[Power] Require B200 and B300 multinode telemetry / 强制启用 B200 和 B300 多节点功耗采集

.github/workflows/test-process-result.yml

@@ -9,13 +9,18 @@ on:
       - '.github/workflows/test-process-result.yml'
       - 'benchmarks/benchmark_lib.sh'
       - 'benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/**/*.yaml'
+      - 'benchmarks/multi_node/srt-slurm-recipes/vllm/deepseek-v4/**/*.yaml'
+      - 'benchmarks/multi_node/srt-slurm-recipes/vllm/kimi-k2.6/b200-fp4/**/*.yaml'
       - 'benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/agentic/agg-h200-tp8-mtp-kvoffload.yaml'
       - 'benchmarks/multi_node/srt-slurm-recipes/sglang/glm5.2/agentic/disagg-h200-2p2d-pcp8-tp8-dp8-mtp.yaml'
       - 'benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb200-*/**/*.yaml'
       - 'benchmarks/multi_node/srt-slurm-recipes/sglang/qwen3.5/gb300-*/**/*.yaml'
       - 'configs/nvidia-master.yaml'
       - 'runners/launch_gb200-nv.sh'
       - 'runners/launch_gb300-nv.sh'
+      - 'runners/launch_b200-dgxc.sh'
+      - 'runners/launch_b200-nscale-slurm.sh'
+      - 'runners/launch_b300-nv.sh'
       - 'runners/launch_h200-dgxc-slurm.sh'
       - 'runners/inject_srt_power_concurrencies.py'
       - 'utils/aggregate_power.py'
@@ -30,6 +35,7 @@ on:
       - 'utils/process_result.py'
       - 'utils/test_aggregate_power.py'
       - 'utils/test_aggregate_power_multinode.py'
+      - 'utils/test_b200_b300_power_official_contract.py'
       - 'utils/test_gb200_power_official_contract.py'
       - 'utils/test_gb300_power_official_contract.py'
       - 'utils/test_h200_power_official_contract.py'
@@ -62,4 +68,4 @@ jobs:
       - name: Run pytest
         run: |
           cd utils
-          python -m pytest test_aggregate_power.py test_aggregate_power_multinode.py agentic/aggregation/test_power_adapter.py agentic/aggregation/test_power_lifecycle.py agentic/aggregation/test_process_agentic_result.py test_gb200_power_official_contract.py test_gb300_power_official_contract.py test_h200_power_official_contract.py test_inject_srt_power_concurrencies.py test_process_result.py -v
+          python -m pytest test_aggregate_power.py test_aggregate_power_multinode.py agentic/aggregation/test_power_adapter.py agentic/aggregation/test_power_lifecycle.py agentic/aggregation/test_process_agentic_result.py test_b200_b300_power_official_contract.py test_gb200_power_official_contract.py test_gb300_power_official_contract.py test_h200_power_official_contract.py test_inject_srt_power_concurrencies.py test_process_result.py -v

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b200-low-latency-1p1d-tp8-tp8-mtp.yaml

@@ -118,3 +118,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 9401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b200-low-latency-1p6d-dep8-tp8-mtp.yaml

@@ -125,3 +125,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 9401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b200-mid-curve-1p1d-dep8-dep8-mtp.yaml

@@ -126,3 +126,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 9401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b200-mid-curve-1p2d-dep8-dep8-mtp.yaml

@@ -137,3 +137,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 9401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b200-mid-curve-5p3d-dep8-dep8-mtp.yaml

@@ -126,3 +126,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 9401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b300-1p1d-dep4-dep8-mtp.yaml

@@ -132,3 +132,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 19401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b300-1p1d-dep4-dep8.yaml

@@ -143,3 +143,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: false
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 19401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b300-1p1d-tp4-tp4-mtp.yaml

@@ -123,3 +123,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 19401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b300-1p1d-tp4-tp4.yaml

@@ -118,3 +118,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: false
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 19401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b300-1p2d-dep4-dep8-mtp.yaml

@@ -133,3 +133,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 19401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b300-1p6d-dep4-tp4-mtp.yaml

@@ -127,3 +127,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 19401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b300-2p1d-dep4-dep8-mtp.yaml

@@ -133,3 +133,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 19401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b300-2p1d-dep4-dep8.yaml

@@ -143,3 +143,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: false
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 19401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b300-4p1d-dep4-dep8-mtp.yaml

@@ -133,3 +133,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 19401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b300-4p1d-dep4-dep8.yaml

@@ -143,3 +143,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: false
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 19401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b300-6p1d-dep4-dep8-mtp.yaml

@@ -133,3 +133,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 19401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b300-6p1d-dep4-dep8.yaml

@@ -143,3 +143,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: false
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 19401

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-b300-8p1d-dep4-dep8-mtp.yaml

@@ -133,3 +133,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 19401

benchmarks/multi_node/srt-slurm-recipes/vllm/deepseek-v4/8k1k/disagg-b200-1p1d-dep8-dep8-mtp.yaml

@@ -31,6 +31,7 @@ frontend:
   type: dynamo
   enable_multiple_frontends: false
 backend:
+  set_cuda_visible_devices: true  # keep the aflowers/vllm-gb200-v0.20.0 provenance: bind GPUs via CVD, not --device-ids
   type: vllm
   connector: null
   prefill_environment:
@@ -127,3 +128,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.vllm_deepseek_v4.VLLMDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 9401

benchmarks/multi_node/srt-slurm-recipes/vllm/deepseek-v4/8k1k/disagg-b200-1p1d-dep8-tp8-mtp.yaml

@@ -31,6 +31,7 @@ frontend:
   type: dynamo
   enable_multiple_frontends: false
 backend:
+  set_cuda_visible_devices: true  # keep the aflowers/vllm-gb200-v0.20.0 provenance: bind GPUs via CVD, not --device-ids
   type: vllm
   connector: null
   prefill_environment:
@@ -112,3 +113,16 @@ benchmark:
   req_rate: "inf"
   use_chat_template: true
   custom_tokenizer: "sa_bench_tokenizers.vllm_deepseek_v4.VLLMDeepseekV4Tokenizer"
+
+telemetry:
+  enabled: true
+  provider: dcgm-power
+  default_frequency: 1.0
+  storage_subdir: power
+  required: true
+  startup_timeout_seconds: 120
+  request_timeout_seconds: 2
+  collector_join_timeout_seconds: 10
+  dcgm_exporter:
+    container_image: dcgm-exporter
+    port: 9401