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

metal: faster kernel_scale via float4

Browse source
This commit is contained in:
Iwan Kawrakow 2023-09-07 16:16:28 +02:00
parent 9a9010609b
commit 7c8c6ce085
2 changed files with 3 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml-metal.m
View file

@ -750,7 +750,7 @@ void ggml_metal_graph_compute(
[encoder setBuffer:id_dst offset:offs_dst atIndex:1];
[encoder setBytes:&scale length:sizeof(scale) atIndex:2];
const int64_t n = ggml_nelements(dst);
const int64_t n = ggml_nelements(dst)/4;
[encoder dispatchThreadgroups:MTLSizeMake(n, 1, 1) threadsPerThreadgroup:MTLSizeMake(1, 1, 1)];
} break;

4
ggml-metal.metal
View file

@ -63,8 +63,8 @@ kernel void kernel_mul_row(
}
kernel void kernel_scale(
device const float * src0,
device float * dst,
device const float4 * src0,
device float4 * dst,
constant float & scale,
uint tpig[[thread_position_in_grid]]) {
dst[tpig] = src0[tpig] * scale;