[Kernel] Set sgl_per_token_group_quant_8bit_v2 as default choice (#22467)

docs/references/environment_variables.md

@@ -122,7 +122,6 @@ SGLang supports various environment variables that can be used to configure its
 | Environment Variable | Description | Default Value |
 | --- | --- | --- |
 | `SGLANG_INT4_WEIGHT` | Enable INT4 weight quantization | `false` |
-| `SGLANG_PER_TOKEN_GROUP_QUANT_8BIT_V2` | Apply per token group quantization kernel with fused silu and mul and masked m | `false` |
 | `SGLANG_FORCE_FP8_MARLIN` | Force using FP8 MARLIN kernels even if other FP8 kernels are available | `false` |
 | `SGLANG_NVFP4_CKPT_FP8_GEMM_IN_ATTN` | Quantize q_b_proj from BF16 to FP8 when launching DeepSeek NVFP4 checkpoint | `false` |
 | `SGLANG_MOE_NVFP4_DISPATCH` | Use nvfp4 for moe dispatch (on flashinfer_cutlass or flashinfer_cutedsl moe runner backend) | `"false"` |

python/sglang/srt/environ.py

@@ -3,7 +3,7 @@ import subprocess
 import warnings
 from contextlib import ExitStack, contextmanager
 from enum import IntEnum
-from typing import Any
+from typing import Any, Optional
 
 
 @contextmanager
@@ -341,7 +341,6 @@ class Envs:
     SGLANG_FORCE_FP8_MARLIN = EnvBool(False)
     SGLANG_MOE_NVFP4_DISPATCH = EnvBool(False)
     SGLANG_NVFP4_CKPT_FP8_GEMM_IN_ATTN = EnvBool(False)
-    SGLANG_PER_TOKEN_GROUP_QUANT_8BIT_V2 = EnvBool(False)
     SGLANG_NVFP4_CKPT_FP8_NEXTN_MOE = EnvBool(False)
     SGLANG_QUANT_ALLOW_DOWNCASTING = EnvBool(False)
     SGLANG_FP8_IGNORED_LAYERS = EnvStr("")
@@ -552,12 +551,15 @@ envs = Envs()
 EnvField._allow_set_name = False
 
 
-def _print_deprecated_env(new_name: str, old_name: str):
+def _print_deprecated_env(old_name: str, new_name: Optional[str] = None):
     if old_name in os.environ:
-        warnings.warn(
-            f"Environment variable {old_name} will be deprecated, please use {new_name} instead"
-        )
-        os.environ[new_name] = os.environ[old_name]
+        if new_name is None:
+            warnings.warn(f"Environment variable {old_name} has been deprecated.")
+        else:
+            warnings.warn(
+                f"Environment variable {old_name} will be deprecated, please use {new_name} instead"
+            )
+            os.environ[new_name] = os.environ[old_name]
 
 
 def _warn_deprecated_env_to_cli_flag(env_name: str, suggestion: str):
@@ -570,14 +572,15 @@ def _warn_deprecated_env_to_cli_flag(env_name: str, suggestion: str):
 
 
 def _convert_SGL_to_SGLANG():
-    _print_deprecated_env("SGLANG_LOG_GC", "SGLANG_GC_LOG")
+    _print_deprecated_env("SGLANG_GC_LOG", "SGLANG_LOG_GC")
     _print_deprecated_env(
-        "SGLANG_MOE_NVFP4_DISPATCH", "SGLANG_CUTEDSL_MOE_NVFP4_DISPATCH"
+        "SGLANG_CUTEDSL_MOE_NVFP4_DISPATCH", "SGLANG_MOE_NVFP4_DISPATCH"
     )
     _print_deprecated_env(
-        "SGLANG_ENABLE_TP_MEMORY_INBALANCE_CHECK",
         "SGL_DISABLE_TP_MEMORY_INBALANCE_CHECK",
+        "SGLANG_ENABLE_TP_MEMORY_INBALANCE_CHECK",
     )
+    _print_deprecated_env("SGLANG_PER_TOKEN_GROUP_QUANT_8BIT_V2")
     _deprecated_ms_to_s = {
         "SGLANG_QUEUED_TIMEOUT_MS": "SGLANG_REQ_WAITING_TIMEOUT",
         "SGLANG_FORWARD_TIMEOUT_MS": "SGLANG_REQ_RUNNING_TIMEOUT",

python/sglang/srt/layers/quantization/fp8_kernel.py

@@ -324,10 +324,6 @@ def _per_token_group_quant_8bit_raw(
     return x_q, x_s
 
 
-# backward compatibility
-per_token_group_quant_fp8 = _per_token_group_quant_8bit_raw
-
-
 def _per_token_group_quant_8bit_fuse_silu_and_mul(
     x: torch.Tensor,
     group_size: int,
@@ -507,6 +503,11 @@ def sglang_per_token_group_quant_fp8(
         scale_ue8m0=scale_ue8m0,
     )
 
+    # Enable v2 kernel by default on supported group sizes
+    _V2_KERNEL_SUPPORTED_GROUP_SIZES = [16, 32, 64, 128]
+    if enable_v2 is None:
+        enable_v2 = group_size in _V2_KERNEL_SUPPORTED_GROUP_SIZES
+
     if x.shape[0] > 0:
         # Temporary
         if enable_sgl_per_token_group_quant_8bit:
@@ -606,6 +607,12 @@ def sglang_per_token_quant_fp8(
     return x_q, x_s
 
 
+if _is_cuda:
+    per_token_group_quant_fp8 = sglang_per_token_group_quant_fp8
+else:
+    per_token_group_quant_fp8 = _per_token_group_quant_8bit_raw
+
+
 @triton.jit
 def _static_quant_fp8(
     # Pointers to inputs and output

sgl-kernel/python/sgl_kernel/gemm.py

@@ -109,10 +109,9 @@ def sgl_per_token_group_quant_8bit(
     masked_m: Optional[torch.Tensor] = None,
     enable_v2: Optional[bool] = None,
 ) -> None:
+    _V2_KERNEL_SUPPORTED_GROUP_SIZES = [16, 32, 64, 128]
     if enable_v2 is None:
-        from sglang.srt.utils import get_bool_env_var
-
-        enable_v2 = get_bool_env_var("SGLANG_PER_TOKEN_GROUP_QUANT_8BIT_V2")
+        enable_v2 = group_size in _V2_KERNEL_SUPPORTED_GROUP_SIZES
 
     if enable_v2:
         return torch.ops.sgl_kernel.sgl_per_token_group_quant_8bit_v2.default(

test/registered/quant/test_fp8_kernel.py

@@ -96,7 +96,9 @@ class TestPerTokenGroupQuantFP8(TestFP8Base):
         A, A_quant_gt, scale_gt = self._make_A(
             M=self.M, K=self.K, group_size=self.group_size, out_dtype=self.quant_type
         )
-        A_quant, scale = per_token_group_quant_fp8(x=A, group_size=self.group_size)
+        A_quant, scale = per_token_group_quant_fp8(
+            x=A.to(torch.bfloat16), group_size=self.group_size
+        )
         torch.testing.assert_close(scale, scale_gt)
         diff = (A_quant.to(torch.float16) - A_quant_gt.to(torch.float16)).abs()
         diff_count = (diff > 1e-5).count_nonzero()