CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-01 09:01:57 +00:00
Code Issues Packages Projects Releases Wiki Activity

Vulkan: Add DP4A MMQ and Q8_1 quantization shader (#12135)

Browse Source 
* Vulkan: Add DP4A MMQ and Q8_1 quantization shader

* Add q4_0 x q8_1 matrix matrix multiplication support

* Vulkan: Add int8 coopmat MMQ support

* Vulkan: Add q4_1, q5_0 and q5_1 quants, improve integer dot code

* Add GL_EXT_integer_dot_product check

* Remove ggml changes, fix mmq pipeline picker

* Remove ggml changes, restore Intel coopmat behaviour

* Fix glsl compile attempt when integer vec dot is not supported

* Remove redundant code, use non-saturating integer dot, enable all matmul sizes for mmq

* Remove redundant comment

* Fix integer dot check

* Fix compile issue with unsupported int dot glslc

* Update Windows build Vulkan SDK version
This commit is contained in:
0cc4m
2025-03-31 14:37:01 +02:00
committed by GitHub
parent 1790e73157
commit a8a1f33567
10 changed files with 1146 additions and 95 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
ggml/src/ggml-vulkan/vulkan-shaders/mul_mm.comp
View File

@@ -212,7 +212,7 @@ void main() {
#else
ACC_TYPE sums[WMITER * TM * WNITER * TN];
FLOAT_TYPE cache_a[WMITER * TM];
FLOAT_TYPE cache_b[WNITER * TN];
FLOAT_TYPE cache_b[TN];
[[unroll]] for (uint i = 0; i < WMITER*TM*WNITER*TN; i++) {
sums[i] = ACC_TYPE(0.0f);
@@ -744,16 +744,14 @@ void main() {
}
[[unroll]] for (uint wsic = 0; wsic < WNITER; wsic++) {
[[unroll]] for (uint j = 0; j < TN; j++) {
cache_b[wsic * TN + j] = buf_b[(warp_c * WN + wsic * WSUBN + tiwc * TN + j) * SHMEM_STRIDE + i];
cache_b[j] = buf_b[(warp_c * WN + wsic * WSUBN + tiwc * TN + j) * SHMEM_STRIDE + i];
}
}
[[unroll]] for (uint wsic = 0; wsic < WNITER; wsic++) {
[[unroll]] for (uint wsir = 0; wsir < WMITER; wsir++) {
[[unroll]] for (uint cc = 0; cc < TN; cc++) {
[[unroll]] for (uint cr = 0; cr < TM; cr++) {
const uint sums_idx = (wsic * TN + cc) * (WMITER * TM) + wsir * TM + cr;
sums[sums_idx] = fma(ACC_TYPE(cache_a[wsir * TM + cr]), ACC_TYPE(cache_b[wsic * TN + cc]), sums[sums_idx]);
sums[sums_idx] = fma(ACC_TYPE(cache_a[wsir * TM + cr]), ACC_TYPE(cache_b[cc]), sums[sums_idx]);
}
}
}