feat(hunyuan-image3): optimize four-GPU hybrid inference - #1412
Closed
Chernobyllight wants to merge 1 commit into
Closed
feat(hunyuan-image3): optimize four-GPU hybrid inference#1412Chernobyllight wants to merge 1 commit into
Chernobyllight wants to merge 1 commit into
Conversation
Add a production-ready HunyuanImage 3.0 path for the classic four-GPU topology. The AR phase runs TP4 with native-GQA paged FlashAttention 3 decode, full-decode CUDA Graph capture, and graph-aware vLLM custom all-reduce. The denoising phase remains TP2+SP2 and uses native-GQA FlashAttention 3 over NCCL collectives. Include portable T2I and TI2I configs and launch scripts, strict configuration validation, persistent paged KV-cache management, and deterministic distributed cleanup. The implementation intentionally excludes benchmark-only timing instrumentation.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Add a production-ready HunyuanImage 3.0 path for the classic four-GPU topology.
The AR phase runs TP4 with native-GQA paged FlashAttention 3 decode, full-decode CUDA Graph capture, and graph-aware vLLM custom all-reduce. The denoising phase remains TP2+SP2 and uses native-GQA FlashAttention 3 over NCCL collectives.
Include portable T2I and TI2I configs and launch scripts, strict configuration validation, persistent paged KV-cache management, and deterministic distributed cleanup. The implementation intentionally excludes benchmark-only timing instrumentation.