From 77f602be692735883794261cea76666b30d25d0d Mon Sep 17 00:00:00 2001
From: functionstackx <47992694+functionstackx@users.noreply.github.com>
Date: Fri, 17 Apr 2026 02:04:01 -0400
Subject: [PATCH 1/2] Add B300 config: dsr1-fp8-sglang (non-MTP)

At the time of submission, the SGLang DSR1 cookbook does not have a
B300-specific recipe, so this config reuses the existing B200 DSR1 FP8
SGLang recipe as-is until B300-specific tuning is available. Image bumped
to v0.5.10.post1-cu130 to match the standard B300 SGLang image used by
other B300 configs.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 .github/configs/nvidia-master.yaml      |  22 +++++
 benchmarks/single_node/dsr1_fp8_b300.sh | 113 ++++++++++++++++++++++++
 perf-changelog.yaml                     |   8 ++
 3 files changed, 143 insertions(+)
 create mode 100644 benchmarks/single_node/dsr1_fp8_b300.sh

diff --git a/.github/configs/nvidia-master.yaml b/.github/configs/nvidia-master.yaml
index 47d66aa2d..3015cbf16 100644
--- a/.github/configs/nvidia-master.yaml
+++ b/.github/configs/nvidia-master.yaml
@@ -1774,6 +1774,28 @@ dsr1-fp8-b200-sglang:
     - { tp: 8, ep: 1, conc-start: 4, conc-end: 4 }
     - { tp: 4, ep: 1, conc-start: 4, conc-end: 32 }
 
+# NOTE: At the time of submission, https://cookbook.sglang.io/autoregressive/DeepSeek/DeepSeek-R1
+# does not have a B300-specific recipe, so this config reuses the existing DSR1 FP8
+# B200 SGLang recipe as-is until B300-specific tuning is available.
+dsr1-fp8-b300-sglang:
+  image: lmsysorg/sglang:v0.5.10.post1-cu130
+  model: deepseek-ai/DeepSeek-R1-0528
+  model-prefix: dsr1
+  runner: b300
+  precision: fp8
+  framework: sglang
+  multinode: false
+  seq-len-configs:
+  - isl: 1024
+    osl: 1024
+    search-space:
+    - { tp: 8, ep: 1, conc-start: 4, conc-end: 64 }
+  - isl: 8192
+    osl: 1024
+    search-space:
+    - { tp: 8, ep: 1, conc-start: 4, conc-end: 4 }
+    - { tp: 4, ep: 1, conc-start: 4, conc-end: 32 }
+
 qwen3.5-bf16-b200-sglang:
   image: lmsysorg/sglang:nightly-dev-20260216-d3bae71e
   model: Qwen/Qwen3.5-397B-A17B
diff --git a/benchmarks/single_node/dsr1_fp8_b300.sh b/benchmarks/single_node/dsr1_fp8_b300.sh
new file mode 100644
index 000000000..f0ebda29e
--- /dev/null
+++ b/benchmarks/single_node/dsr1_fp8_b300.sh
@@ -0,0 +1,113 @@
+#!/usr/bin/env bash
+
+# NOTE: At the time of submission, https://cookbook.sglang.io/autoregressive/DeepSeek/DeepSeek-R1
+# does not have a B300-specific recipe, so this script reuses the existing
+# DSR1 FP8 B200 SGLang recipe as-is until B300-specific tuning is available.
+
+source "$(dirname "$0")/../benchmark_lib.sh"
+
+check_env_vars \
+    MODEL \
+    TP \
+    CONC \
+    ISL \
+    OSL \
+    RANDOM_RANGE_RATIO \
+    RESULT_FILENAME \
+    EP_SIZE
+
+if [[ -n "$SLURM_JOB_ID" ]]; then
+  echo "JOB $SLURM_JOB_ID running on $SLURMD_NODENAME"
+fi
+
+nvidia-smi
+
+hf download "$MODEL"
+
+export SGL_ENABLE_JIT_DEEPGEMM=false
+export SGLANG_ENABLE_FLASHINFER_GEMM=true
+SERVER_LOG=/workspace/server.log
+PORT=${PORT:-8888}
+
+# Default: recv every ~10 requests; if CONC ≥ 16, relax to ~30 requests between scheduler recv polls.
+if [[ $TP -eq 8 ]]; then
+  if [[ $CONC -ge 16 ]]; then
+    SCHEDULER_RECV_INTERVAL=30
+  else
+    SCHEDULER_RECV_INTERVAL=10
+  fi
+
+  # Setting these values (passed in to --cuda-graph-max-bs and --max-running-requests) as the maximum concurrency
+  # this will help us save memory from being unnecessary used.
+  MAX_RUNNING_REQUESTS=128
+  CUDA_GRAPH_MAX_BATCH_SIZE=128
+
+  MEM_FRAC_STATIC=0.82
+  CHUNKED_PREFILL_SIZE=32768
+  MAX_PREFILL_TOKENS=32768
+elif [[ $TP -eq 4 ]]; then
+  if [[ $ISL -ne 8192 ]] || [[ $OSL -ne 1024 ]]; then
+    echo "TP=4 not yet supported for ISL=$ISL OSL=$OSL!"
+    exit 1
+  fi
+
+  # Setting these values (passed in to --cuda-graph-max-bs and --max-running-requests) as the maximum concurrency
+  # this will help us save memory from being unnecessary used.
+  MAX_RUNNING_REQUESTS=32
+  CUDA_GRAPH_MAX_BATCH_SIZE=32
+
+  MEM_FRAC_STATIC=0.95
+  CHUNKED_PREFILL_SIZE=8192
+  MAX_PREFILL_TOKENS=8192
+
+  SCHEDULER_RECV_INTERVAL=10
+else
+  echo "Unrecognized TP size $TP!"
+  exit 1
+fi
+echo "SCHEDULER_RECV_INTERVAL: $SCHEDULER_RECV_INTERVAL, CONC: $CONC, ISL: $ISL, OSL: $OSL"
+
+EVAL_CONTEXT_ARGS=""
+if [ "${EVAL_ONLY}" = "true" ]; then
+    setup_eval_context
+    EVAL_CONTEXT_ARGS="--context-length $EVAL_MAX_MODEL_LEN"
+fi
+# Start GPU monitoring (power, temperature, clocks every second)
+start_gpu_monitor
+
+set -x
+PYTHONNOUSERSITE=1 python3 -m sglang.launch_server --model-path=$MODEL --host=0.0.0.0 --port=$PORT \
+--tensor-parallel-size=$TP --data-parallel-size=1 \
+--cuda-graph-max-bs $CUDA_GRAPH_MAX_BATCH_SIZE --max-running-requests $MAX_RUNNING_REQUESTS \
+--mem-fraction-static $MEM_FRAC_STATIC --kv-cache-dtype fp8_e4m3 --chunked-prefill-size $CHUNKED_PREFILL_SIZE --max-prefill-tokens $MAX_PREFILL_TOKENS \
+--enable-flashinfer-allreduce-fusion --scheduler-recv-interval $SCHEDULER_RECV_INTERVAL --disable-radix-cache \
+--attention-backend trtllm_mla --stream-interval 30 --ep-size $EP_SIZE --moe-runner-backend flashinfer_trtllm --quantization fp8 $EVAL_CONTEXT_ARGS > $SERVER_LOG 2>&1 &
+
+SERVER_PID=$!
+
+# Wait for server to be ready
+wait_for_server_ready --port "$PORT" --server-log "$SERVER_LOG" --server-pid "$SERVER_PID"
+
+pip install -q datasets pandas
+
+run_benchmark_serving \
+    --model "$MODEL" \
+    --port "$PORT" \
+    --backend vllm \
+    --input-len "$ISL" \
+    --output-len "$OSL" \
+    --random-range-ratio "$RANDOM_RANGE_RATIO" \
+    --num-prompts "$((CONC * 10))" \
+    --max-concurrency "$CONC" \
+    --result-filename "$RESULT_FILENAME" \
+    --result-dir /workspace/
+
+# After throughput, run evaluation only if RUN_EVAL is true
+if [ "${RUN_EVAL}" = "true" ]; then
+    run_eval --framework lm-eval --port "$PORT"
+    append_lm_eval_summary
+fi
+
+# Stop GPU monitoring
+stop_gpu_monitor
+set +x
diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index 75fae41e3..2928b7fa0 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -1404,3 +1404,11 @@
     - "Image: lmsysorg/sglang:v0.5.10.post1-cu130"
     - "At the time of submission, https://cookbook.sglang.io/autoregressive/DeepSeek/DeepSeek-R1 does not have a B300-specific recipe, so this reuses the existing DSR1 FP4 B200 SGLang recipe as-is"
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1049
+
+- config-keys:
+    - dsr1-fp8-b300-sglang
+  description:
+    - "Add DeepSeek-R1-0528 FP8 B300 SGLang benchmark (non-MTP)"
+    - "Image: lmsysorg/sglang:v0.5.10.post1-cu130"
+    - "At the time of submission, https://cookbook.sglang.io/autoregressive/DeepSeek/DeepSeek-R1 does not have a B300-specific recipe, so this reuses the existing DSR1 FP8 B200 SGLang recipe as-is"
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXXX

From 43e2882508e5f1b9025befc4d1f2eac6f6b9ec8f Mon Sep 17 00:00:00 2001
From: functionstackx <47992694+functionstackx@users.noreply.github.com>
Date: Fri, 17 Apr 2026 02:04:23 -0400
Subject: [PATCH 2/2] Fill in PR link for dsr1-fp8-b300-sglang changelog entry

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 perf-changelog.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index 2928b7fa0..67237ec1b 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -1411,4 +1411,4 @@
     - "Add DeepSeek-R1-0528 FP8 B300 SGLang benchmark (non-MTP)"
     - "Image: lmsysorg/sglang:v0.5.10.post1-cu130"
     - "At the time of submission, https://cookbook.sglang.io/autoregressive/DeepSeek/DeepSeek-R1 does not have a B300-specific recipe, so this reuses the existing DSR1 FP8 B200 SGLang recipe as-is"
-  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXXX
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1050