From b0308eda694f3ac7bf6d8a84aa5c8129be2a5440 Mon Sep 17 00:00:00 2001 From: functionstackx <47992694+functionstackx@users.noreply.github.com> Date: Fri, 17 Apr 2026 04:25:25 -0400 Subject: [PATCH 1/2] Add B300 config: minimaxm2.5-fp4-vllm At the time of submission, the vLLM MiniMax-M2 recipes page (https://docs.vllm.ai/projects/recipes/en/latest/MiniMax/MiniMax-M2.html) does not have a B300-specific recipe, so this config reuses the existing MiniMax-M2.5 FP4 (NVFP4) B200 vLLM recipe as-is until B300-specific tuning is available. Co-Authored-By: Claude Opus 4.7 (1M context) --- .github/configs/nvidia-master.yaml | 32 +++++++ .../single_node/minimaxm2.5_fp4_b300.sh | 84 +++++++++++++++++++ perf-changelog.yaml | 8 ++ 3 files changed, 124 insertions(+) create mode 100755 benchmarks/single_node/minimaxm2.5_fp4_b300.sh diff --git a/.github/configs/nvidia-master.yaml b/.github/configs/nvidia-master.yaml index 7ce2824a40..8d5bebf124 100644 --- a/.github/configs/nvidia-master.yaml +++ b/.github/configs/nvidia-master.yaml @@ -3337,6 +3337,38 @@ minimaxm2.5-fp4-b200-vllm: - { tp: 4, conc-start: 4, conc-end: 512 } - { tp: 8, conc-start: 4, conc-end: 4 } +# NOTE: At the time of submission, https://docs.vllm.ai/projects/recipes/en/latest/MiniMax/MiniMax-M2.html +# does not have a B300-specific recipe, so this config reuses the existing +# MiniMax-M2.5 FP4 B200 vLLM recipe as-is until B300-specific tuning is available. +minimaxm2.5-fp4-b300-vllm: + image: vllm/vllm-openai:v0.19.0-cu130 + model: nvidia/MiniMax-M2.5-NVFP4 + model-prefix: minimaxm2.5 + runner: b300 + precision: fp4 + framework: vllm + multinode: false + seq-len-configs: + - isl: 1024 + osl: 1024 + search-space: + - { tp: 1, conc-start: 4, conc-end: 4 } + - { tp: 2, conc-start: 4, conc-end: 512 } + - { tp: 2, ep: 2, conc-start: 128, conc-end: 256 } + - { tp: 2, ep: 2, dp-attn: true, conc-start: 512, conc-end: 512 } + - { tp: 4, conc-start: 4, conc-end: 512 } + - { tp: 4, ep: 4, conc-start: 32, conc-end: 128 } + - { tp: 8, conc-start: 4, conc-end: 4 } + - isl: 8192 + osl: 1024 + search-space: + - { tp: 1, conc-start: 4, conc-end: 32 } + - { tp: 1, conc-start: 256, conc-end: 512 } + - { tp: 2, conc-start: 4, conc-end: 512 } + - { tp: 2, ep: 2, conc-start: 128, conc-end: 512 } + - { tp: 4, conc-start: 4, conc-end: 512 } + - { tp: 8, conc-start: 4, conc-end: 4 } + gptoss-fp4-h100-vllm: image: vllm/vllm-openai:v0.18.0 model: openai/gpt-oss-120b diff --git a/benchmarks/single_node/minimaxm2.5_fp4_b300.sh b/benchmarks/single_node/minimaxm2.5_fp4_b300.sh new file mode 100755 index 0000000000..1647889991 --- /dev/null +++ b/benchmarks/single_node/minimaxm2.5_fp4_b300.sh @@ -0,0 +1,84 @@ +#!/usr/bin/env bash + +# NOTE: At the time of submission, https://docs.vllm.ai/projects/recipes/en/latest/MiniMax/MiniMax-M2.html +# does not have a B300-specific recipe, so this script reuses the existing +# MiniMax-M2.5 FP4 B200 vLLM recipe as-is until B300-specific tuning is available. + +source "$(dirname "$0")/../benchmark_lib.sh" + +check_env_vars \ + MODEL \ + TP \ + EP_SIZE \ + DP_ATTENTION \ + CONC \ + ISL \ + OSL \ + MAX_MODEL_LEN \ + RANDOM_RANGE_RATIO \ + RESULT_FILENAME + +if [[ -n "$SLURM_JOB_ID" ]]; then + echo "JOB $SLURM_JOB_ID running on $SLURMD_NODENAME" +fi + +nvidia-smi + +hf download "$MODEL" + +SERVER_LOG=/workspace/server.log +PORT=${PORT:-8888} + +if [ "${DP_ATTENTION}" = "true" ]; then + PARALLEL_ARGS="--tensor-parallel-size=1 --data-parallel-size=$TP --enable-expert-parallel" +elif [ "$EP_SIZE" -gt 1 ]; then + PARALLEL_ARGS="--tensor-parallel-size=$TP --enable-expert-parallel" +else + PARALLEL_ARGS="--tensor-parallel-size=$TP" +fi + +if [ "${EVAL_ONLY}" = "true" ]; then + setup_eval_context + MAX_MODEL_LEN="$EVAL_MAX_MODEL_LEN" +fi +# Start GPU monitoring (power, temperature, clocks every second) +start_gpu_monitor + +set -x +vllm serve $MODEL --port $PORT \ +$PARALLEL_ARGS \ +--gpu-memory-utilization 0.90 \ +--max-model-len $MAX_MODEL_LEN \ +--kv-cache-dtype fp8 \ +--max-cudagraph-capture-size 2048 \ +--max-num-batched-tokens "$((ISL * 2 ))" \ +--stream-interval 20 --no-enable-prefix-caching \ +--trust-remote-code > $SERVER_LOG 2>&1 & + +SERVER_PID=$! + +# Wait for server to be ready +wait_for_server_ready --port "$PORT" --server-log "$SERVER_LOG" --server-pid "$SERVER_PID" + +run_benchmark_serving \ + --model "$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 "$RESULT_FILENAME" \ + --result-dir /workspace/ \ + --trust-remote-code + +# After throughput, run evaluation only if RUN_EVAL is true +if [ "${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 e3cbeea619..29a690a77f 100644 --- a/perf-changelog.yaml +++ b/perf-changelog.yaml @@ -1437,3 +1437,11 @@ - "EAGLE speculative decoding with MTP, TP=8, concurrency 4-512 for 1k1k and 8k1k" - "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 MTP recipe as-is" pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1059 + +- config-keys: + - minimaxm2.5-fp4-b300-vllm + description: + - "Add MiniMax-M2.5 FP4 (NVFP4) B300 vLLM benchmark" + - "Image: vllm/vllm-openai:v0.19.0-cu130" + - "At the time of submission, https://docs.vllm.ai/projects/recipes/en/latest/MiniMax/MiniMax-M2.html does not have a B300-specific recipe, so this reuses the existing MiniMax-M2.5 FP4 B200 vLLM recipe as-is" + pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXXX From f1bdccf1f65faf4e5523133bb701c09b9291a512 Mon Sep 17 00:00:00 2001 From: functionstackx <47992694+functionstackx@users.noreply.github.com> Date: Fri, 17 Apr 2026 04:25:57 -0400 Subject: [PATCH 2/2] Fill in PR link for minimaxm2.5-fp4-b300-vllm changelog entry Co-Authored-By: Claude Opus 4.7 (1M context) --- perf-changelog.yaml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/perf-changelog.yaml b/perf-changelog.yaml index 29a690a77f..1ed802e236 100644 --- a/perf-changelog.yaml +++ b/perf-changelog.yaml @@ -1444,4 +1444,4 @@ - "Add MiniMax-M2.5 FP4 (NVFP4) B300 vLLM benchmark" - "Image: vllm/vllm-openai:v0.19.0-cu130" - "At the time of submission, https://docs.vllm.ai/projects/recipes/en/latest/MiniMax/MiniMax-M2.html does not have a B300-specific recipe, so this reuses the existing MiniMax-M2.5 FP4 B200 vLLM recipe as-is" - pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXXX + pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1055