We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
There was an error while loading. Please reload this page.
1 parent 80acb7b commit 106964eCopy full SHA for 106964e
ggml/src/ggml-metal/ggml-metal.m
@@ -1927,7 +1927,7 @@ static void ggml_metal_encode_node(
1927
1928
// find the break-even point where the matrix-matrix kernel becomes more efficient compared
1929
// to the matrix-vector kernel
1930
- int ne11_mm_min = 1;
+ int ne11_mm_min = 4;
1931
1932
#if 0
1933
// the numbers below are measured on M2 Ultra for 7B and 13B models
0 commit comments