EngineX-Ascend/enginex-ascend-910-llama.cpp
10
0
Fork 0
Code Issues Pull Requests Actions 4 Projects Releases Wiki Activity

Vulkan IQ4_NL Support (#8613)

Browse Source 
* Fix Vulkan matmul tests compile errors

* Add Vulkan IQ4_NL support

* Fix Vulkan DeepSeek-Coder-V2-Lite MoE support
This commit is contained in:
0cc4m
2024-07-23 10:56:49 +02:00
committed by GitHub
parent 46e47417aa
commit 751fcfc6c3
7 changed files with 222 additions and 193 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
ggml/src/vulkan-shaders/mul_mm.comp
View File

@@ -71,7 +71,7 @@ shared FLOAT_TYPE buf_a[BM * (BK+1)];
shared FLOAT_TYPE buf_b[BN * (BK+1)];
#ifdef MUL_MAT_ID
shared u16vec2 row_ids[2048];
shared u16vec2 row_ids[3072];
#endif
void main() {
@@ -380,6 +380,19 @@ void main() {
buf_a[buf_idx ] = FLOAT_TYPE(dscale * float(int8_t(((data_a[ib].ql[qsi ] >> (b * 4)) & 0xF) | (((data_a[ib].qh[qhi ] >> qhshift) & 3) << 4)) - 32));
buf_a[buf_idx + 1] = FLOAT_TYPE(dscale * float(int8_t(((data_a[ib].ql[qsi + 1] >> (b * 4)) & 0xF) | (((data_a[ib].qh[qhi + 1] >> qhshift) & 3) << 4)) - 32));
#elif defined(DATA_A_IQ4_NL)
const uint idx = pos_a + (loadc_a + l) * p.stride_a / LOAD_VEC_A + loadr_a;
const uint buf_idx = (loadc_a + l) * (BK+1) + loadr_a;
const uint ib = idx / 16;
const uint iqs = idx & 0xF;
const float d = float(data_a[ib].d);
const uint vui = uint(data_a[ib].qs[iqs]);
const vec2 v = vec2(kvalues_iq4nl[vui & 0xF], kvalues_iq4nl[vui >> 4]) * d;
buf_a[buf_idx ] = FLOAT_TYPE(v.x);
buf_a[buf_idx + 16] = FLOAT_TYPE(v.y);
#endif
}
[[unroll]] for (uint l = 0; l < BN; l += loadstride_b) {