EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[CPU] Fix build issue (#6419)

Browse Source
This commit is contained in:
blzheng
2025-05-22 02:17:10 +08:00
committed by GitHub
parent d4c038daed
commit cfe48c5902
14 changed files with 157 additions and 143 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
sgl-kernel/csrc/cpu/moe.cpp
View File

@@ -946,10 +946,10 @@ at::Tensor fused_experts_cpu(
at::Tensor& topk_ids,
bool inplace,
bool use_int8_w8a8,
std::optional<at::Tensor>& w1_scale,
std::optional<at::Tensor>& w2_scale,
std::optional<at::Tensor>& a1_scale,
std::optional<at::Tensor>& a2_scale,
const std::optional<at::Tensor>& w1_scale,
const std::optional<at::Tensor>& w2_scale,
const std::optional<at::Tensor>& a1_scale,
const std::optional<at::Tensor>& a2_scale,
bool is_vnni) {
RECORD_FUNCTION(
"sgl-kernel::fused_experts_cpu", std::vector<c10::IValue>({hidden_states, w1, w2, topk_weights, topk_ids}));
@@ -1138,11 +1138,11 @@ at::Tensor shared_expert_cpu(
bool inplace,
bool use_int8_w8a8,
bool use_fp8_w8a16,
std::optional<at::Tensor>& w1_scale,
std::optional<at::Tensor>& w2_scale,
std::optional<std::vector<int64_t>> block_size,
std::optional<at::Tensor>& a1_scale,
std::optional<at::Tensor>& a2_scale,
const std::optional<at::Tensor>& w1_scale,
const std::optional<at::Tensor>& w2_scale,
const std::optional<std::vector<int64_t>> block_size,
const std::optional<at::Tensor>& a1_scale,
const std::optional<at::Tensor>& a2_scale,
bool is_vnni) {
RECORD_FUNCTION("sgl-kernel::shared_expert_cpu", std::vector<c10::IValue>({hidden_states, w1, w2}));