xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-03-28 20:49:11 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix output zeroing race condition in GPTQ GEMM kernels

Browse Source 
Signed-off-by: Andreas Karatzas <akaratza@amd.com>
This commit is contained in:
Andreas Karatzas 2025-12-15 20:25:30 +00:00
parent 17fec3af09
commit 7326fef568
No known key found for this signature in database
GPG Key ID: 74A33CBB22F03519
1 changed files with 5 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
csrc/quantization/gptq/q_gemm.cu
View File

@ -233,10 +233,10 @@ __global__ void gemm_half_q_half_gptq_4bit_kernel(
// Zero output
if (n >= size_n) return;
if (blockIdx.z == 0) {
for (int m = 0; m < m_count; m++)
*((uint64_t*)c_.item_ptr(offset_m + m, n)) = 0;
}
// if (blockIdx.z == 0) {
// for (int m = 0; m < m_count; m++)
// *((uint64_t*)c_.item_ptr(offset_m + m, n)) = 0;
// }
__syncthreads();
@ -1857,7 +1857,7 @@ torch::Tensor gptq_gemm(torch::Tensor a, torch::Tensor b_q_weight,
bool use_exllama, bool use_v2_format, int64_t bit) {
const at::cuda::OptionalCUDAGuard device_guard(device_of(a));
auto options = torch::TensorOptions().dtype(a.dtype()).device(a.device());
at::Tensor c = torch::empty({a.size(0), b_q_weight.size(1)}, options);
at::Tensor c = torch::zeros({a.size(0), b_q_weight.size(1)}, options);
at::Tensor temp_dq = torch::empty(
{b_q_weight.size(0) * 32 / bit, b_q_weight.size(1)}, options);