[NV] add dsv4-fp4-gb300-dynamo-sglang-mtp-1k1k / 新增 DeepSeek-V4-Pro FP4 GB300 Dynamo SGLang MTP 1k1k 基准测试配置 #1697
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,156 @@ | ||
| name: "dsv4-pro-gb300-disagg-1k1k-mtp-1p1d-dep16-conc8192" | ||
|
|
||
| sbatch_directives: | ||
| cpus-per-task: '144' | ||
| mem: '0' | ||
|
|
||
| dynamo: | ||
| version: 1.2.1 | ||
|
|
||
| frontend: | ||
| type: dynamo | ||
| enable_multiple_frontends: true | ||
| num_additional_frontends: 8 | ||
|
|
||
| model: | ||
| path: deepseek-v4-pro | ||
| container: "lmsysorg/sglang:nightly-dev-cu13-20260710-cfc66e05" | ||
| precision: fp4 | ||
|
|
||
| resources: | ||
| gpu_type: gb300 | ||
| gpus_per_node: 4 | ||
| prefill_nodes: 1 | ||
| prefill_workers: 1 | ||
| gpus_per_prefill: 4 | ||
| decode_nodes: 4 | ||
| decode_workers: 1 | ||
| gpus_per_decode: 16 | ||
|
|
||
| backend: | ||
| type: sglang | ||
|
|
||
| prefill_environment: | ||
| PYTHONUNBUFFERED: '1' | ||
| SGLANG_RADIX_DISABLE_REUSE: '1' | ||
| SGLANG_JIT_DEEPGEMM_FAST_WARMUP: '1' | ||
| SGLANG_DEFAULT_THINKING: '1' | ||
| SGLANG_DSV4_REASONING_EFFORT: max | ||
| SGLANG_OPT_SWA_SPLIT_LEAF_ON_INSERT: '1' | ||
| SGLANG_OPT_USE_JIT_NORM: '1' | ||
| SGLANG_OPT_USE_JIT_INDEXER_METADATA: '1' | ||
| SGLANG_OPT_USE_TOPK_V2: '1' | ||
|
|
||
| SGLANG_OPT_SWA_EVICT_DROP_PAGE_MARGIN: '1' | ||
| SGLANG_OPT_USE_CUSTOM_ALL_REDUCE_V2: '1' | ||
| SGLANG_OPT_USE_FAST_MASK_EP: '1' | ||
| SGLANG_OPT_USE_DEEPGEMM_MEGA_MOE: '1' | ||
| SGLANG_OPT_FIX_HASH_MEGA_MOE: '1' | ||
| SGLANG_OPT_DEEPGEMM_MEGA_MOE_NUM_MAX_TOKENS_PER_RANK: '9216' | ||
| SGLANG_OPT_FIX_MEGA_MOE_MEMORY: '1' | ||
| SGLANG_OPT_FIX_NEXTN_MEGA_MOE: '1' | ||
| SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK: '0' | ||
|
|
||
| NCCL_MNNVL_ENABLE: '1' | ||
| NCCL_CUMEM_ENABLE: '1' | ||
| SGLANG_MOONCAKE_CUSTOM_MEM_POOL: 'True' | ||
| MC_FORCE_MNNVL: '1' | ||
| SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: '100000' | ||
| SGLANG_DISAGGREGATION_WAITING_TIMEOUT: '100000' | ||
| SGLANG_OPT_SWA_RELEASE_LEAF_LOCK_AFTER_WINDOW: '1' | ||
|
|
||
| decode_environment: | ||
| PYTHONUNBUFFERED: '1' | ||
| SGLANG_RADIX_DISABLE_REUSE: '1' | ||
| SGLANG_JIT_DEEPGEMM_FAST_WARMUP: '1' | ||
| SGLANG_DEFAULT_THINKING: '1' | ||
| SGLANG_DSV4_REASONING_EFFORT: max | ||
| SGLANG_OPT_SWA_SPLIT_LEAF_ON_INSERT: '1' | ||
| SGLANG_OPT_USE_JIT_NORM: '1' | ||
| SGLANG_OPT_USE_JIT_INDEXER_METADATA: '1' | ||
| SGLANG_OPT_USE_TOPK_V2: '1' | ||
|
|
||
| SGLANG_OPT_SWA_EVICT_DROP_PAGE_MARGIN: '1' | ||
| SGLANG_OPT_USE_FAST_MASK_EP: '1' | ||
| SGLANG_OPT_USE_DEEPGEMM_MEGA_MOE: '1' | ||
| SGLANG_OPT_FIX_HASH_MEGA_MOE: '1' | ||
| SGLANG_OPT_DEEPGEMM_MEGA_MOE_NUM_MAX_TOKENS_PER_RANK: '4096' | ||
| SGLANG_OPT_FIX_MEGA_MOE_MEMORY: '1' | ||
| SGLANG_OPT_FIX_NEXTN_MEGA_MOE: '1' | ||
| SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK: '0' | ||
|
|
||
| NCCL_MNNVL_ENABLE: '1' | ||
| NCCL_CUMEM_ENABLE: '1' | ||
| SGLANG_MOONCAKE_CUSTOM_MEM_POOL: 'True' | ||
| MC_FORCE_MNNVL: '1' | ||
| SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: '100000' | ||
| SGLANG_DISAGGREGATION_WAITING_TIMEOUT: '100000' | ||
| SGLANG_OPT_SWA_RELEASE_LEAF_LOCK_AFTER_WINDOW: '1' | ||
| SGLANG_OPT_USE_CUSTOM_ALL_REDUCE_V2: "0" # CAR_V2 is single-node only. | ||
|
|
||
| sglang_config: | ||
| prefill: | ||
| served-model-name: deepseek-ai/DeepSeek-V4-Pro | ||
| model-path: "/model/" | ||
| trust-remote-code: true | ||
| tool-call-parser: deepseekv4 | ||
|
|
||
| disaggregation-mode: prefill | ||
| disaggregation-transfer-backend: mooncake | ||
|
|
||
| tensor-parallel-size: 4 | ||
| data-parallel-size: 4 | ||
| expert-parallel-size: 4 | ||
|
|
||
| enable-dp-attention: true | ||
| enable-dp-lm-head: true | ||
|
|
||
| moe-a2a-backend: deepep | ||
| deepep-config: '{"normal_dispatch":{"num_sms":96},"normal_combine":{"num_sms":96}}' | ||
|
|
||
| mem-fraction-static: 0.9 | ||
| max-running-requests: 1024 | ||
| cuda-graph-max-bs: 1024 | ||
| chunked-prefill-size: 32768 | ||
| stream-interval: 60 | ||
|
|
||
| decode: | ||
| served-model-name: deepseek-ai/DeepSeek-V4-Pro | ||
| model-path: "/model/" | ||
| trust-remote-code: true | ||
| tool-call-parser: deepseekv4 | ||
|
|
||
| disaggregation-mode: decode | ||
| disaggregation-transfer-backend: mooncake | ||
|
|
||
| tensor-parallel-size: 16 | ||
| data-parallel-size: 16 | ||
| expert-parallel-size: 16 | ||
|
|
||
| enable-dp-attention: true | ||
| enable-dp-lm-head: true | ||
|
|
||
| moe-a2a-backend: deepep | ||
| deepep-config: '{"normal_dispatch":{"num_sms":96},"normal_combine":{"num_sms":96}}' | ||
|
|
||
| speculative-algo: "EAGLE" | ||
| speculative-num-steps: 3 | ||
| speculative-eagle-topk: 1 | ||
| speculative-num-draft-tokens: 4 | ||
|
|
||
| mem-fraction-static: 0.9 | ||
| max-running-requests: 18432 | ||
| cuda-graph-max-bs: 1024 | ||
| swa-full-tokens-ratio: 0.15 | ||
| context-length: 16384 | ||
| stream-interval: 60 | ||
|
|
||
| benchmark: | ||
| type: "sa-bench" | ||
| isl: 1024 | ||
| osl: 1024 | ||
| random_range_ratio: 0.8 | ||
| concurrencies: "8192" | ||
| req_rate: "inf" | ||
| use_chat_template: true | ||
| custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer" | ||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,157 @@ | ||
| name: "dsv4-pro-gb300-disagg-1k1k-mtp-1p1d-dep8-conc8192" | ||
|
|
||
| sbatch_directives: | ||
| cpus-per-task: '144' | ||
| mem: '0' | ||
|
|
||
| dynamo: | ||
| version: 1.2.1 | ||
|
|
||
| frontend: | ||
| type: dynamo | ||
| enable_multiple_frontends: true | ||
| num_additional_frontends: 8 | ||
|
|
||
| model: | ||
| path: deepseek-v4-pro | ||
| container: "lmsysorg/sglang:nightly-dev-cu13-20260710-cfc66e05" | ||
| precision: fp4 | ||
|
There was a problem hiding this comment. High-conc precision mismatchMedium Severity The three high-throughput 1k1k MTP recipes declare Additional Locations (1)  Reviewed by Cursor Bugbot for commit 7680f41. Configure here. |
||
|
|
||
| resources: | ||
| gpu_type: gb300 | ||
| gpus_per_node: 4 | ||
| prefill_nodes: 1 | ||
| prefill_workers: 1 | ||
| gpus_per_prefill: 4 | ||
| decode_nodes: 2 | ||
| decode_workers: 1 | ||
| gpus_per_decode: 8 | ||
|
|
||
| backend: | ||
| type: sglang | ||
|
|
||
| prefill_environment: | ||
| PYTHONUNBUFFERED: '1' | ||
| SGLANG_RADIX_DISABLE_REUSE: '1' | ||
| SGLANG_JIT_DEEPGEMM_FAST_WARMUP: '1' | ||
| SGLANG_DEFAULT_THINKING: '1' | ||
| SGLANG_DSV4_REASONING_EFFORT: max | ||
| SGLANG_OPT_SWA_SPLIT_LEAF_ON_INSERT: '1' | ||
| SGLANG_OPT_USE_JIT_NORM: '1' | ||
| SGLANG_OPT_USE_JIT_INDEXER_METADATA: '1' | ||
| SGLANG_OPT_USE_TOPK_V2: '1' | ||
|
|
||
| SGLANG_OPT_SWA_EVICT_DROP_PAGE_MARGIN: '1' | ||
| SGLANG_OPT_USE_CUSTOM_ALL_REDUCE_V2: '1' | ||
| SGLANG_OPT_USE_FAST_MASK_EP: '1' | ||
| SGLANG_OPT_USE_DEEPGEMM_MEGA_MOE: '1' | ||
| SGLANG_OPT_FIX_HASH_MEGA_MOE: '1' | ||
| SGLANG_OPT_DEEPGEMM_MEGA_MOE_NUM_MAX_TOKENS_PER_RANK: '9216' | ||
| SGLANG_OPT_FIX_MEGA_MOE_MEMORY: '1' | ||
| SGLANG_OPT_FIX_NEXTN_MEGA_MOE: '1' | ||
| SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK: '0' | ||
|
|
||
| NCCL_MNNVL_ENABLE: '1' | ||
| NCCL_CUMEM_ENABLE: '1' | ||
| SGLANG_MOONCAKE_CUSTOM_MEM_POOL: 'True' | ||
| MC_FORCE_MNNVL: '1' | ||
| SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: '100000' | ||
| SGLANG_DISAGGREGATION_WAITING_TIMEOUT: '100000' | ||
| SGLANG_OPT_SWA_RELEASE_LEAF_LOCK_AFTER_WINDOW: '1' | ||
|
|
||
| decode_environment: | ||
| PYTHONUNBUFFERED: '1' | ||
| SGLANG_RADIX_DISABLE_REUSE: '1' | ||
| SGLANG_JIT_DEEPGEMM_FAST_WARMUP: '1' | ||
| SGLANG_DEFAULT_THINKING: '1' | ||
| SGLANG_DSV4_REASONING_EFFORT: max | ||
| SGLANG_OPT_SWA_SPLIT_LEAF_ON_INSERT: '1' | ||
| SGLANG_OPT_USE_JIT_NORM: '1' | ||
| SGLANG_OPT_USE_JIT_INDEXER_METADATA: '1' | ||
| SGLANG_OPT_USE_TOPK_V2: '1' | ||
|
|
||
| SGLANG_OPT_SWA_EVICT_DROP_PAGE_MARGIN: '1' | ||
| SGLANG_OPT_USE_FAST_MASK_EP: '1' | ||
| SGLANG_OPT_USE_DEEPGEMM_MEGA_MOE: '1' | ||
| SGLANG_OPT_FIX_HASH_MEGA_MOE: '1' | ||
| SGLANG_OPT_DEEPGEMM_MEGA_MOE_NUM_MAX_TOKENS_PER_RANK: '2048' | ||
| SGLANG_OPT_FIX_MEGA_MOE_MEMORY: '1' | ||
| SGLANG_OPT_FIX_NEXTN_MEGA_MOE: '1' | ||
| SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK: '0' | ||
|
|
||
| NCCL_MNNVL_ENABLE: '1' | ||
| NCCL_CUMEM_ENABLE: '1' | ||
| SGLANG_MOONCAKE_CUSTOM_MEM_POOL: 'True' | ||
| MC_FORCE_MNNVL: '1' | ||
| SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: '100000' | ||
| SGLANG_DISAGGREGATION_WAITING_TIMEOUT: '100000' | ||
| SGLANG_OPT_SWA_RELEASE_LEAF_LOCK_AFTER_WINDOW: '1' | ||
| SGLANG_OPT_USE_CUSTOM_ALL_REDUCE_V2: "0" # CAR_V2 is single-node only. | ||
| TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: "1800" | ||
|
|
||
| sglang_config: | ||
| prefill: | ||
| served-model-name: deepseek-ai/DeepSeek-V4-Pro | ||
| model-path: "/model/" | ||
| trust-remote-code: true | ||
| tool-call-parser: deepseekv4 | ||
|
|
||
| disaggregation-mode: prefill | ||
| disaggregation-transfer-backend: mooncake | ||
|
|
||
| tensor-parallel-size: 4 | ||
| data-parallel-size: 4 | ||
| expert-parallel-size: 4 | ||
|
|
||
| enable-dp-attention: true | ||
| enable-dp-lm-head: true | ||
|
|
||
| moe-a2a-backend: deepep | ||
| deepep-config: '{"normal_dispatch":{"num_sms":96},"normal_combine":{"num_sms":96}}' | ||
|
|
||
| mem-fraction-static: 0.9 | ||
| max-running-requests: 256 | ||
| cuda-graph-max-bs: 256 | ||
|
There was a problem hiding this comment. Dep8 prefill cap too lowMedium Severity In Reviewed by Cursor Bugbot for commit e48f6b3. Configure here. |
||
| chunked-prefill-size: 32768 | ||
| stream-interval: 60 | ||
|
|
||
| decode: | ||
| served-model-name: deepseek-ai/DeepSeek-V4-Pro | ||
| model-path: "/model/" | ||
| trust-remote-code: true | ||
| tool-call-parser: deepseekv4 | ||
|
|
||
| disaggregation-mode: decode | ||
| disaggregation-transfer-backend: mooncake | ||
|
|
||
| tensor-parallel-size: 8 | ||
| data-parallel-size: 8 | ||
| expert-parallel-size: 8 | ||
|
|
||
| enable-dp-attention: true | ||
| enable-dp-lm-head: true | ||
|
|
||
| moe-a2a-backend: deepep | ||
| deepep-config: '{"normal_dispatch":{"num_sms":96},"normal_combine":{"num_sms":96}}' | ||
|
|
||
| speculative-algo: "EAGLE" | ||
| speculative-num-steps: 3 | ||
| speculative-eagle-topk: 1 | ||
| speculative-num-draft-tokens: 4 | ||
|
|
||
| mem-fraction-static: 0.94 | ||
| max-running-requests: 1536 | ||
|
cursor[bot] marked this conversation as resolved.
|
||
| cuda-graph-max-bs: 256 | ||
| swa-full-tokens-ratio: 0.15 | ||
| context-length: 16384 | ||
| stream-interval: 60 | ||
|
|
||
| benchmark: | ||
| type: "sa-bench" | ||
| isl: 1024 | ||
| osl: 1024 | ||
| random_range_ratio: 0.8 | ||
| concurrencies: "8192" | ||
| req_rate: "inf" | ||
| use_chat_template: true | ||
| custom_tokenizer: "sa_bench_tokenizers.sglang_deepseek_v4.SGLangDeepseekV4Tokenizer" | ||
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Missing distributed timeout dep16
Medium Severity
disagg-1p1d-dep8-conc8192-mtp.yamlsetsTORCH_DISTRIBUTED_DEFAULT_TIMEOUTto1800indecode_environmentfor multi-node decode, but the newdep16recipes (four decode nodes, TP16) omit it. Longer multi-node decode init can hit the default shorter timeout.Additional Locations (1)
benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/1k1k/disagg-2p1d-dep16-conc8192-mtp.yaml#L65-L93Reviewed by Cursor Bugbot for commit 37100c4. Configure here.