Support FP4 gemm (1/2) (#3899)

2025-03-24 19:50:23 -07:00
parent 22c3702e1e
commit e9f8e42318
11 changed files with 1245 additions and 5 deletions
--- a/sgl-kernel/python/sgl_kernel/init.py
+++ b/sgl-kernel/python/sgl_kernel/init.py
@@ -26,9 +26,11 @@ from sgl_kernel.gemm import (
    awq_dequantize,
    bmm_fp8,
    cublas_grouped_gemm,
+    cutlass_scaled_fp4_mm,
    fp8_blockwise_scaled_mm,
    fp8_scaled_mm,
    int8_scaled_mm,
+    scaled_fp4_quant,
    sgl_per_tensor_quant_fp8,
    sgl_per_token_group_quant_fp8,
    sgl_per_token_group_quant_int8,