ne11 + nr1 - 1) / nr1), ((ne01 + nr0 - 1) / nr0), ne12 * ne13, 32, nsg, 1);
} else {
auto pipeline = ggml_metal_library_get_pipeline_mul_mv(lib, op);