vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

metal : disable support for MUL_MAT F32 x F16

Browse source
This commit is contained in:
Georgi Gerganov 2024-01-23 15:50:56 +02:00
parent 44879ee885
commit 26d607608d
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
1 changed files with 2 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

3
ggml-metal.m
View file

@ -668,7 +668,8 @@ static bool ggml_metal_supports_op(const struct ggml_metal_context * ctx, const
return true; return true;
case GGML_OP_MUL_MAT: case GGML_OP_MUL_MAT:
case GGML_OP_MUL_MAT_ID: case GGML_OP_MUL_MAT_ID:
return ctx->support_simdgroup_reduction; return ctx->support_simdgroup_reduction &&
(op->src[0]->type != GGML_TYPE_F32 || op->src[1]->type == GGML_TYPE_F32);
case GGML_OP_CPY: case GGML_OP_CPY:
case GGML_OP_DUP: case GGML_OP_DUP:
case GGML_OP_CONT: case GGML_OP_CONT: