tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-07-28 21:23:55 -04:00
Code Issues Packages Projects Releases Wiki Activity

metal : add f16 support

Browse Source
This commit is contained in:
Georgi Gerganov
2023-06-06 20:16:57 +03:00
parent d5b111f53d
commit 44f906e853
3 changed files with 31 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
ggml-metal.metal
View File

@@ -169,6 +169,22 @@ kernel void kernel_diag_mask_inf(
}
}
kernel void kernel_get_rows_f16(
device const void * src0,
device const int * src1,
device float * dst,
constant int64_t & ne00,
constant uint64_t & nb01,
constant uint64_t & nb1,
uint tpig[[thread_position_in_grid]]) {
const int i = tpig;
const int r = ((device int32_t *) src1)[i];
for (int j = 0; j < ne00; j++) {
dst[i*nb1 + j] = ((device half *) ((device char *) src0 + r*nb01))[j];
}
}
kernel void kernel_get_rows_q4_0(
device const void * src0,
device const int * src1,