Add MiniMax-M2.5 FP8 vLLM benchmark for B200

.github/configs/nvidia-master.yaml

@@ -3106,6 +3106,31 @@ gptoss-fp4-b200-vllm:
     - { tp: 4, conc-start: 4, conc-end: 64 }
     - { tp: 8, conc-start: 4, conc-end: 4 }
 
+minimaxm2.5-fp8-b200-vllm:
+  image: vllm/vllm-openai:v0.16.0-cu130
+  model: MiniMaxAI/MiniMax-M2.5
+  model-prefix: minimaxm2.5
+  runner: b200
+  precision: fp8
+  framework: vllm
+  multinode: false
+  seq-len-configs:
+  - isl: 1024
+    osl: 1024
+    search-space:
+    - { tp: 2, conc-start: 4, conc-end: 64 }
+    - { tp: 4, conc-start: 4, conc-end: 64 }
+  - isl: 1024
+    osl: 8192
+    search-space:
+    - { tp: 2, conc-start: 4, conc-end: 64 }
+    - { tp: 4, conc-start: 4, conc-end: 64 }
+  - isl: 8192
+    osl: 1024
+    search-space:
+    - { tp: 2, conc-start: 4, conc-end: 64 }
+    - { tp: 4, conc-start: 4, conc-end: 64 }
+
 gptoss-fp4-h100-vllm:
   image: vllm/vllm-openai:v0.15.1
   model: openai/gpt-oss-120b

benchmarks/single_node/minimaxm2.5_fp8_b200.sh

@@ -0,0 +1,69 @@
+#!/usr/bin/env bash
+
+source "$(dirname "$0")/../benchmark_lib.sh"
+
+check_env_vars \
+    MODEL \
+    TP \
+    EP_SIZE \
+    CONC \
+    ISL \
+    OSL \
+    MAX_MODEL_LEN \
+    RANDOM_RANGE_RATIO \
+    RESULT_FILENAME
+
+if [[ -n "$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=${PORT:-8888}
+
+export VLLM_USE_FLASHINFER_MOE_FP8=0
+export VLLM_MOE_USE_DEEP_GEMM=0
+
+if [ "$EP_SIZE" -ge 1 ]; then
+  EP=" --enable-expert-parallel"
+else
+  EP=" "
+fi
+
+set -x
+vllm serve $MODEL --port $PORT \
+--tensor-parallel-size=$TP \
+$EP \
+--gpu-memory-utilization 0.95 \
+--max-model-len $MAX_MODEL_LEN \
+--block-size=32 \
+--disable-log-requests \
+--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" --concurrent-requests $CONC
+    append_lm_eval_summary
+fi
+set +x

perf-changelog.yaml

@@ -679,6 +679,7 @@
     - "Image: rocm/sgl-dev:v0.5.8.post1-rocm720-mi35x-20260218"
     - "Uses triton attention backend, TP=8, concurrency 4-64"
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/768
+
 
 - config-keys:
     - qwen3.5-bf16-b200-sglang
@@ -875,3 +876,10 @@
     - "Server: reasoning-parser qwen3, tool-call-parser qwen3_coder, enable-flashinfer-allreduce-fusion, mem-fraction-static 0.8"
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/855
 
+- config-keys:
+    - minimaxm2.5-fp8-b200-vllm
+  description:
+    - "Add MiniMax-M2.5 FP8 vLLM benchmark for B200"
+    - "Model: MiniMaxAI/MiniMax-M2.5 with --trust-remote-code"
+    - "Image: vllm/vllm-openai:v0.16.0-cu130"
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/757