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

metal : fix null-pipeline crash for F16 src1 mul_mat/mul_mat_id (#25648)

Browse Source 
* metal : fail closed on mul_mat shapes with missing F16 kernels

* metal : abort on nil pipeline in encoder_set_pipeline

* metal : address review comments

* metal : share mul_mat mm dispatch with supports_op
This commit is contained in:
QuintinShaw
2026-08-30 08:56:35 +03:00
committed by GitHub
parent bdf3955159
commit d882575cc8
4 changed files with 59 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-metal/ggml-metal-ops.cpp
+2 -17
View File
@@ -2362,10 +2362,6 @@ int ggml_metal_op_mul_mat(ggml_metal_op_t ctx, int idx) {
const int16_t r2 = ne12/ne02;
const int16_t r3 = ne13/ne03;
// find the break-even point where the matrix-matrix kernel becomes more efficient compared
// to the matrix-vector kernel
const int ne11_mm_min = 8;
// first try to use small-batch mat-mv kernels
// these should be efficient for BS [2, ~8]
if (op->src[1]->type == GGML_TYPE_F32 && (ne00%128 == 0) &&
@@ -2468,12 +2464,7 @@ int ggml_metal_op_mul_mat(ggml_metal_op_t ctx, int idx) {
ggml_metal_encoder_set_buffer (enc, ggml_metal_get_buffer_id(op), 3);
ggml_metal_encoder_dispatch_threadgroups(enc, ((ne01 + r0ptg - 1)/r0ptg), ((ne11 + r1ptg - 1)/r1ptg), ne12*ne13, 32, nsg, 1);
} else if (
!ggml_is_transposed(op->src[0]) &&
!ggml_is_transposed(op->src[1]) &&
// for now the matrix-matrix multiplication kernel only works on A14+/M1+ SoCs
// AMD GPU and older A-chips will reuse matrix-vector multiplication kernel
props_dev->has_simdgroup_mm && ne00 >= 64 && ne11 > ne11_mm_min) {
} else if (ggml_metal_op_mul_mat_use_mm(op, props_dev->has_simdgroup_mm)) {
//GGML_LOG_INFO("matrix: ne00 = %6d, ne01 = %6d, ne02 = %6d, ne11 = %6d, ne12 = %6d\n", ne00, ne01, ne02, ne11, ne12);
// some Metal matrix data types require aligned pointers
@@ -2622,13 +2613,7 @@ int ggml_metal_op_mul_mat_id(ggml_metal_op_t ctx, int idx) {
const uint32_t r2 = 1;
const uint32_t r3 = 1;
// find the break-even point where the matrix-matrix kernel becomes more efficient compared
// to the matrix-vector kernel
// ne20 = n_used_experts
// ne21 = n_rows (batch size)
const int ne21_mm_id_min = 32;
if (props_dev->has_simdgroup_mm && ne00 >= 64 && (ne21 >= ne21_mm_id_min)) {
if (ggml_metal_op_mul_mat_id_use_mm(op, props_dev->has_simdgroup_mm)) {
// some Metal matrix data types require aligned pointers
// ref: https://developer.apple.com/metal/Metal-Shading-Language-Specification.pdf (Table 2.5)
//switch (op->src[0]->type) {