[2/n]decouple quantization implementation from vLLM dependency (#8112)

Co-authored-by: walker-ai <yiyun.wyt@antgroup.com>
Co-authored-by: leoneo <1320612015@qq.com>
This commit is contained in:
Peng Zhang
2025-08-14 18:19:03 +08:00
committed by GitHub
parent 4dbf43601d
commit 5aa1ebd242
32 changed files with 6506 additions and 202 deletions

View File

@@ -254,13 +254,15 @@ set(SOURCES
"csrc/gemm/per_token_quant_fp8.cu"
"csrc/gemm/qserve_w4a8_per_chn_gemm.cu"
"csrc/gemm/qserve_w4a8_per_group_gemm.cu"
"csrc/gemm/marlin/gptq_marlin.cu"
"csrc/gemm/marlin/gptq_marlin_repack.cu"
"csrc/gemm/marlin/awq_marlin_repack.cu"
"csrc/gemm/gptq/gptq_kernel.cu"
"csrc/grammar/apply_token_bitmask_inplace_cuda.cu"
"csrc/moe/cutlass_moe/w4a8/scaled_mm_entry.cu"
"csrc/moe/cutlass_moe/w4a8/w4a8_moe_data.cu"
"csrc/moe/cutlass_moe/w4a8/w4a8_grouped_mm_c3x.cu"
"csrc/moe/marlin_moe_wna16/ops.cu"
"csrc/moe/marlin_moe_wna16/gptq_marlin_repack.cu"
"csrc/moe/marlin_moe_wna16/awq_marlin_repack.cu"
"csrc/moe/marlin_moe_wna16/kernel_bf16_ku4.cu"
"csrc/moe/marlin_moe_wna16/kernel_bf16_ku4b8.cu"
"csrc/moe/marlin_moe_wna16/kernel_bf16_ku8b128.cu"