Reland [1/2] Optimizations and refactors about quant kernel (#10312)

Co-authored-by: Yineng Zhang <me@zhyncs.com>
2025-10-11 15:59:03 +08:00
parent 129d299278
commit 21337b22b9
13 changed files with 1065 additions and 178 deletions
--- a/sgl-kernel/python/sgl_kernel/init.py
+++ b/sgl-kernel/python/sgl_kernel/init.py
@@ -263,8 +263,7 @@ from sgl_kernel.gemm import (
    scaled_fp4_grouped_quant,
    scaled_fp4_quant,
    sgl_per_tensor_quant_fp8,
-    sgl_per_token_group_quant_fp8,
-    sgl_per_token_group_quant_int8,
+    sgl_per_token_group_quant_8bit,
    sgl_per_token_quant_fp8,
    shuffle_rows,
    silu_and_mul_scaled_fp4_grouped_quant,