happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

metal : improve concurrency (#19555)

This commit is contained in:
Georgi Gerganov 2026-02-13 07:35:57 +02:00 committed by GitHub
parent 490eb96b88
commit 0644baefde
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 12 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
ggml/src/ggml-metal/ggml-metal-common.cpp
View File

@ -264,15 +264,25 @@ static std::vector<int> ggml_metal_graph_optimize_reorder(const std::vector<node
case GGML_OP_NORM:
case GGML_OP_RMS_NORM:
case GGML_OP_GROUP_NORM:
case GGML_OP_L2_NORM:
case GGML_OP_SUM_ROWS:
case GGML_OP_SSM_CONV:
case GGML_OP_SSM_SCAN:
case GGML_OP_CLAMP:
case GGML_OP_TRI:
case GGML_OP_DIAG:
case GGML_OP_MUL:
case GGML_OP_ADD:
case GGML_OP_DIV:
case GGML_OP_GLU:
case GGML_OP_SCALE:
case GGML_OP_UNARY:
case GGML_OP_GET_ROWS:
case GGML_OP_CPY:
case GGML_OP_SET_ROWS:
case GGML_OP_SET:
case GGML_OP_CPY:
case GGML_OP_CONT:
case GGML_OP_REPEAT:
return true;
default:
return ggml_op_is_empty(op);
@ -312,7 +322,7 @@ static std::vector<int> ggml_metal_graph_optimize_reorder(const std::vector<node
h_add(mrs1, node0);
// that many nodes forward to search for a concurrent node
constexpr int N_FORWARD = 8;
constexpr int N_FORWARD = 64;
for (int i1 = i0 + 1; i1 < i0 + N_FORWARD && i1 < n; i1++) {
if (used[i1]) {