CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-31 08:51:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

Remove Q4_3 which is no better than Q5 (#1218)

Browse Source
This commit is contained in:
Stephan Walter
2023-04-28 23:10:43 +00:00
committed by GitHub
parent 7f15c5c477
commit 36d19a603b
11 changed files with 21 additions and 359 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
ggml-opencl-dequant.cl
View File

@@ -60,25 +60,4 @@ __kernel void dequantize_row_q4_2(__global struct block_q4_2* blocks, __global f
result[index + 1] = ((vi >> 4) - 8)*d;
}
struct block_q4_3
{
ushort d;
ushort m;
uchar qs[8];
};
__kernel void dequantize_row_q4_3(__global struct block_q4_3* blocks, __global float* result) {
const uint i = get_global_id(0) / 16;
const uint l = get_local_id(0);
const float d = vload_half(0, (__global half*) &(blocks[i].d));
const float m = vload_half(0, (__global half*) &(blocks[i].m));
const uchar vi = blocks[i].qs[l];
const uint index = i*16 + l*2;
result[index + 0] = (vi & 0xf) * d + m;
result[index + 1] = (vi >> 4) * d + m;
}
);