Skip to content

[AMD] DeepSeek-V4 FP4 MI355X vLLM STP: bump image to latest nightly / DeepSeek-V4 FP4 MI355X vLLM STP：升级镜像至最新 nightly #1980

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
functionstackx merged 7 commits into from
Jul 4, 2026
Merged

[AMD] DeepSeek-V4 FP4 MI355X vLLM STP: bump image to latest nightly / DeepSeek-V4 FP4 MI355X vLLM STP：升级镜像至最新 nightly #1980

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
ecb56e1
[AMD] DeepSeek-V4 FP4 MI355X vLLM STP: bump image to latest nightly
Fangzhou-Ai Jul 2, 2026
dd5653f
chore(changelog): set pr-link for dsv4-fp4-mi355x-vllm image bump (#1…
Fangzhou-Ai Jul 2, 2026
50961ac
docs(changelog): note two-stage attention kernels and AITER MLA for d…
Fangzhou-Ai Jul 2, 2026
9260991
docs(changelog): two-stage attention improves across all concurrency …
Fangzhou-Ai Jul 2, 2026
fe4b460
[AMD] dsv4 fp4 mi355x vllm STP: use AITER MoE backend
Fangzhou-Ai Jul 2, 2026
f20db0f
Merge remote-tracking branch 'origin/main' into amd/dsv4-fp4-mi355x-v…
Fangzhou-Ai Jul 2, 2026
cfddf3a
Merge remote-tracking branch 'origin/main' into pr-1980-reuse-47330
functionstackx Jul 4, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion .github/configs/amd-master.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1952,7 +1952,7 @@
# gpu-mem-util=0.6. TP8 sweeps conc 4-64; DEP8 has a single conc=64
# probe to validate the ROCm DP+EP path.
dsv4-fp4-mi355x-vllm:
image: vllm/vllm-openai-rocm:v0.22.0
image: vllm/vllm-openai-rocm:nightly-09663abde0f50944a8d5ea30120666024b503faa

Check warning on line 1955 in .github/configs/amd-master.yaml

View check run for this annotation

Claude / Claude Code Review

Stale MTP-variant comment after base image bump 

Nit / pre-existing (feel free to address in a follow-up): the comment above `dsv4-fp4-mi355x-vllm-mtp` (around lines 1979-1980, outside this diff) states "Image reuses the base entry's v0.22.0 ROCm build, which already contains the MTP commit." After this bump the base moves to `nightly-09663abde...` while MTP stays on `v0.22.0`, so "reuses the base entry" is no longer accurate — the two images have intentionally diverged. Consider updating the comment to note the deliberate divergence, or fold
model: deepseek-ai/DeepSeek-V4-Pro
model-prefix: dsv4
runner: mi355x
Expand Down
6 changes: 6 additions & 0 deletions perf-changelog.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4400,3 +4400,9 @@
description:
- "Bump SGLang image from lmsysorg/sglang:deepseek-v4-blackwell (digest sha256:df18bfc4...) to mainline nightly lmsysorg/sglang:nightly-dev-cu13-20260628-da802ddc."
pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1923

- config-keys:
- dsv4-fp4-mi355x-vllm
description:
- "Bump DeepSeek-V4-Pro FP4 MI355X single-node vLLM STP image from vllm/vllm-openai-rocm:v0.22.0 to the latest nightly vllm/vllm-openai-rocm:nightly-09663abde0f50944a8d5ea30120666024b503faa."
pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/PLACEHOLDER

Check failure on line 4408 in perf-changelog.yaml

View check run for this annotation

Claude / Claude Code Review

Invalid PR-link placeholder blocks changelog validation 

This changelog entry uses `pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/PLACEHOLDER`, which is neither the canonical URL for PR #1980 nor one of the whitelisted `XXX` placeholders accepted by `utils/validate_perf_changelog.py`. As-is, `utils/merge_with_reuse.sh`'s canonicalize step will raise `ChangelogValidationError` and block the merge. Replace `PLACEHOLDER` with `1980` (or `XXX`).