[blocked by vllm#45879] MiniMax-M3 MXFP8 full sweep config for GB200#1734
Open
Oseltamivir wants to merge 33 commits into
Open
[blocked by vllm#45879] MiniMax-M3 MXFP8 full sweep config for GB200#1734Oseltamivir wants to merge 33 commits into
Oseltamivir wants to merge 33 commits into
Conversation
Add minimaxm3-fp8-gb200-dynamo-vllm to nvidia-master.yaml with 6
topologies covering the full concurrency range:
- TP4/TP8 (low latency, conc 4-64)
- TP4+EP4 agg + 1P+1D disagg (mid curve, conc 64-512)
- DEP4/DEP8 (high throughput, conc 256-2048)
All recipe YAMLs included under minimax-m3-gb200-fp8/{1k1k,8k1k}/.
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
Adopt the NVIDIA Dynamo vLLM runtime image (nvcr.io/nvidia/ai-dynamo/vllm-runtime:1.3.0-minimax-m3-dev.1), the canonical M3 runtime from ai-dynamo/dynamo release/1.3.0-minimax-m3-dev.1. Changes mirrored from that release's recipes/minimax-m3/vllm/disagg/MXFP8/deploy.yaml: - dynamo.install: false — the runtime image bundles dynamo 1.3.0, so the prior 1.2.0 wheel install is dropped (srtctl defaults install=true) - attention-backend: FLASH_ATTN on every prefill/decode/agg engine Benchmark-specific knobs kept over the reference's serving defaults: language-model-only (text-only), no-enable-prefix-caching (random data), scenario-trimmed max-model-len.
enroot's docker:// URI needs `#` to separate the registry host from
the image path; `nvcr.io/...` was parsed as a Docker Hub repo and 401'd
against registry-1.docker.io. Matches the existing nvcr.io# convention
in nvidia-master.yaml. Recipe container fields kept byte-identical to
the master image: field (srtslurm.yaml maps "${IMAGE}" -> squashfile).
Replace the mostly-aggregated GB200 sweep (5 agg + 1 disagg) with a fully disaggregated sweep that splits prefill/decode over NixlConnector, mirroring the minimaxm2.5-fp8-gb200 reference. Every worker = one 4-GPU node since the 444 GB MXFP8 checkpoint can't fit in fewer. Topologies (1k1k): 1P1D TP4 (low-lat), 1P1D TP4+EP4 (mid), 1P2D TP4+EP4 (decode-scaled), 2P1D TP4+EP4 (prefill-scaled), 1P1D DEP4 (max-tput), spanning conc 4-2048. - add 4 disagg recipes; remove 8 orphaned agg recipes (1k1k + 8k1k) - rewire nvidia-master.yaml search-space to the 5 disagg entries - perf-changelog: describe disagg sweep; fix stale Image line (vllm/vllm-openai:minimax-m3 -> nvcr.io#.../vllm-runtime:1.3.0-minimax-m3-dev.1) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
… transfer Run 27478698552 failed: every disagg worker crashed at NixlConnector init with "NIXL is not available" (RuntimeError, vllm .../nixl/worker.py:248). The ai-dynamo vllm-runtime:1.3.0-minimax-m3-dev.1 image ships dynamo but NOT the nixl bindings (cupy missing too), so kv_connector=NixlConnector cannot initialize and the engine core never becomes healthy. Revert to the pre-ed63c1e0 runtime path that pulls NIXL in via the dynamo wheel (same as the working minimaxm2.5-gb200 disagg recipes): - image/container: vllm/vllm-openai:minimax-m3 (the m3_release build all other m3 entries already use) - dynamo.install=true + wheel 1.2.0.dev20260526 (nixl is a dynamo dep) - keep attention-backend FLASH_ATTN (added in the image-switch commit) Also enable NVLink (MNNVL) KV transfer so NIXL doesn't fall back to TCP, mirroring the deepseek-v4 gb200 disagg recipes — on every prefill/decode env block: UCX_TLS=cuda_copy,cuda_ipc,tcp UCX_CUDA_IPC_ENABLE_MNNVL=y UCX_MEMTYPE_CACHE=n / UCX_MEMTYPE_REG_WHOLE=n NCCL_CUMEM_ENABLE=1 (cuMem-allocate buffers so they are IPC-exportable) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The narrow DEP8-max sweep showed no GB200 advantage over B200 because both cap at an 8-GPU NVLink island. Exploit NVL72's rack-scale NVLink with wide expert parallelism spanning multiple nodes, mirroring the deepseek-v4 "megamoe" ladder (DEP = data-parallel attention + expert-parallel): - 1P1D TP4 (2n) low-latency, conc 4-64 - 1P1D DEP8 (4n) mid, EP8/16-experts-per-rank, conc 128-512 - 1P1D DEP8->DEP16 (6n) wide decode (EP16), conc 512-2048 - 2P1D DEP8->DEP16 (8n) prefill-scaled, conc 2048-4096 - 4P1D DEP8->DEP16 (12n) max throughput, conc 4096-8192 M3 has 128 routed experts (top-4), so EP8/EP16 shard cleanly. EP16 across 16 GPU / 4 nodes is the regime B200 physically can't reach. Attention: FLASH_ATTN -> FLASHINFER (trtllm-gen) on all GB200 recipes to exploit Blackwell. Requires the :minimax-m3 image rebuilt from m3_release HEAD 022448dd (vllm-project/vllm#45381), which gates trtllm-gen page>=128. Also add GB200 perf/NVLink-KV knobs from the deepseek-v4 reference: numa-bind (Grace) and enable-sleep-mode (cuMem allocator so the KV cache is IPC-exportable over the MNNVL fabric), alongside the existing UCX MNNVL env. Replaces the four narrow EP4 recipes; keeps 1P1D TP4 for low latency. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
1k1k TP4 low-conc tuning: stream-interval 1 (was 128 decode / 32 prefill), cudagraph cap 128 (was 512), conc range extended to 1-64 (was 4-64) to match B200 coverage. 8k1k sweep: 5 disagg recipes mirroring the 1k1k megamoe ladder (TP4, DEP8, DEP8→DEP16, 2P1D, 4P1D) with max-model-len 9472 (74×128 blocks = ISL+OSL+256 headroom). Concurrencies shifted ~4x lower for 8x heavier prefill: TP4 1-16, DEP8 32-128, DEP8→DEP16 128-512, 2P1D 512-1024, 4P1D 1024-2048. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Comment out all conc > 64 entries (1k1k DEP8/DEP16/2P1D/4P1D and all
8k1k high-conc) to focus sweep budget on low-concurrency tuning.
Add two new 1k1k experiments at conc 1-64 alongside the existing
1P1D TP4 baseline:
- 1P2D TP4 (3 nodes): 2 decode workers halve per-worker batch
- 1P1D TP4→TP8 (3 nodes): wider decode TP spreads forward pass
across 8 GPU over NVL72
All three share the low-conc tuning (stream-interval 1, cudagraph
cap 128, FLASHINFER, block-size 128).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…onc gap B200 TEP8 (TP8+EP8) achieves 11.68ms TPOT at conc 1 vs GB200 TP8's 15.29ms — the gap is entirely from expert parallelism splitting 128 MoE experts across 8 ranks. Add enable-expert-parallel: true to the TP8 decode recipe and update nvidia-master.yaml decode ep: 1→8 so result JSON reflects TEP8. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
… ISL GB200 8k1k only had TP4 (2n) giving 18.50ms TPOT at conc 1 vs B200 TEP8's 11.57ms. Add 1P1D TP4→TEP8 (3n) 8k1k recipe mirroring the 1k1k TEP8 config that already closed the gap there (12.34ms vs 11.68ms). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Drop 1P1D-TP4 (2n) and 1P2D-TP4 (3n) entries from both 1k1k and 8k1k. TEP8 dominates at every concurrency — TP4 baseline is 50% slower at conc 1 and 1P2D gave <2% TPOT improvement for 50% more GPUs. Active sweep is now TEP8-only: 1k1k conc 1-64, 8k1k conc 1-16. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Enable DEP8 (4n), DEP8→DEP16 (6n), 2P1D (8n), 4P1D (12n) for both 1k1k and 8k1k alongside the optimized TEP8 low-conc configs. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Test whether EP4 on 4 decode GPUs (2 nodes total) improves TPOT over pure TP4 on GB200's NVL72 NVLink. B200 showed TEP4 slightly worse than TP4 intra-node; NVL72 all-to-all may differ. All other entries commented out for this isolated test. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes and found 1 potential issue.
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit 7062524. Configure here.
|
|
||
| dynamo: | ||
| install: true | ||
| wheel: "1.2.0.dev20260526" |
There was a problem hiding this comment.
Missing Nixl patch mounts
High Severity
The 1k1k disagg-gb200-1p1d-tp4-tp8-3n recipe omits extra_mount bind mounts for the Nixl worker.py and fused_allreduce_gemma_rms_norm.py patches, while the matching 8k1k recipe includes them. The sweep uses this 1k1k file for TP4→TEP8 disagg, so jobs can hit NixlConnector handshake failures and zero KV transfer without the documented runtime fixes.
Additional Locations (1)
Reviewed by Cursor Bugbot for commit 7062524. Configure here.
Contributor
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=27648819218 |
Oseltamivir
changed the title
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Summary
minimaxm3-fp8-gb200-dynamo-vllmto nvidia-master.yaml with 6 topologies: TP4, TP8, TP4+EP4, 1P+1D disagg, DEP4, DEP8/mnt/lustre01/models/MiniMax-M3-MXFP8extra_mountandHF_HOMEfrom recipesminimax-m3-gb200-fp8/{1k1k,8k1k}/Test plan
Note
Medium Risk
Large new multinode benchmark surface (12-node jobs, secrets in CI, shared squash locking) but no production serving path; misconfiguration mainly wastes cluster time or fails jobs.
Overview
Adds
minimaxm3-fp8-gb200-dynamo-vllmtonvidia-master.yaml: multinode, fully disaggregated dynamo-vllm on GB200 for MiniMax-M3 MXFP8, with a concurrency ladder across 1k1k and 8k1k (TP4 prefill → TEP8 decode through DEP8/DEP16 and 2P/4P prefill scaling). Each scenario points at new Slurm recipes underbenchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb200-fp8/(NixlConnector, FLASHINFER, block-size 128, Dynamo wheel install, UCX/MNNVL KV over NVL72).The GB200 launcher gains minimaxm3/fp8 staged weights (
/mnt/lustre01/models/MiniMax-M3-MXFP8), copies M3 recipes into srt-slurm, and reuses the watchtower/shared-FS paths previously used for minimaxm2.5. Serializedimport_squashavoids corrupting shared enroot images when matrix jobs run in parallel.benchmark-multinode-tmpl.ymlsetsHF_TOKENfrom a secret so Slurm workers can pull large Hub snapshots without anonymous rate limits.perf-changelog.yamldocuments an evals-only note for the same config key (hetero-TP NixlConnector handshake / optional lustre extra_mount patches on one 8k1k recipe).Reviewed by Cursor Bugbot for commit 7062524. Bugbot is set up for automated code reviews on this repo. Configure here.