biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[2/n] Migrate per_token_group_quant to torch stable ABI (#36058)

Browse Source 
Signed-off-by: Mikayla Gawarecki <mikaylagawarecki@gmail.com>
This commit is contained in:
mikaylagawarecki
2026-03-25 13:15:13 -04:00
committed by GitHub
parent 1ac2ef2e53
commit bf4cc9ed2d
22 changed files with 207 additions and 133 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
csrc/ops.h
View File

@@ -306,25 +306,6 @@ void silu_and_mul_scaled_fp4_experts_quant(
torch::Tensor const& input_offset_by_experts,
torch::Tensor const& output_scale_offset_by_experts);
void per_token_group_quant_fp8(const torch::Tensor& input,
torch::Tensor& output_q, torch::Tensor& output_s,
int64_t group_size, double eps, double fp8_min,
double fp8_max, bool scale_ue8m0,
bool dummy_is_scale_transposed,
bool dummy_is_tma_aligned);
void per_token_group_quant_int8(const torch::Tensor& input,
torch::Tensor& output_q,
torch::Tensor& output_s, int64_t group_size,
double eps, double int8_min, double int8_max);
// Fused activation quantisation + DeepGEMM-compatible UE8M0-packed scales.
void per_token_group_quant_8bit_packed(const torch::Tensor& input,
torch::Tensor& output_q,
torch::Tensor& output_s_packed,
int64_t group_size, double eps,
double min_8bit, double max_8bit);
#endif
void static_scaled_int8_quant(torch::Tensor& out, torch::Tensor const& input,