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

fix undefined memory reads for small matrices

Browse source
This commit is contained in:
Johannes Gäßler 2024-06-20 14:05:50 +02:00
parent da1db13d6a
commit 141d0810ec
1 changed files with 7 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

11
ggml-cuda/mmq.cuh
View file

@ -2071,15 +2071,18 @@ static __global__ void mul_mat_q_stream_k_fixup(
const int bidx_stop = (blockIdx.y*nty + blockIdx.x + 1) * block_num_mmq / (gridDim.y*gridDim.x) + 1;
for (int bidx = bidx_start; bidx < bidx_stop; ++bidx) {
const int64_t kb_stop = GGML_PAD((int64_t)(bidx + 1)*blocks_per_ne00*ntx*nty / block_num_mmq, blocks_per_warp);
const int64_t kbc = GGML_PAD((int64_t) bidx *blocks_per_ne00*ntx*nty / block_num_mmq, blocks_per_warp);
const int64_t kbc_stop = GGML_PAD((int64_t)(bidx + 1)*blocks_per_ne00*ntx*nty / block_num_mmq, blocks_per_warp);
if (kb_stop % blocks_per_ne00 == 0) {
// Skip fixup tile if the MMQ CUDA block never wrote anything to it:
if (kbc == kbc_stop || kbc_stop % blocks_per_ne00 == 0) {
continue;
}
const int jt = kb_stop / (blocks_per_ne00*nty);
const int it = (kb_stop - jt*(blocks_per_ne00*nty)) / blocks_per_ne00;
const int jt = kbc_stop / (blocks_per_ne00*nty);
const int it = (kbc_stop - jt*(blocks_per_ne00*nty)) / blocks_per_ne00;
// Skip fixup tile if it's unrelated to the output tile assigned to this CUDA block:
if (it != blockIdx.x || jt != blockIdx.y) {
continue;
}