mozempk/ggml-org-llama-cpp-mirror
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-10-11 07:20:33 +02:00
Code Issues Packages Projects Releases Wiki Activity

remove unneeded scratch buffer from FP8 cpu cpy/dup path

Browse Source
This commit is contained in:
Oliver Simons
2026-10-09 21:06:18 +02:00
parent 5ed32fd09f
commit b51d08c82c
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-cpu/ggml-cpu.c
+1 -1
View File
@@ -2858,7 +2858,7 @@ struct ggml_cplan ggml_graph_plan(
case GGML_OP_CPY: case GGML_OP_CPY:
case GGML_OP_DUP: case GGML_OP_DUP:
{ {
if (ggml_is_quantized(node->type) || node->type == GGML_TYPE_F8_E4M3 || if (ggml_is_quantized(node->type) ||
// F16 -> BF16 and BF16 -> F16 copies go through intermediate F32 // F16 -> BF16 and BF16 -> F16 copies go through intermediate F32
(node->src[0]->type == GGML_TYPE_F16 && node->src[1] && node->src[1]->type == GGML_TYPE_BF16) || (node->src[0]->type == GGML_TYPE_F16 && node->src[1] && node->src[1]->type == GGML_TYPE_BF16) ||
(node->src[0]->type == GGML_TYPE_BF16 && node->src[1] && node->src[1]->type == GGML_TYPE_F16) || (node->src[0]->type == GGML_TYPE_BF16 && node->src[1] && node->src[1]->type == GGML_TYPE_F16) ||