From 7f9a3bcefd4bcd990704c46186f3af2b920c689e Mon Sep 17 00:00:00 2001
From: Rohit Pujar Nagraj <rpujarnagraj@nvidia.com>
Date: Thu, 9 Jul 2026 11:15:21 -0700
Subject: [PATCH 1/4] Add Qwen3.5 FP8 GB300 disaggregated multinode SGLang
 benchmarks via Dynamo
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

6 topologies across 1k/1k and 8k/1k: 1P1D TP4 STP plus wide-EP (DEP4
prefill / DEP16 decode) from 1P1D up to 8P1D, mirroring the gb200-fp8
recipe set. Runs qwen3.5 fp8 against srt-slurm main (hash-pinned dynamo
source install needs its cargo/maturin bootstrap) with --no-preflight
for the compute-node-local model path, and overlays a header-based
NVSHMEM resolver for the DeepEP kNumMaxTopK=16 rebuild.

中文：新增 Qwen3.5 FP8 GB300 分离式多节点 SGLang（Dynamo）基准测试。
覆盖 1k/1k 与 8k/1k 共 6 种拓扑：1P1D TP4 STP 及 1P1D 至 8P1D 的宽 EP
（预填充 DEP4 / 解码 DEP16），与 gb200-fp8 配方集一致。fp8 使用
srt-slurm main 分支（哈希固定的 dynamo 源码安装需要其 cargo/maturin
引导），对计算节点本地模型路径使用 --no-preflight，并为 DeepEP
kNumMaxTopK=16 重建覆盖基于头文件的 NVSHMEM 解析脚本。
---
 .../configs/rebuild-deepep.sh                 |  59 +++++++
 .../gb300-fp8/1k1k/1p1d-dep4-dep16.yaml       | 155 +++++++++++++++++
 .../qwen3.5/gb300-fp8/1k1k/1p1d-tp4-tp4.yaml  | 123 +++++++++++++
 .../gb300-fp8/1k1k/2p1d-dep4-dep16.yaml       | 155 +++++++++++++++++
 .../qwen3.5/gb300-fp8/8k1k/1p1d-tp4-tp4.yaml  | 121 +++++++++++++
 .../gb300-fp8/8k1k/4p1d-dep4-dep16.yaml       | 160 +++++++++++++++++
 .../gb300-fp8/8k1k/8p1d-dep4-dep16.yaml       | 161 ++++++++++++++++++
 configs/nvidia-master.yaml                    | 108 ++++++++++++
 perf-changelog.yaml                           |   8 +
 runners/launch_gb300-nv.sh                    |  38 ++++-
 10 files changed, 1082 insertions(+), 6 deletions(-)
 create mode 100644 benchmarks/multi_node/srt-slurm-recipes/configs/rebuild-deepep.sh
 create mode 100644 benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-dep4-dep16.yaml
 create mode 100644 benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-tp4-tp4.yaml
 create mode 100644 benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/2p1d-dep4-dep16.yaml
 create mode 100644 benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/1p1d-tp4-tp4.yaml
 create mode 100644 benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/4p1d-dep4-dep16.yaml
 create mode 100644 benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/8p1d-dep4-dep16.yaml

diff --git a/benchmarks/multi_node/srt-slurm-recipes/configs/rebuild-deepep.sh b/benchmarks/multi_node/srt-slurm-recipes/configs/rebuild-deepep.sh
new file mode 100644
index 0000000000..0be5dba59a
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/configs/rebuild-deepep.sh
@@ -0,0 +1,59 @@
+#!/bin/bash
+# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+set -eux
+
+echo "=== Rebuilding DeepEP with kNumMaxTopK=16 for Qwen3.5 (topk=10) ==="
+
+DEEPEP_SRC="/sgl-workspace/DeepEP"
+
+if [ ! -d "$DEEPEP_SRC" ]; then
+    echo "ERROR: DeepEP source not found at $DEEPEP_SRC (mount via extra_mount)"
+    exit 1
+fi
+
+cd "$DEEPEP_SRC"
+
+# Resolve NVSHMEM by locating its header, not by directory name: a bare
+# `find -name nvshmem -type d` can latch onto header-less source stubs
+# (e.g. tilelang's bundled tvm src/runtime/contrib/nvshmem) and the build
+# then dies with "fatal error: nvshmem.h: No such file or directory".
+NVSHMEM_H=$(find /usr/local /opt /usr/lib /sgl-workspace -name "nvshmem.h" -path "*/include/*" \
+    -not -path "*tilelang*" -not -path "*flashinfer*" -not -path "*DeepEP*" 2>/dev/null | head -1)
+if [ -z "${NVSHMEM_H:-}" ]; then
+    echo "No NVSHMEM headers found in the image; installing nvidia-nvshmem-cu13..."
+    pip install --break-system-packages --quiet nvidia-nvshmem-cu13
+    NVSHMEM_H=$(find /usr/local/lib/python3*/dist-packages/nvidia -name "nvshmem.h" -path "*/include/*" 2>/dev/null | head -1)
+fi
+if [ -z "${NVSHMEM_H:-}" ]; then
+    echo "ERROR: NVSHMEM headers not found in image or via pip" >&2
+    exit 1
+fi
+NVSHMEM_DIR=$(dirname "$(dirname "$NVSHMEM_H")")
+echo "NVSHMEM_DIR=$NVSHMEM_DIR"
+
+# Fix missing nvshmem symlinks (container has .so.3 but not .so)
+NVSHMEM_LIB="$NVSHMEM_DIR/lib"
+if [ ! -f "$NVSHMEM_LIB/libnvshmem_host.so" ] && [ -f "$NVSHMEM_LIB/libnvshmem_host.so.3" ]; then
+    echo "Creating missing nvshmem symlinks..."
+    ln -sf libnvshmem_host.so.3 "$NVSHMEM_LIB/libnvshmem_host.so"
+fi
+
+# Apply kNumMaxTopK=16 patch (Qwen3.5 uses topk=10, default kNumMaxTopK=8 is insufficient)
+# Note: source has both kNumMaxTopK (uppercase) and kNumMaxTopk (lowercase) as separate variables
+sed -i 's/kNumMaxTopK[[:space:]]*=[[:space:]]*[0-9][0-9]*/kNumMaxTopK = 16/g' csrc/kernels/internode_ll.cu
+sed -i 's/kNumMaxTopk[[:space:]]*=[[:space:]]*[0-9][0-9]*/kNumMaxTopk = 16/g' csrc/kernels/internode_ll.cu
+
+# Verify the patch was applied
+grep -q "kNumMaxTop. = 16" csrc/kernels/internode_ll.cu && echo "Patch verified: kNumMaxTopK/k=16" || {
+    echo "ERROR: kNumMaxTopK patch failed to apply!"; exit 1;
+}
+
+# Build with full output so we can debug failures
+# set -e will auto-exit on failure
+TORCH_CUDA_ARCH_LIST="10.0" \
+NVSHMEM_DIR="$NVSHMEM_DIR" \
+pip install -e . --no-build-isolation 2>&1
+
+echo "=== DeepEP rebuild complete ==="
+python3 -c "import deep_ep; print('deep_ep imported successfully')"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-dep4-dep16.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-dep4-dep16.yaml
new file mode 100644
index 0000000000..f1d6ee9cce
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-dep4-dep16.yaml
@@ -0,0 +1,155 @@
+name: "qwen3.5-1p1d-dep4-dep16"
+
+setup_script: rebuild-deepep.sh
+
+dynamo:
+  hash: 46520ca59afe992fb5ef61b3197b2316f8df9b2b
+  install: true
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  num_additional_frontends: 1
+  nginx_container: nginx
+
+model:
+  path: "qwen3.5-fp8"
+  container: "lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
+  precision: "fp8"
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  prefill_nodes: 1
+  decode_nodes: 4
+  prefill_workers: 1
+  decode_workers: 1
+
+backend:
+  type: sglang
+
+  prefill_environment:
+    SGLANG_JIT_DEEPGEMM_PRECOMPILE: "0"
+    NO_COLOR: "1"
+    TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: "3600"
+    TORCH_NCCL_WATCHDOG_TIMEOUT_SEC: "3600"
+    TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC: "3600"
+    PYTHONUNBUFFERED: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_CUMEM_ENABLE: "1"
+    MC_FORCE_MNNVL: "1"
+    SGLANG_DG_CACHE_DIR: "/configs/deepgemm-cache"
+    FLASHINFER_WORKSPACE_BASE: "/configs/flashinfer-cache"
+    SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE: "100000"
+    SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: "100000"
+    SGLANG_DISAGGREGATION_WAITING_TIMEOUT: "100000"
+    SGLANG_MOONCAKE_CUSTOM_MEM_POOL: "True"
+    SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: "0"
+    SGLANG_DISABLE_TP_MEMORY_INBALANCE_CHECK: "1"
+    SGLANG_HEALTH_STARTING_OK: "1"
+    SGLANG_ENABLE_HEALTH_ENDPOINT_GENERATION: "0"
+
+  decode_environment:
+    SGLANG_JIT_DEEPGEMM_PRECOMPILE: "0"
+    NO_COLOR: "1"
+    TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: "3600"
+    TORCH_NCCL_WATCHDOG_TIMEOUT_SEC: "3600"
+    TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC: "3600"
+    PYTHONUNBUFFERED: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_CUMEM_ENABLE: "1"
+    MC_FORCE_MNNVL: "1"
+    MC_TE_METRIC: "true"
+    SGLANG_DG_CACHE_DIR: "/tmp/deepgemm-cache"
+    FLASHINFER_WORKSPACE_BASE: "/configs/flashinfer-cache"
+    SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE: "100000"
+    SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: "100000"
+    SGLANG_DISAGGREGATION_WAITING_TIMEOUT: "100000"
+    SGLANG_DECODE_BOOTSTRAP_TIMEOUT: "1000"
+    SGLANG_HACK_SEQ_BOOTSTRAP_ROOM: "1"
+    SGLANG_MOONCAKE_CUSTOM_MEM_POOL: "True"
+    SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: "0"
+    SGLANG_DISABLE_TP_MEMORY_INBALANCE_CHECK: "1"
+    SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK: "512"
+    SGLANG_HEALTH_CHECK_TIMEOUT: "1800"
+    SGLANG_HEALTH_STARTING_OK: "1"
+    SGLANG_ENABLE_HEALTH_ENDPOINT_GENERATION: "0"
+
+  sglang_config:
+    prefill:
+      served-model-name: "Qwen/Qwen3.5-397B-A17B-FP8"
+      model-path: "/model/"
+      quantization: "fp8"
+      kv-cache-dtype: "fp8_e4m3"
+      trust-remote-code: true
+
+      tensor-parallel-size: 4
+      data-parallel-size: 4
+      expert-parallel-size: 4
+      enable-dp-attention: true
+      enable-dp-lm-head: true
+
+      mamba-scheduler-strategy: "no_buffer"
+      mamba-track-interval: 2048
+      mamba-ssm-dtype: "bfloat16"
+      disaggregation-mode: "prefill"
+      disable-radix-cache: true
+      disaggregation-bootstrap-port: 31000
+      mem-fraction-static: 0.8
+      chunked-prefill-size: 65536
+      load-balance-method: "round_robin"
+      watchdog-timeout: 1000000
+      disable-cuda-graph: true
+      log-level: "info"
+      page-size: 64
+      attention-backend: "trtllm_mha"
+      moe-runner-backend: "flashinfer_trtllm"
+
+    decode:
+      served-model-name: "Qwen/Qwen3.5-397B-A17B-FP8"
+      model-path: "/model/"
+      trust-remote-code: true
+      quantization: "fp8"
+      kv-cache-dtype: "fp8_e4m3"
+
+      tensor-parallel-size: 16
+      data-parallel-size: 16
+      expert-parallel-size: 16
+      enable-dp-attention: true
+      enable-dp-lm-head: true
+      prefill-round-robin-balance: true
+
+      mamba-scheduler-strategy: "no_buffer"
+      mamba-track-interval: 128
+      mamba-ssm-dtype: "bfloat16"
+
+      disaggregation-mode: "decode"
+      disable-radix-cache: true
+      disaggregation-bootstrap-port: 31000
+
+      chunked-prefill-size: 4096
+      context-length: 4096
+      mem-fraction-static: 0.80
+      max-mamba-cache-size: 2048
+      max-running-requests: 2048
+      cuda-graph-max-bs: 128
+      watchdog-timeout: 1000000
+
+      page-size: 64
+      attention-backend: "trtllm_mha"
+      moe-runner-backend: "deep_gemm"
+      moe-a2a-backend: "deepep"
+      deepep-mode: "low_latency"
+      ep-dispatch-algorithm: "static"
+      eplb-algorithm: "deepseek"
+
+      decode-log-interval: 1
+      stream-interval: 50
+
+benchmark:
+  type: "sa-bench"
+  isl: 1024
+  osl: 1024
+  req_rate: "inf"
+  random_range_ratio: 0.8
+  concurrencies: "512x1024x2048"
\ No newline at end of file
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-tp4-tp4.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-tp4-tp4.yaml
new file mode 100644
index 0000000000..a51a2fcb94
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-tp4-tp4.yaml
@@ -0,0 +1,123 @@
+name: "qwen3.5-1p1d-tp4-tp4"
+
+sbatch_directives:
+  mem: "0"
+
+dynamo:
+  hash: 46520ca59afe992fb5ef61b3197b2316f8df9b2b
+  install: true
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  num_additional_frontends: 1
+  nginx_container: nginx
+
+model:
+  path: "qwen3.5-fp8"
+  container: "lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
+  precision: "fp8"
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  prefill_nodes: 1
+  decode_nodes: 1
+  prefill_workers: 1
+  decode_workers: 1
+
+backend:
+  type: sglang
+
+  prefill_environment:
+    SGLANG_JIT_DEEPGEMM_PRECOMPILE: "0"
+    TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: "3600"
+    TORCH_NCCL_WATCHDOG_TIMEOUT_SEC: "3600"
+    TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC: "3600"
+    PYTHONUNBUFFERED: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_CUMEM_ENABLE: "1"
+    SGLANG_DG_CACHE_DIR: "/configs/deepgemm-cache"
+    FLASHINFER_WORKSPACE_BASE: "/configs/flashinfer-cache"
+    SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE: "100000"
+    SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: "100000"
+    SGLANG_DISAGGREGATION_WAITING_TIMEOUT: "100000"
+    SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: "0"
+    SGLANG_MOONCAKE_CUSTOM_MEM_POOL: "True"
+    MC_FORCE_MNNVL: "1"
+
+  decode_environment:
+    SGLANG_JIT_DEEPGEMM_PRECOMPILE: "0"
+    TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: "3600"
+    TORCH_NCCL_WATCHDOG_TIMEOUT_SEC: "3600"
+    TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC: "3600"
+    PYTHONUNBUFFERED: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_CUMEM_ENABLE: "1"
+    SGLANG_DG_CACHE_DIR: "/configs/deepgemm-cache"
+    FLASHINFER_WORKSPACE_BASE: "/configs/flashinfer-cache"
+    SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE: "100000"
+    SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: "100000"
+    SGLANG_DISAGGREGATION_WAITING_TIMEOUT: "100000"
+    SGLANG_DECODE_BOOTSTRAP_TIMEOUT: "1000"
+    SGLANG_MOONCAKE_CUSTOM_MEM_POOL: "True"
+    SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: "0"
+    SGLANG_HEALTH_CHECK_TIMEOUT: "3600"
+    SGLANG_HEALTH_STARTING_OK: "1"
+    SGLANG_ENABLE_HEALTH_ENDPOINT_GENERATION: "0"
+    MC_FORCE_MNNVL: "1"
+    MC_TE_METRIC: "true"
+    SGLANG_HACK_SEQ_BOOTSTRAP_ROOM: "1"
+
+  sglang_config:
+    prefill:
+      served-model-name: "Qwen/Qwen3.5-397B-A17B-FP8"
+      model-path: "/model/"
+      quantization: "fp8"
+      kv-cache-dtype: "fp8_e4m3"
+      trust-remote-code: true
+      attention-backend: "trtllm_mha"
+      tensor-parallel-size: 4
+      mamba-ssm-dtype: "bfloat16"
+      moe-runner-backend: "flashinfer_trtllm"
+      disable-radix-cache: true
+      max-running-requests: 1024
+      mem-fraction-static: 0.8
+      chunked-prefill-size: 16384
+      max-prefill-tokens: 16384
+      context-length: 4096
+      cuda-graph-max-bs: 1024
+      decode-log-interval: 1
+      stream-interval: 50
+      disaggregation-mode: "prefill"
+
+    decode:
+      served-model-name: "Qwen/Qwen3.5-397B-A17B-FP8"
+      model-path: "/model/"
+      quantization: "fp8"
+      kv-cache-dtype: "fp8_e4m3"
+      trust-remote-code: true
+      attention-backend: "trtllm_mha"
+      tensor-parallel-size: 4
+      mamba-ssm-dtype: "bfloat16"
+      moe-runner-backend: "flashinfer_trtllm"
+      disable-radix-cache: true
+      max-running-requests: 1024
+      mem-fraction-static: 0.8
+      chunked-prefill-size: 16384
+      max-prefill-tokens: 16384
+      context-length: 4096
+      cuda-graph-max-bs: 1024
+      decode-log-interval: 1
+      stream-interval: 50
+      disaggregation-mode: "decode"
+
+benchmark:
+  type: "sa-bench"
+  isl: 1024
+  osl: 1024
+  req_rate: "inf"
+  num_prompts_mult: 10
+  num_warmup_mult: 1
+  random_range_ratio: 0.8
+  concurrencies: "1x2x4x8x16x32x64"
\ No newline at end of file
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/2p1d-dep4-dep16.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/2p1d-dep4-dep16.yaml
new file mode 100644
index 0000000000..c7ebbb698e
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/2p1d-dep4-dep16.yaml
@@ -0,0 +1,155 @@
+name: "qwen3.5-2p1d-dep4-dep16"
+
+setup_script: rebuild-deepep.sh
+
+dynamo:
+  hash: 46520ca59afe992fb5ef61b3197b2316f8df9b2b
+  install: true
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  num_additional_frontends: 2
+  nginx_container: nginx
+
+model:
+  path: "qwen3.5-fp8"
+  container: "lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
+  precision: "fp8"
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  prefill_nodes: 2
+  decode_nodes: 4
+  prefill_workers: 2
+  decode_workers: 1
+
+backend:
+  type: sglang
+
+  prefill_environment:
+    SGLANG_JIT_DEEPGEMM_PRECOMPILE: "0"
+    NO_COLOR: "1"
+    TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: "3600"
+    TORCH_NCCL_WATCHDOG_TIMEOUT_SEC: "3600"
+    TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC: "3600"
+    PYTHONUNBUFFERED: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_CUMEM_ENABLE: "1"
+    MC_FORCE_MNNVL: "1"
+    SGLANG_DG_CACHE_DIR: "/configs/deepgemm-cache"
+    FLASHINFER_WORKSPACE_BASE: "/configs/flashinfer-cache"
+    SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE: "100000"
+    SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: "100000"
+    SGLANG_DISAGGREGATION_WAITING_TIMEOUT: "100000"
+    SGLANG_MOONCAKE_CUSTOM_MEM_POOL: "True"
+    SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: "0"
+    SGLANG_DISABLE_TP_MEMORY_INBALANCE_CHECK: "1"
+    SGLANG_HEALTH_STARTING_OK: "1"
+    SGLANG_ENABLE_HEALTH_ENDPOINT_GENERATION: "0"
+
+  decode_environment:
+    SGLANG_JIT_DEEPGEMM_PRECOMPILE: "0"
+    NO_COLOR: "1"
+    TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: "3600"
+    TORCH_NCCL_WATCHDOG_TIMEOUT_SEC: "3600"
+    TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC: "3600"
+    PYTHONUNBUFFERED: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_CUMEM_ENABLE: "1"
+    MC_FORCE_MNNVL: "1"
+    MC_TE_METRIC: "true"
+    SGLANG_DG_CACHE_DIR: "/tmp/deepgemm-cache"
+    FLASHINFER_WORKSPACE_BASE: "/configs/flashinfer-cache"
+    SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE: "100000"
+    SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: "100000"
+    SGLANG_DISAGGREGATION_WAITING_TIMEOUT: "100000"
+    SGLANG_DECODE_BOOTSTRAP_TIMEOUT: "1000"
+    SGLANG_HACK_SEQ_BOOTSTRAP_ROOM: "1"
+    SGLANG_MOONCAKE_CUSTOM_MEM_POOL: "True"
+    SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: "0"
+    SGLANG_DISABLE_TP_MEMORY_INBALANCE_CHECK: "1"
+    SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK: "512"
+    SGLANG_HEALTH_CHECK_TIMEOUT: "1800"
+    SGLANG_HEALTH_STARTING_OK: "1"
+    SGLANG_ENABLE_HEALTH_ENDPOINT_GENERATION: "0"
+
+  sglang_config:
+    prefill:
+      served-model-name: "Qwen/Qwen3.5-397B-A17B-FP8"
+      model-path: "/model/"
+      quantization: "fp8"
+      kv-cache-dtype: "fp8_e4m3"
+      trust-remote-code: true
+
+      tensor-parallel-size: 4
+      data-parallel-size: 4
+      expert-parallel-size: 4
+      enable-dp-attention: true
+      enable-dp-lm-head: true
+
+      mamba-scheduler-strategy: "no_buffer"
+      mamba-track-interval: 2048
+      mamba-ssm-dtype: "bfloat16"
+      disaggregation-mode: "prefill"
+      disable-radix-cache: true
+      disaggregation-bootstrap-port: 31000
+      mem-fraction-static: 0.8
+      chunked-prefill-size: 65536
+      load-balance-method: "round_robin"
+      watchdog-timeout: 1000000
+      disable-cuda-graph: true
+      log-level: "info"
+      page-size: 64
+      attention-backend: "trtllm_mha"
+      moe-runner-backend: "flashinfer_trtllm"
+
+    decode:
+      served-model-name: "Qwen/Qwen3.5-397B-A17B-FP8"
+      model-path: "/model/"
+      trust-remote-code: true
+      quantization: "fp8"
+      kv-cache-dtype: "fp8_e4m3"
+
+      tensor-parallel-size: 16
+      data-parallel-size: 16
+      expert-parallel-size: 16
+      enable-dp-attention: true
+      enable-dp-lm-head: true
+      prefill-round-robin-balance: true
+
+      mamba-scheduler-strategy: "no_buffer"
+      mamba-track-interval: 128
+      mamba-ssm-dtype: "bfloat16"
+
+      disaggregation-mode: "decode"
+      disable-radix-cache: true
+      disaggregation-bootstrap-port: 31000
+
+      chunked-prefill-size: 4096
+      context-length: 8192
+      mem-fraction-static: 0.75
+      max-mamba-cache-size: 4096
+      max-running-requests: 4096
+      cuda-graph-max-bs: 256
+      watchdog-timeout: 1000000
+
+      page-size: 64
+      attention-backend: "trtllm_mha"
+      moe-runner-backend: "deep_gemm"
+      moe-a2a-backend: "deepep"
+      deepep-mode: "low_latency"
+      ep-dispatch-algorithm: "static"
+      eplb-algorithm: "deepseek"
+
+      decode-log-interval: 1
+      stream-interval: 50
+
+benchmark:
+  type: "sa-bench"
+  isl: 1024
+  osl: 1024
+  req_rate: "inf"
+  random_range_ratio: 0.8
+  concurrencies: "4096"
\ No newline at end of file
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/1p1d-tp4-tp4.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/1p1d-tp4-tp4.yaml
new file mode 100644
index 0000000000..c9915dcdbb
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/1p1d-tp4-tp4.yaml
@@ -0,0 +1,121 @@
+name: "qwen3.5-1p1d-tp4-tp4"
+
+sbatch_directives:
+  mem: "0"
+
+dynamo:
+  hash: 46520ca59afe992fb5ef61b3197b2316f8df9b2b
+  install: true
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  num_additional_frontends: 1
+  nginx_container: nginx
+
+model:
+  path: "qwen3.5-fp8"
+  container: "lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
+  precision: "fp8"
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  prefill_nodes: 1
+  decode_nodes: 1
+  prefill_workers: 1
+  decode_workers: 1
+
+backend:
+  type: sglang
+
+  prefill_environment:
+    SGLANG_JIT_DEEPGEMM_PRECOMPILE: "0"
+    TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: "3600"
+    TORCH_NCCL_WATCHDOG_TIMEOUT_SEC: "3600"
+    TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC: "3600"
+    PYTHONUNBUFFERED: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_CUMEM_ENABLE: "1"
+    SGLANG_DG_CACHE_DIR: "/configs/deepgemm-cache"
+    FLASHINFER_WORKSPACE_BASE: "/configs/flashinfer-cache"
+    SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE: "100000"
+    SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: "100000"
+    SGLANG_DISAGGREGATION_WAITING_TIMEOUT: "100000"
+    SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: "0"
+    SGLANG_MOONCAKE_CUSTOM_MEM_POOL: "True"
+    MC_FORCE_MNNVL: "1"
+
+  decode_environment:
+    SGLANG_JIT_DEEPGEMM_PRECOMPILE: "0"
+    TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: "3600"
+    TORCH_NCCL_WATCHDOG_TIMEOUT_SEC: "3600"
+    TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC: "3600"
+    PYTHONUNBUFFERED: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_CUMEM_ENABLE: "1"
+    SGLANG_DG_CACHE_DIR: "/configs/deepgemm-cache"
+    FLASHINFER_WORKSPACE_BASE: "/configs/flashinfer-cache"
+    SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE: "100000"
+    SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: "100000"
+    SGLANG_DISAGGREGATION_WAITING_TIMEOUT: "100000"
+    SGLANG_DECODE_BOOTSTRAP_TIMEOUT: "1000"
+    SGLANG_MOONCAKE_CUSTOM_MEM_POOL: "True"
+    SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: "0"
+    SGLANG_HEALTH_CHECK_TIMEOUT: "3600"
+    SGLANG_HEALTH_STARTING_OK: "1"
+    SGLANG_ENABLE_HEALTH_ENDPOINT_GENERATION: "0"
+    MC_FORCE_MNNVL: "1"
+    MC_TE_METRIC: "true"
+    SGLANG_HACK_SEQ_BOOTSTRAP_ROOM: "1"
+
+  sglang_config:
+    prefill:
+      served-model-name: "Qwen/Qwen3.5-397B-A17B-FP8"
+      model-path: "/model/"
+      quantization: "fp8"
+      kv-cache-dtype: "fp8_e4m3"
+      trust-remote-code: true
+      attention-backend: "trtllm_mha"
+      tensor-parallel-size: 4
+      mamba-ssm-dtype: "bfloat16"
+      moe-runner-backend: "flashinfer_trtllm"
+      disable-radix-cache: true
+      max-running-requests: 1024
+      mem-fraction-static: 0.8
+      chunked-prefill-size: 16384
+      max-prefill-tokens: 16384
+      context-length: 16384
+      cuda-graph-max-bs: 1024
+      decode-log-interval: 1
+      stream-interval: 50
+      disaggregation-mode: "prefill"
+
+    decode:
+      served-model-name: "Qwen/Qwen3.5-397B-A17B-FP8"
+      model-path: "/model/"
+      quantization: "fp8"
+      kv-cache-dtype: "fp8_e4m3"
+      trust-remote-code: true
+      attention-backend: "trtllm_mha"
+      tensor-parallel-size: 4
+      mamba-ssm-dtype: "bfloat16"
+      moe-runner-backend: "flashinfer_trtllm"
+      disable-radix-cache: true
+      max-running-requests: 1024
+      mem-fraction-static: 0.8
+      chunked-prefill-size: 16384
+      max-prefill-tokens: 16384
+      context-length: 16384
+      cuda-graph-max-bs: 1024
+      decode-log-interval: 1
+      stream-interval: 50
+      disaggregation-mode: "decode"
+
+benchmark:
+  type: "sa-bench"
+  isl: 8192
+  osl: 1024
+  req_rate: "inf"
+  random_range_ratio: 0.8
+  concurrencies: "1x2x4x8x16x32x64x128"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/4p1d-dep4-dep16.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/4p1d-dep4-dep16.yaml
new file mode 100644
index 0000000000..cd3cd91e3b
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/4p1d-dep4-dep16.yaml
@@ -0,0 +1,160 @@
+name: "qwen3.5-4p1d-dep4-dep16"
+
+setup_script: rebuild-deepep.sh
+
+sbatch_directives:
+  mem: "0"
+
+dynamo:
+  hash: 46520ca59afe992fb5ef61b3197b2316f8df9b2b
+  install: true
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  num_additional_frontends: 3
+  nginx_container: nginx
+
+model:
+  path: "qwen3.5-fp8"
+  container: "lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
+  precision: "fp8"
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  prefill_nodes: 4
+  decode_nodes: 4
+  prefill_workers: 4
+  decode_workers: 1
+
+backend:
+  type: sglang
+
+  prefill_environment:
+    SGLANG_JIT_DEEPGEMM_PRECOMPILE: "0"
+    NO_COLOR: "1"
+    TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: "3600"
+    TORCH_NCCL_WATCHDOG_TIMEOUT_SEC: "3600"
+    TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC: "3600"
+    PYTHONUNBUFFERED: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_CUMEM_ENABLE: "1"
+    MC_FORCE_MNNVL: "1"
+    SGLANG_DG_CACHE_DIR: "/configs/deepgemm-cache"
+    FLASHINFER_WORKSPACE_BASE: "/configs/flashinfer-cache"
+    SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE: "100000"
+    SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: "100000"
+    SGLANG_DISAGGREGATION_WAITING_TIMEOUT: "100000"
+    SGLANG_MOONCAKE_CUSTOM_MEM_POOL: "True"
+    SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: "0"
+    SGLANG_DISABLE_TP_MEMORY_INBALANCE_CHECK: "1"
+    SGLANG_HEALTH_STARTING_OK: "1"
+    SGLANG_ENABLE_HEALTH_ENDPOINT_GENERATION: "0"
+
+  decode_environment:
+    SGLANG_JIT_DEEPGEMM_PRECOMPILE: "0"
+    NO_COLOR: "1"
+    TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: "3600"
+    TORCH_NCCL_WATCHDOG_TIMEOUT_SEC: "3600"
+    TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC: "3600"
+    PYTHONUNBUFFERED: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_CUMEM_ENABLE: "1"
+    MC_FORCE_MNNVL: "1"
+    MC_TE_METRIC: "true"
+    SGLANG_DG_CACHE_DIR: "/tmp/deepgemm-cache"
+    FLASHINFER_WORKSPACE_BASE: "/configs/flashinfer-cache"
+    SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE: "100000"
+    SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: "100000"
+    SGLANG_DISAGGREGATION_WAITING_TIMEOUT: "100000"
+    SGLANG_DECODE_BOOTSTRAP_TIMEOUT: "1000"
+    SGLANG_HACK_SEQ_BOOTSTRAP_ROOM: "1"
+    SGLANG_MOONCAKE_CUSTOM_MEM_POOL: "True"
+    SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: "0"
+    SGLANG_DISABLE_TP_MEMORY_INBALANCE_CHECK: "1"
+    SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK: "512"
+    SGLANG_HEALTH_CHECK_TIMEOUT: "1800"
+    SGLANG_HEALTH_STARTING_OK: "1"
+    SGLANG_ENABLE_HEALTH_ENDPOINT_GENERATION: "0"
+
+  sglang_config:
+    prefill:
+      served-model-name: "Qwen/Qwen3.5-397B-A17B-FP8"
+      model-path: "/model/"
+      quantization: "fp8"
+      kv-cache-dtype: "fp8_e4m3"
+      trust-remote-code: true
+
+      tensor-parallel-size: 4
+      data-parallel-size: 4
+      expert-parallel-size: 4
+      enable-dp-attention: true
+      enable-dp-lm-head: true
+
+      mamba-scheduler-strategy: "no_buffer"
+      mamba-track-interval: 2048
+      mamba-ssm-dtype: "bfloat16"
+      disaggregation-mode: "prefill"
+      disable-radix-cache: true
+      disaggregation-bootstrap-port: 31000
+      mem-fraction-static: 0.8
+      chunked-prefill-size: 65536
+      load-balance-method: "round_robin"
+      watchdog-timeout: 1000000
+      disable-cuda-graph: true
+      log-level: "info"
+      page-size: 64
+      attention-backend: "trtllm_mha"
+      moe-runner-backend: "flashinfer_trtllm"
+
+    decode:
+      served-model-name: "Qwen/Qwen3.5-397B-A17B-FP8"
+      model-path: "/model/"
+      trust-remote-code: true
+      quantization: "fp8"
+      kv-cache-dtype: "fp8_e4m3"
+
+      tensor-parallel-size: 16
+      data-parallel-size: 16
+      expert-parallel-size: 16
+      enable-dp-attention: true
+      enable-dp-lm-head: true
+      prefill-round-robin-balance: true
+
+      mamba-scheduler-strategy: "no_buffer"
+      mamba-track-interval: 128
+      mamba-ssm-dtype: "bfloat16"
+
+      disaggregation-mode: "decode"
+      disable-radix-cache: true
+      disaggregation-bootstrap-port: 31000
+
+      chunked-prefill-size: 4096
+      context-length: 16384
+      mem-fraction-static: 0.80
+      max-mamba-cache-size: 2048
+      max-running-requests: 2048
+      cuda-graph-max-bs: 128
+      watchdog-timeout: 1000000
+
+      page-size: 64
+      attention-backend: "trtllm_mha"
+      moe-runner-backend: "deep_gemm"
+      moe-a2a-backend: "deepep"
+      deepep-mode: "low_latency"
+      ep-dispatch-algorithm: "static"
+      eplb-algorithm: "deepseek"
+
+      decode-log-interval: 1
+      stream-interval: 50
+
+benchmark:
+  type: "sa-bench"
+  isl: 8192
+  osl: 1024
+  req_rate: "inf"
+  num_prompts_mult: 20
+  num_warmup_mult: 2
+  random_range_ratio: 0.8
+  concurrencies: "1024"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/8p1d-dep4-dep16.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/8p1d-dep4-dep16.yaml
new file mode 100644
index 0000000000..258e347244
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/8p1d-dep4-dep16.yaml
@@ -0,0 +1,161 @@
+name: "qwen3.5-8p1d-dep4-dep16"
+
+setup_script: rebuild-deepep.sh
+
+sbatch_directives:
+  mem: "0"
+
+infra:
+  etcd_nats_dedicated_node: true
+
+dynamo:
+  hash: 46520ca59afe992fb5ef61b3197b2316f8df9b2b
+  install: true
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  num_additional_frontends: 4
+  nginx_container: nginx
+
+model:
+  path: "qwen3.5-fp8"
+  container: "lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
+  precision: "fp8"
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  prefill_nodes: 8
+  decode_nodes: 4
+  prefill_workers: 8
+  decode_workers: 1
+
+backend:
+  type: sglang
+
+  prefill_environment:
+    SGLANG_JIT_DEEPGEMM_PRECOMPILE: "0"
+    NO_COLOR: "1"
+    TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: "3600"
+    TORCH_NCCL_WATCHDOG_TIMEOUT_SEC: "3600"
+    TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC: "3600"
+    PYTHONUNBUFFERED: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_CUMEM_ENABLE: "1"
+    MC_FORCE_MNNVL: "1"
+    SGLANG_DG_CACHE_DIR: "/configs/deepgemm-cache"
+    FLASHINFER_WORKSPACE_BASE: "/configs/flashinfer-cache"
+    SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE: "100000"
+    SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: "100000"
+    SGLANG_DISAGGREGATION_WAITING_TIMEOUT: "100000"
+    SGLANG_MOONCAKE_CUSTOM_MEM_POOL: "True"
+    SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: "0"
+    SGLANG_DISABLE_TP_MEMORY_INBALANCE_CHECK: "1"
+    SGLANG_HEALTH_STARTING_OK: "1"
+    SGLANG_ENABLE_HEALTH_ENDPOINT_GENERATION: "0"
+
+  decode_environment:
+    SGLANG_JIT_DEEPGEMM_PRECOMPILE: "0"
+    NO_COLOR: "1"
+    TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: "3600"
+    TORCH_NCCL_WATCHDOG_TIMEOUT_SEC: "3600"
+    TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC: "3600"
+    PYTHONUNBUFFERED: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_CUMEM_ENABLE: "1"
+    MC_FORCE_MNNVL: "1"
+    MC_TE_METRIC: "true"
+    SGLANG_DG_CACHE_DIR: "/tmp/deepgemm-cache"
+    FLASHINFER_WORKSPACE_BASE: "/configs/flashinfer-cache"
+    SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE: "100000"
+    SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: "100000"
+    SGLANG_DISAGGREGATION_WAITING_TIMEOUT: "100000"
+    SGLANG_DECODE_BOOTSTRAP_TIMEOUT: "1000"
+    SGLANG_HACK_SEQ_BOOTSTRAP_ROOM: "1"
+    SGLANG_MOONCAKE_CUSTOM_MEM_POOL: "True"
+    SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: "0"
+    SGLANG_DISABLE_TP_MEMORY_INBALANCE_CHECK: "1"
+    SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK: "512"
+    SGLANG_HEALTH_CHECK_TIMEOUT: "1800"
+    SGLANG_HEALTH_STARTING_OK: "1"
+    SGLANG_ENABLE_HEALTH_ENDPOINT_GENERATION: "0"
+
+  sglang_config:
+    prefill:
+      served-model-name: "Qwen/Qwen3.5-397B-A17B-FP8"
+      model-path: "/model/"
+      quantization: "fp8"
+      kv-cache-dtype: "fp8_e4m3"
+      trust-remote-code: true
+
+      tensor-parallel-size: 4
+      data-parallel-size: 4
+      expert-parallel-size: 4
+      enable-dp-attention: true
+      enable-dp-lm-head: true
+
+      mamba-scheduler-strategy: "no_buffer"
+      mamba-track-interval: 2048
+      mamba-ssm-dtype: "bfloat16"
+      disaggregation-mode: "prefill"
+      disable-radix-cache: true
+      disaggregation-bootstrap-port: 31000
+      mem-fraction-static: 0.8
+      chunked-prefill-size: 65536
+      load-balance-method: "round_robin"
+      watchdog-timeout: 1000000
+      disable-cuda-graph: true
+      log-level: "info"
+      page-size: 64
+      attention-backend: "trtllm_mha"
+      moe-runner-backend: "flashinfer_trtllm"
+
+    decode:
+      served-model-name: "Qwen/Qwen3.5-397B-A17B-FP8"
+      model-path: "/model/"
+      trust-remote-code: true
+      quantization: "fp8"
+      kv-cache-dtype: "fp8_e4m3"
+
+      tensor-parallel-size: 16
+      data-parallel-size: 16
+      expert-parallel-size: 16
+      enable-dp-attention: true
+      enable-dp-lm-head: true
+      prefill-round-robin-balance: true
+
+      mamba-scheduler-strategy: "no_buffer"
+      mamba-track-interval: 128
+      mamba-ssm-dtype: "bfloat16"
+
+      disaggregation-mode: "decode"
+      disable-radix-cache: true
+      disaggregation-bootstrap-port: 31000
+
+      chunked-prefill-size: 4096
+      context-length: 16384
+      mem-fraction-static: 0.80
+      max-mamba-cache-size: 2048
+      max-running-requests: 2048
+      cuda-graph-max-bs: 128
+      watchdog-timeout: 1000000
+
+      page-size: 64
+      attention-backend: "trtllm_mha"
+      moe-runner-backend: "deep_gemm"
+      moe-a2a-backend: "deepep"
+      deepep-mode: "low_latency"
+      ep-dispatch-algorithm: "static"
+      eplb-algorithm: "deepseek"
+
+      decode-log-interval: 1
+      stream-interval: 50
+
+benchmark:
+  type: "sa-bench"
+  isl: 8192
+  osl: 1024
+  req_rate: "inf"
+  random_range_ratio: 0.8
+  concurrencies: "2048x4096"
diff --git a/configs/nvidia-master.yaml b/configs/nvidia-master.yaml
index 6be8e006f9..ddd6f454ed 100644
--- a/configs/nvidia-master.yaml
+++ b/configs/nvidia-master.yaml
@@ -14563,3 +14563,111 @@ glm5-fp4-gb300-dynamo-trt-mtp:
           ep: 16
           dp-attn: true
 
+
+qwen3.5-fp8-gb300-dynamo-sglang:
+  image: lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc
+  model: Qwen/Qwen3.5-397B-A17B-FP8
+  model-prefix: qwen3.5
+  runner: gb300
+  precision: fp8
+  framework: dynamo-sglang
+  multinode: true
+  disagg: true
+  scenarios:
+    fixed-seq-len:
+    - isl: 1024
+      osl: 1024
+      search-space:
+      # 1P1D STP: TP4 prefill + TP4 decode (pure tensor parallel). 2 nodes (1+1).
+      - spec-decoding: "none"
+        conc-list: [1, 2, 4, 8, 16, 32, 64]
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 1
+          dp-attn: false
+          additional-settings:
+          - "CONFIG_FILE=recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-tp4-tp4.yaml"
+        decode:
+          num-worker: 1
+          tp: 4
+          ep: 1
+          dp-attn: false
+      # 1P1D wide-EP: prefill DEP4 + decode DEP16. 5 nodes (1+4).
+      - spec-decoding: "none"
+        conc-list: [512, 1024, 2048]
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-dep4-dep16.yaml"
+        decode:
+          num-worker: 1
+          tp: 16
+          ep: 16
+          dp-attn: true
+      # 2P1D wide-EP: 2 prefill DEP4 + decode DEP16. 6 nodes (2+4).
+      - spec-decoding: "none"
+        conc-list: [4096]
+        prefill:
+          num-worker: 2
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/sglang/qwen3.5/gb300-fp8/1k1k/2p1d-dep4-dep16.yaml"
+        decode:
+          num-worker: 1
+          tp: 16
+          ep: 16
+          dp-attn: true
+    - isl: 8192
+      osl: 1024
+      search-space:
+      # 1P1D STP: TP4 prefill + TP4 decode (pure tensor parallel). 2 nodes (1+1).
+      - spec-decoding: "none"
+        conc-list: [1, 2, 4, 8, 16, 32, 64, 128]
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 1
+          dp-attn: false
+          additional-settings:
+          - "CONFIG_FILE=recipes/sglang/qwen3.5/gb300-fp8/8k1k/1p1d-tp4-tp4.yaml"
+        decode:
+          num-worker: 1
+          tp: 4
+          ep: 1
+          dp-attn: false
+      # 4P1D wide-EP: 4 prefill DEP4 + decode DEP16. 8 nodes (4+4).
+      - spec-decoding: "none"
+        conc-list: [1024]
+        prefill:
+          num-worker: 4
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/sglang/qwen3.5/gb300-fp8/8k1k/4p1d-dep4-dep16.yaml"
+        decode:
+          num-worker: 1
+          tp: 16
+          ep: 16
+          dp-attn: true
+      # 8P1D wide-EP: 8 prefill DEP4 + decode DEP16. 12 nodes (8+4).
+      - spec-decoding: "none"
+        conc-list: [2048, 4096]
+        prefill:
+          num-worker: 8
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/sglang/qwen3.5/gb300-fp8/8k1k/8p1d-dep4-dep16.yaml"
+        decode:
+          num-worker: 1
+          tp: 16
+          ep: 16
+          dp-attn: true
diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index 3425fd7c7c..d7735a1884 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -4669,3 +4669,11 @@
     - "Image: lmsysorg/sglang:v0.5.11-cu130"
     - "11 topologies across 1k/1k and 8k/1k (prefill TP4 + decode wide-EP: 8k/1k DEP16/24/32/40, 1k/1k DEP48/56; per-node TP4 low-latency); MTP flags: speculative-algorithm EAGLE, num-steps 2, eagle-topk 1, num-draft-tokens 3"
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1907
+
+- config-keys:
+    - qwen3.5-fp8-gb300-dynamo-sglang
+  description:
+    - "Add Qwen3.5-397B-A17B-FP8 GB300 disaggregated multinode SGLang benchmarks via Dynamo"
+    - "Image: lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
+    - "6 topologies across 1k/1k and 8k/1k: 1P1D TP4 STP + wide-EP (DEP4 prefill / DEP16 decode) from 1P1D up to 8P1D, recipes under benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/"
+  pr-link: XXX
diff --git a/runners/launch_gb300-nv.sh b/runners/launch_gb300-nv.sh
index 5328ded02d..3e3fa61255 100644
--- a/runners/launch_gb300-nv.sh
+++ b/runners/launch_gb300-nv.sh
@@ -74,8 +74,13 @@ elif [[ $MODEL_PREFIX == "qwen3.5" && $PRECISION == "fp4" ]]; then
     # Qwen3.5 sglang recipes (qwen3.5-fp4).
     export MODEL_PATH=/scratch/models/Qwen3.5-397B-A17B-NVFP4
     export SRT_SLURM_MODEL_PREFIX="qwen3.5-fp4"
+elif [[ $MODEL_PREFIX == "qwen3.5" && $PRECISION == "fp8" ]]; then
+    # SRT_SLURM_MODEL_PREFIX must match the model.path alias used in our
+    # Qwen3.5 sglang recipes (qwen3.5-fp8).
+    export MODEL_PATH=/scratch/models/Qwen3.5-397B-A17B-FP8
+    export SRT_SLURM_MODEL_PREFIX="qwen3.5-fp8"
 else
-    echo "Unsupported model: $MODEL_PREFIX-$PRECISION. Supported models are: dsr1-fp4, dsr1-fp8, dsv4-fp4, glm5-fp4, glm5-fp8, minimaxm2.5-fp4, minimaxm2.5-fp8, kimik2.5-fp4, qwen3.5-fp4"
+    echo "Unsupported model: $MODEL_PREFIX-$PRECISION. Supported models are: dsr1-fp4, dsr1-fp8, dsv4-fp4, glm5-fp4, glm5-fp8, minimaxm2.5-fp4, minimaxm2.5-fp8, kimik2.5-fp4, qwen3.5-fp4, qwen3.5-fp8"
     exit 1
 fi
 
@@ -184,10 +189,28 @@ elif [[ $FRAMEWORK == "dynamo-sglang" && $MODEL_PREFIX == "glm5.1" ]]; then
     cd "$SRT_REPO_DIR"
     git checkout main
 elif [[ $FRAMEWORK == "dynamo-sglang" && $MODEL_PREFIX == "qwen3.5" ]]; then
-    # Overlay our version-controlled Qwen3.5 recipes onto the submission branch.
+    # Overlay our version-controlled Qwen3.5 recipes onto the srt-slurm checkout.
+    # fp8 recipes pin dynamo by commit hash (source install), which needs the
+    # cargo/maturin bootstrap that only exists on srt-slurm:main — the
+    # sa-submission-q2-2026 sglang install path assumes maturin ships in the
+    # image, and the lmsysorg/sglang nightly-dev-cu13 image doesn't include it.
+    # Same branch the identical gb200-fp8 recipes run on. fp4 recipes pin
+    # dynamo by version (pip install) and stay on the submission branch they
+    # were validated against.
     git clone https://github.com/NVIDIA/srt-slurm.git "$SRT_REPO_DIR"
     cd "$SRT_REPO_DIR"
-    git checkout sa-submission-q2-2026
+    if [[ $PRECISION == "fp8" ]]; then
+        git checkout main
+        # The stock configs/rebuild-deepep.sh (setup_script of the wide-EP
+        # fp8 recipes) resolves NVSHMEM by directory name, which can match a
+        # header-less stub in the cu13 nightly image (tilelang's bundled tvm
+        # contrib/nvshmem) and fail the DeepEP build with "nvshmem.h: No
+        # such file or directory". Overlay our header-based resolver.
+        cp "$GITHUB_WORKSPACE/benchmarks/multi_node/srt-slurm-recipes/configs/rebuild-deepep.sh" \
+            configs/rebuild-deepep.sh
+    else
+        git checkout sa-submission-q2-2026
+    fi
     mkdir -p recipes/sglang/qwen3.5
     cp -rT "$GITHUB_WORKSPACE/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5" recipes/sglang/qwen3.5
 elif [[ $FRAMEWORK == "dynamo-vllm" && $MODEL_PREFIX == "minimaxm3" ]]; then
@@ -308,9 +331,12 @@ sed -i "s/^name:.*/name: \"${RUNNER_NAME}\"/" "$CONFIG_PATH"
 # the GHA runner host (im-gb300-login-02, an x86 login node). It's required
 # whenever model.path resolves to the node-local /scratch NVMe that the login
 # node can't see:
-#   - the agentic path (DSv4-Pro checkpoint), and
+#   - the agentic path (DSv4-Pro checkpoint),
 #   - glm5.1, whose GLM-5.1-NVFP4 weights are prestaged on the compute-node
-#     /scratch/models.
+#     /scratch/models, and
+#   - qwen3.5 fp8, whose weights are also on the compute-node /scratch/models
+#     and which runs on srt-slurm:main (the branch that has the preflight;
+#     qwen3.5 fp4 runs on sa-submission-q2-2026, which has none).
 # The engine still fails loudly at runtime if the path is genuinely missing on
 # the compute node. Other fixed-seq-len recipes resolve model.path to a
 # login-visible location, so keep the precheck enforced for them.
@@ -318,7 +344,7 @@ SRTCTL_APPLY_ARGS=(
     -f "$CONFIG_FILE"
     --tags "gb300,${MODEL_PREFIX},${PRECISION},${ISL}x${OSL},infmax-$(date +%Y%m%d)"
 )
-if [[ "$IS_AGENTIC" == "1" || "$MODEL_PREFIX" == "glm5.1" ]]; then
+if [[ "$IS_AGENTIC" == "1" || "$MODEL_PREFIX" == "glm5.1" || ( "$MODEL_PREFIX" == "qwen3.5" && "$PRECISION" == "fp8" ) ]]; then
     SRTCTL_APPLY_ARGS+=(--no-preflight)
 fi
 if [[ -n "$SRTCTL_SETUP_SCRIPT" ]]; then

From ef2603ae51a79f7d1251190d0be1e9d1fb43690d Mon Sep 17 00:00:00 2001
From: Rohit Pujar Nagraj <rpujarnagraj@nvidia.com>
Date: Thu, 9 Jul 2026 11:18:17 -0700
Subject: [PATCH 2/4] Update perf-changelog pr-link for #2137
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

中文：更新 perf-changelog 中 #2137 的 pr-link。
---
 perf-changelog.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index d7735a1884..ebc4a0df93 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -4676,4 +4676,4 @@
     - "Add Qwen3.5-397B-A17B-FP8 GB300 disaggregated multinode SGLang benchmarks via Dynamo"
     - "Image: lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
     - "6 topologies across 1k/1k and 8k/1k: 1P1D TP4 STP + wide-EP (DEP4 prefill / DEP16 decode) from 1P1D up to 8P1D, recipes under benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/"
-  pr-link: XXX
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2137

From a4e55c6a3a1965070e6aec4647da54a94963e9e3 Mon Sep 17 00:00:00 2001
From: Rohit Pujar Nagraj <rpujarnagraj@nvidia.com>
Date: Thu, 9 Jul 2026 11:55:44 -0700
Subject: [PATCH 3/4] Bump qwen3.5-fp8-gb300-dynamo-sglang to sglang
 nightly-dev-cu13-20260709
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Update the pinned container image from nightly-dev-cu13-20260608-303757cc
to nightly-dev-cu13-20260709-074bb928 across the master config entry, the
six gb300-fp8 recipes, and the perf-changelog entry.

中文：将 qwen3.5-fp8-gb300-dynamo-sglang 的容器镜像从
nightly-dev-cu13-20260608-303757cc 更新为
nightly-dev-cu13-20260709-074bb928，同步更新主配置条目、六个
gb300-fp8 配方及 perf-changelog 条目。
---
 .../sglang/qwen3.5/gb300-fp8/1k1k/1p1d-dep4-dep16.yaml          | 2 +-
 .../sglang/qwen3.5/gb300-fp8/1k1k/1p1d-tp4-tp4.yaml             | 2 +-
 .../sglang/qwen3.5/gb300-fp8/1k1k/2p1d-dep4-dep16.yaml          | 2 +-
 .../sglang/qwen3.5/gb300-fp8/8k1k/1p1d-tp4-tp4.yaml             | 2 +-
 .../sglang/qwen3.5/gb300-fp8/8k1k/4p1d-dep4-dep16.yaml          | 2 +-
 .../sglang/qwen3.5/gb300-fp8/8k1k/8p1d-dep4-dep16.yaml          | 2 +-
 configs/nvidia-master.yaml                                      | 2 +-
 perf-changelog.yaml                                             | 2 +-
 8 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-dep4-dep16.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-dep4-dep16.yaml
index f1d6ee9cce..a30ac61660 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-dep4-dep16.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-dep4-dep16.yaml
@@ -14,7 +14,7 @@ frontend:
 
 model:
   path: "qwen3.5-fp8"
-  container: "lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
+  container: "lmsysorg/sglang:nightly-dev-cu13-20260709-074bb928"
   precision: "fp8"
 
 resources:
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-tp4-tp4.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-tp4-tp4.yaml
index a51a2fcb94..90eaa12aae 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-tp4-tp4.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-tp4-tp4.yaml
@@ -15,7 +15,7 @@ frontend:
 
 model:
   path: "qwen3.5-fp8"
-  container: "lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
+  container: "lmsysorg/sglang:nightly-dev-cu13-20260709-074bb928"
   precision: "fp8"
 
 resources:
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/2p1d-dep4-dep16.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/2p1d-dep4-dep16.yaml
index c7ebbb698e..35d2aba8e7 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/2p1d-dep4-dep16.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/2p1d-dep4-dep16.yaml
@@ -14,7 +14,7 @@ frontend:
 
 model:
   path: "qwen3.5-fp8"
-  container: "lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
+  container: "lmsysorg/sglang:nightly-dev-cu13-20260709-074bb928"
   precision: "fp8"
 
 resources:
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/1p1d-tp4-tp4.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/1p1d-tp4-tp4.yaml
index c9915dcdbb..0a938fb48c 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/1p1d-tp4-tp4.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/1p1d-tp4-tp4.yaml
@@ -15,7 +15,7 @@ frontend:
 
 model:
   path: "qwen3.5-fp8"
-  container: "lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
+  container: "lmsysorg/sglang:nightly-dev-cu13-20260709-074bb928"
   precision: "fp8"
 
 resources:
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/4p1d-dep4-dep16.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/4p1d-dep4-dep16.yaml
index cd3cd91e3b..8b4b802669 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/4p1d-dep4-dep16.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/4p1d-dep4-dep16.yaml
@@ -17,7 +17,7 @@ frontend:
 
 model:
   path: "qwen3.5-fp8"
-  container: "lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
+  container: "lmsysorg/sglang:nightly-dev-cu13-20260709-074bb928"
   precision: "fp8"
 
 resources:
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/8p1d-dep4-dep16.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/8p1d-dep4-dep16.yaml
index 258e347244..cf9ca607e2 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/8p1d-dep4-dep16.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/8p1d-dep4-dep16.yaml
@@ -20,7 +20,7 @@ frontend:
 
 model:
   path: "qwen3.5-fp8"
-  container: "lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
+  container: "lmsysorg/sglang:nightly-dev-cu13-20260709-074bb928"
   precision: "fp8"
 
 resources:
diff --git a/configs/nvidia-master.yaml b/configs/nvidia-master.yaml
index ddd6f454ed..3d9e46399b 100644
--- a/configs/nvidia-master.yaml
+++ b/configs/nvidia-master.yaml
@@ -14565,7 +14565,7 @@ glm5-fp4-gb300-dynamo-trt-mtp:
 
 
 qwen3.5-fp8-gb300-dynamo-sglang:
-  image: lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc
+  image: lmsysorg/sglang:nightly-dev-cu13-20260709-074bb928
   model: Qwen/Qwen3.5-397B-A17B-FP8
   model-prefix: qwen3.5
   runner: gb300
diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index ebc4a0df93..b87c735fd1 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -4674,6 +4674,6 @@
     - qwen3.5-fp8-gb300-dynamo-sglang
   description:
     - "Add Qwen3.5-397B-A17B-FP8 GB300 disaggregated multinode SGLang benchmarks via Dynamo"
-    - "Image: lmsysorg/sglang:nightly-dev-cu13-20260608-303757cc"
+    - "Image: lmsysorg/sglang:nightly-dev-cu13-20260709-074bb928"
     - "6 topologies across 1k/1k and 8k/1k: 1P1D TP4 STP + wide-EP (DEP4 prefill / DEP16 decode) from 1P1D up to 8P1D, recipes under benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/"
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2137

From 8b283ebddcc6bd8fbe86a66eb1beb153b95cc76c Mon Sep 17 00:00:00 2001
From: Rohit Pujar Nagraj <rpujarnagraj@nvidia.com>
Date: Fri, 10 Jul 2026 17:57:05 -0700
Subject: [PATCH 4/4] Remove DeepEP rebuild: image default kNumMaxTopK=11
 covers Qwen3.5 topk=10
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The DeepEP snapshot baked into the GB cu13 sglang images caps the
low-latency dispatch/combine paths at kNumMaxTopK=11, which already
accommodates Qwen3.5's top-10 expert routing. The only lower cap
(kNumMaxTopk=9) is in the opt-in overlapped-combine kernel, which these
recipes do not enable. Drop setup_script: rebuild-deepep.sh from the
wide-EP recipes, the vendored rebuild script, and its launch-script
overlay. No other sglang recipe carries a setup script.

中文：移除 DeepEP 重建步骤——GB cu13 sglang 镜像内置的 DeepEP 低延迟
dispatch/combine 路径默认 kNumMaxTopK=11，已满足 Qwen3.5 的 top-10 专家
路由；唯一更低的上限（kNumMaxTopk=9）位于这些配方未启用的 overlapped-
combine 内核中。删除宽 EP 配方中的 setup_script、内置重建脚本及启动
脚本中的覆盖逻辑。
---
 .../configs/rebuild-deepep.sh                 | 59 -------------------
 .../gb300-fp8/1k1k/1p1d-dep4-dep16.yaml       |  1 -
 .../gb300-fp8/1k1k/2p1d-dep4-dep16.yaml       |  1 -
 .../gb300-fp8/8k1k/4p1d-dep4-dep16.yaml       |  1 -
 .../gb300-fp8/8k1k/8p1d-dep4-dep16.yaml       |  1 -
 runners/launch_gb300-nv.sh                    |  7 ---
 6 files changed, 70 deletions(-)
 delete mode 100644 benchmarks/multi_node/srt-slurm-recipes/configs/rebuild-deepep.sh

diff --git a/benchmarks/multi_node/srt-slurm-recipes/configs/rebuild-deepep.sh b/benchmarks/multi_node/srt-slurm-recipes/configs/rebuild-deepep.sh
deleted file mode 100644
index 0be5dba59a..0000000000
--- a/benchmarks/multi_node/srt-slurm-recipes/configs/rebuild-deepep.sh
+++ /dev/null
@@ -1,59 +0,0 @@
-#!/bin/bash
-# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
-# SPDX-License-Identifier: Apache-2.0
-set -eux
-
-echo "=== Rebuilding DeepEP with kNumMaxTopK=16 for Qwen3.5 (topk=10) ==="
-
-DEEPEP_SRC="/sgl-workspace/DeepEP"
-
-if [ ! -d "$DEEPEP_SRC" ]; then
-    echo "ERROR: DeepEP source not found at $DEEPEP_SRC (mount via extra_mount)"
-    exit 1
-fi
-
-cd "$DEEPEP_SRC"
-
-# Resolve NVSHMEM by locating its header, not by directory name: a bare
-# `find -name nvshmem -type d` can latch onto header-less source stubs
-# (e.g. tilelang's bundled tvm src/runtime/contrib/nvshmem) and the build
-# then dies with "fatal error: nvshmem.h: No such file or directory".
-NVSHMEM_H=$(find /usr/local /opt /usr/lib /sgl-workspace -name "nvshmem.h" -path "*/include/*" \
-    -not -path "*tilelang*" -not -path "*flashinfer*" -not -path "*DeepEP*" 2>/dev/null | head -1)
-if [ -z "${NVSHMEM_H:-}" ]; then
-    echo "No NVSHMEM headers found in the image; installing nvidia-nvshmem-cu13..."
-    pip install --break-system-packages --quiet nvidia-nvshmem-cu13
-    NVSHMEM_H=$(find /usr/local/lib/python3*/dist-packages/nvidia -name "nvshmem.h" -path "*/include/*" 2>/dev/null | head -1)
-fi
-if [ -z "${NVSHMEM_H:-}" ]; then
-    echo "ERROR: NVSHMEM headers not found in image or via pip" >&2
-    exit 1
-fi
-NVSHMEM_DIR=$(dirname "$(dirname "$NVSHMEM_H")")
-echo "NVSHMEM_DIR=$NVSHMEM_DIR"
-
-# Fix missing nvshmem symlinks (container has .so.3 but not .so)
-NVSHMEM_LIB="$NVSHMEM_DIR/lib"
-if [ ! -f "$NVSHMEM_LIB/libnvshmem_host.so" ] && [ -f "$NVSHMEM_LIB/libnvshmem_host.so.3" ]; then
-    echo "Creating missing nvshmem symlinks..."
-    ln -sf libnvshmem_host.so.3 "$NVSHMEM_LIB/libnvshmem_host.so"
-fi
-
-# Apply kNumMaxTopK=16 patch (Qwen3.5 uses topk=10, default kNumMaxTopK=8 is insufficient)
-# Note: source has both kNumMaxTopK (uppercase) and kNumMaxTopk (lowercase) as separate variables
-sed -i 's/kNumMaxTopK[[:space:]]*=[[:space:]]*[0-9][0-9]*/kNumMaxTopK = 16/g' csrc/kernels/internode_ll.cu
-sed -i 's/kNumMaxTopk[[:space:]]*=[[:space:]]*[0-9][0-9]*/kNumMaxTopk = 16/g' csrc/kernels/internode_ll.cu
-
-# Verify the patch was applied
-grep -q "kNumMaxTop. = 16" csrc/kernels/internode_ll.cu && echo "Patch verified: kNumMaxTopK/k=16" || {
-    echo "ERROR: kNumMaxTopK patch failed to apply!"; exit 1;
-}
-
-# Build with full output so we can debug failures
-# set -e will auto-exit on failure
-TORCH_CUDA_ARCH_LIST="10.0" \
-NVSHMEM_DIR="$NVSHMEM_DIR" \
-pip install -e . --no-build-isolation 2>&1
-
-echo "=== DeepEP rebuild complete ==="
-python3 -c "import deep_ep; print('deep_ep imported successfully')"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-dep4-dep16.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-dep4-dep16.yaml
index a30ac61660..39e5315059 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-dep4-dep16.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/1p1d-dep4-dep16.yaml
@@ -1,6 +1,5 @@
 name: "qwen3.5-1p1d-dep4-dep16"
 
-setup_script: rebuild-deepep.sh
 
 dynamo:
   hash: 46520ca59afe992fb5ef61b3197b2316f8df9b2b
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/2p1d-dep4-dep16.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/2p1d-dep4-dep16.yaml
index 35d2aba8e7..334636ebb9 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/2p1d-dep4-dep16.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/1k1k/2p1d-dep4-dep16.yaml
@@ -1,6 +1,5 @@
 name: "qwen3.5-2p1d-dep4-dep16"
 
-setup_script: rebuild-deepep.sh
 
 dynamo:
   hash: 46520ca59afe992fb5ef61b3197b2316f8df9b2b
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/4p1d-dep4-dep16.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/4p1d-dep4-dep16.yaml
index 8b4b802669..7ad092cfd4 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/4p1d-dep4-dep16.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/4p1d-dep4-dep16.yaml
@@ -1,6 +1,5 @@
 name: "qwen3.5-4p1d-dep4-dep16"
 
-setup_script: rebuild-deepep.sh
 
 sbatch_directives:
   mem: "0"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/8p1d-dep4-dep16.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/8p1d-dep4-dep16.yaml
index cf9ca607e2..61da0a4ad8 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/8p1d-dep4-dep16.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-fp8/8k1k/8p1d-dep4-dep16.yaml
@@ -1,6 +1,5 @@
 name: "qwen3.5-8p1d-dep4-dep16"
 
-setup_script: rebuild-deepep.sh
 
 sbatch_directives:
   mem: "0"
diff --git a/runners/launch_gb300-nv.sh b/runners/launch_gb300-nv.sh
index 3e3fa61255..150784fb4e 100644
--- a/runners/launch_gb300-nv.sh
+++ b/runners/launch_gb300-nv.sh
@@ -201,13 +201,6 @@ elif [[ $FRAMEWORK == "dynamo-sglang" && $MODEL_PREFIX == "qwen3.5" ]]; then
     cd "$SRT_REPO_DIR"
     if [[ $PRECISION == "fp8" ]]; then
         git checkout main
-        # The stock configs/rebuild-deepep.sh (setup_script of the wide-EP
-        # fp8 recipes) resolves NVSHMEM by directory name, which can match a
-        # header-less stub in the cu13 nightly image (tilelang's bundled tvm
-        # contrib/nvshmem) and fail the DeepEP build with "nvshmem.h: No
-        # such file or directory". Overlay our header-based resolver.
-        cp "$GITHUB_WORKSPACE/benchmarks/multi_node/srt-slurm-recipes/configs/rebuild-deepep.sh" \
-            configs/rebuild-deepep.sh
     else
         git checkout sa-submission-q2-2026
     fi