[Misc]Register and refactor some environs for dpsk-fp4 and DeepEp

docs/references/environment_variables.md

@@ -55,7 +55,9 @@ SGLang supports various environment variables that can be used to configure its
 | Environment Variable | Description | Default Value |
 | --- | --- | --- |
 | `SGLANG_DEEPEP_BF16_DISPATCH` | Use Bfloat16 for dispatch | `"false"` |
-| `SGLANG_MOE_NVFP4_DISPATCH` | Use nvfp4 for moe dispatch | `"false"` |
+| `SGLANG_MOE_NVFP4_DISPATCH` | Use nvfp4 for moe dispatch (on flashinfer_cutlass or flashinfer_cutedsl moe runner backend) | `"false"` |
+| `SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK` | The maximum number of dispatched tokens on each GPU | `"128"` |
+| `SGLANG_DEEPEP_LL_COMBINE_SEND_NUM_SMS` | Number of SMs used for DeepEP combine when single batch overlap is enabled | `"32"` |
 
 ## Memory Management
 
@@ -65,6 +67,8 @@ SGLang supports various environment variables that can be used to configure its
 | `SGLANG_CLIP_MAX_NEW_TOKENS_ESTIMATION` | Clip max new tokens estimation for memory planning | `4096` |
 | `SGLANG_DETOKENIZER_MAX_STATES` | Maximum states for detokenizer | Default value based on system |
 | `SGL_DISABLE_TP_MEMORY_INBALANCE_CHECK` | Disable checks for memory imbalance across Tensor Parallel ranks | Not set (defaults to enabled check) |
+| `SGLANG_NVFP4_CKPT_FP8_GEMM_IN_ATTN` | Quantize q_b_proj from BF16 to FP8 when launching DeepSeek NVFP4 checkpoint | `false` |
+| `SGLANG_PER_TOKEN_GROUP_QUANT_8BIT_V2` | Apply per token group quantization kernel with fused silu and mul and masked m | `false` |
 
 ## Model-Specific Options
 

python/sglang/srt/batch_overlap/single_batch_overlap.py

@@ -19,9 +19,10 @@
 
 import torch
 
+from sglang.srt.environ import envs
 from sglang.srt.layers.moe import get_moe_runner_backend
 from sglang.srt.layers.moe.utils import is_sbo_enabled
-from sglang.srt.utils import get_int_env_var, is_blackwell
+from sglang.srt.utils import is_blackwell
 
 
 class SboFlags:
@@ -87,9 +88,11 @@ def compute_overlap_args(dispatch_output, alt_stream):
     total_num_sms = torch.cuda.get_device_properties(
         device="cuda"
     ).multi_processor_count
-    communicate_num_sms = get_int_env_var(
-        "SGLANG_DEEPEP_LL_COMBINE_SEND_NUM_SMS", 32 if is_blackwell() else 3
-    )
+
+    if envs.SGLANG_DEEPEP_LL_COMBINE_SEND_NUM_SMS.is_set():
+        communicate_num_sms = envs.SGLANG_DEEPEP_LL_COMBINE_SEND_NUM_SMS.get()
+    else:
+        communicate_num_sms = 32 if is_blackwell() else 3
     compute_num_sms = total_num_sms - communicate_num_sms
 
     assert alt_stream is not None

python/sglang/srt/environ.py

@@ -241,6 +241,8 @@ class Envs:
     SGLANG_USE_DYNAMIC_MXFP4_LINEAR = EnvBool(False)
     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)
 
     # Flashinfer
     SGLANG_IS_FLASHINFER_AVAILABLE = EnvBool(True)
@@ -276,6 +278,11 @@ class Envs:
     SGLANG_DG_USE_NVRTC = EnvBool(False)
     SGLANG_USE_DEEPGEMM_BMM = EnvBool(False)
 
+    # DeepEP
+    SGLANG_DEEPEP_BF16_DISPATCH = EnvBool(False)
+    SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK = EnvInt(128)
+    SGLANG_DEEPEP_LL_COMBINE_SEND_NUM_SMS = EnvInt(32)
+
     # sgl-kernel
     SGLANG_SKIP_SGL_KERNEL_VERSION_CHECK = EnvBool(False)
 
@@ -365,6 +372,9 @@ def _convert_SGL_to_SGLANG():
     _print_deprecated_env(
         "SGLANG_ENABLE_FLASHINFER_FP8_GEMM", "SGLANG_ENABLE_FLASHINFER_GEMM"
     )
+    _print_deprecated_env(
+        "SGLANG_MOE_NVFP4_DISPATCH", "SGLANG_CUTEDSL_MOE_NVFP4_DISPATCH"
+    )
 
     for key, value in os.environ.items():
         if key.startswith("SGL_"):

python/sglang/srt/layers/moe/ep_moe/layer.py

@@ -5,6 +5,7 @@
 
 import torch
 
+from sglang.srt.environ import envs
 from sglang.srt.layers import deep_gemm_wrapper
 from sglang.srt.layers.moe import (
     get_deepep_mode,
@@ -303,8 +304,8 @@ def forward_cutlass_w4afp8_masked(
     ):
         assert self.moe_runner_config.activation == "silu"
         assert isinstance(self.quant_method, W4AFp8MoEMethod)
-        assert get_bool_env_var(
-            "SGLANG_DEEPEP_BF16_DISPATCH"
+        assert (
+            envs.SGLANG_DEEPEP_BF16_DISPATCH.get()
         ), "W4AFP8 does not support FP8 dispatch; please set SGLANG_DEEPEP_BF16_DISPATCH=1."
         return self.quant_method.apply_deepep_ll(
             layer=self,

python/sglang/srt/layers/moe/token_dispatcher/deepep.py

@@ -5,6 +5,7 @@
 from dataclasses import dataclass
 from typing import TYPE_CHECKING, List, NamedTuple, Optional, Tuple, Union
 
+from sglang.srt.environ import envs
 from sglang.srt.eplb.expert_distribution import get_global_expert_distribution_recorder
 from sglang.srt.layers import deep_gemm_wrapper
 from sglang.srt.layers.dp_attention import get_is_extend_in_batch
@@ -26,7 +27,6 @@
 )
 from sglang.srt.utils import (
     get_bool_env_var,
-    get_int_env_var,
     is_blackwell,
     is_hip,
     is_npu,
@@ -317,8 +317,8 @@ def __init__(
 
         self.params_bytes = 2
         # A large value will lead to large memory occupation, thus users should change it accordingly
-        self.num_max_dispatch_tokens_per_rank = get_int_env_var(
-            "SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK", 128
+        self.num_max_dispatch_tokens_per_rank = (
+            envs.SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK.get()
         )
         # DeepEP internode_ll dispatch uses FINISHED_SUM_TAG=1024
         # and the logic requires num-tokens-sent-from-one-rank-to-another-rank less than it
@@ -387,7 +387,7 @@ def dispatch_a(
         if (
             deep_gemm_wrapper.ENABLE_JIT_DEEPGEMM
             and not get_moe_runner_backend().is_cutlass()
-            and not get_bool_env_var("SGLANG_DEEPEP_BF16_DISPATCH")
+            and not envs.SGLANG_DEEPEP_BF16_DISPATCH.get()
         ):
             # TODO hard code 128 block quant,use fp8 communication
             hidden_states = sglang_per_token_group_quant_fp8(
@@ -609,7 +609,7 @@ def _dispatch_core(
         input_global_scale = self.quant_config.get("input_global_scale", None)
         if input_global_scale is not None:
             use_nvfp4 = True
-        elif not get_bool_env_var("SGLANG_DEEPEP_BF16_DISPATCH"):
+        elif not envs.SGLANG_DEEPEP_BF16_DISPATCH.get():
             use_fp8 = True
 
         buffer = self._get_buffer()

python/sglang/srt/layers/moe/token_dispatcher/fuseep.py

@@ -5,6 +5,7 @@
 
 import torch
 
+from sglang.srt.environ import envs
 from sglang.srt.layers.moe.token_dispatcher.base import (
     BaseDispatcher,
     CombineInput,
@@ -15,7 +16,6 @@
 from sglang.srt.layers.moe.token_dispatcher.deepep import DeepEPBuffer
 from sglang.srt.layers.moe.topk import TopKOutput
 from sglang.srt.layers.moe.utils import DeepEPMode
-from sglang.srt.utils import get_int_env_var
 
 logger = logging.getLogger(__name__)
 
@@ -62,8 +62,8 @@ def __init__(
         self.deepep_mode = deepep_mode
 
         self.params_bytes = 2
-        self.num_max_dispatch_tokens_per_rank = get_int_env_var(
-            "SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK", 128
+        self.num_max_dispatch_tokens_per_rank = (
+            envs.SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK.get()
         )
 
     def dispatch(

python/sglang/srt/models/deepseek_v2.py

@@ -48,6 +48,7 @@
     get_tensor_model_parallel_world_size,
     tensor_model_parallel_all_reduce,
 )
+from sglang.srt.environ import envs
 from sglang.srt.eplb.expert_distribution import get_global_expert_distribution_recorder
 from sglang.srt.eplb.expert_location import ModelConfigForExpertLocation
 from sglang.srt.eplb.expert_location_dispatch import ExpertLocationDispatchInfo
@@ -2655,7 +2656,7 @@ def _concat_and_cast_mha_k(self, k_nope, k_pe, forward_batch):
 
     @staticmethod
     def _get_q_b_proj_quant_config(quant_config):
-        if get_bool_env_var("SGLANG_NVFP4_CKPT_FP8_GEMM_IN_ATTN"):
+        if envs.SGLANG_NVFP4_CKPT_FP8_GEMM_IN_ATTN.get():
             # refer to real DeepSeek V3 quant config
             return Fp8Config(
                 is_checkpoint_fp8_serialized=True,
@@ -3591,7 +3592,7 @@ def load_weights(self, weights: Iterable[Tuple[str, torch.Tensor]], is_nextn=Fal
             else:
                 raise ValueError("num_nextn_predict_layers is not in the config")
 
-        if get_bool_env_var("SGLANG_NVFP4_CKPT_FP8_GEMM_IN_ATTN"):
+        if envs.SGLANG_NVFP4_CKPT_FP8_GEMM_IN_ATTN.get():
             weights = self._quant_attn_to_fp8_ue8m0(weights, is_nextn=is_nextn)
         if is_nextn and enable_nextn_moe_bf16_cast_to_fp8(self.quant_config):
             weights = self._quant_nextn_moe_to_fp8_ue8m0(