-
Notifications
You must be signed in to change notification settings - Fork 179
Pull requests: SemiAnalysisAI/InferenceX
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
MinimaxM2.5-FP8-MI325x-vLLM: pin AITER FA attention backend
#1594
opened May 30, 2026 by
chunfangamd
Collaborator
Loading…
GLM5.1-FP4-MI355X-SGLang: bump Image to v0.5.12.post1-rocm720-mi35x-20260529
#1593
opened May 30, 2026 by
chunfangamd
Collaborator
Loading…
Add SPEED-Bench reference synthetic AL values for DeepSeek-V4-Pro MTP 1-8
#1592
opened May 30, 2026 by
qiching
Loading…
ci(disagg): fail before writing result file + surface real failure class
#1591
opened May 29, 2026 by
arygupt
Collaborator
Loading…
fix(process_result): fail loudly on zero-throughput disagg runs (no more masked ZeroDivisionError)
#1590
opened May 29, 2026 by
arygupt
Collaborator
Loading…
[WIP] Update DSv4 B300 vllm image tag
full-sweep-enabled
#1588
opened May 29, 2026 by
wzhao18
Collaborator
Loading…
Add DSV4 GB300 wide-EP sweep configs (EP=12/16/24/32/40)
full-sweep-enabled
#1586
opened May 29, 2026 by
yhyang201
Collaborator
Loading…
[Fix] Remove MoRI-IO patches from vLLM Disagg benchmarks
#1585
opened May 29, 2026 by
simondanielsson
Collaborator
•
Draft
[AMD] improve dsr1 fp4 disagg
AMD
full-sweep-enabled
#1584
opened May 29, 2026 by
billishyahao
Collaborator
Loading…
kimik2.5-fp4-gb200-dynamo-vllm: bump vLLM image to v0.21.0
full-sweep-enabled
#1582
opened May 28, 2026 by
Ankur-singh
Collaborator
Loading…
[AMD][MI355X] add the kimik2.5_int4_mi355x_vllm-disagg support for AMD GPU.
sweep-enabled
#1581
opened May 28, 2026 by
haic0
Collaborator
Loading…
feat(power): per-worker prefill/decode power + role-split joules (stacked on #1574)
#1577
opened May 28, 2026 by
arygupt
Collaborator
Loading…
1 of 3 tasks
[NV] Update B300 DSV4 SGLang Pareto sweep
full-sweep-enabled
#1575
opened May 27, 2026 by
Ankur-singh
Collaborator
Loading…
feat(power): multinode measured-power aggregation
full-sweep-enabled
#1574
opened May 27, 2026 by
arygupt
Collaborator
Loading…
3 of 6 tasks
Update glm-5 b200 sglang image to nightly-dev-cu13-20260523-c112f762
non-canary-full-sweep-enabled
Run the full sweep without the canary gate (full search space, no trim)
#1567
opened May 26, 2026 by
Ankur-singh
Collaborator
Loading…
[DNM][AMD] feat(agentic): AgentX v0.3 — Kimi MI355X LMCache MP benchmark
#1565
opened May 26, 2026 by
seungrokj
Collaborator
Loading…
Yeswanth/minimax fp4 gb300 b300 dynamo vllm disagg
full-sweep-enabled
#1560
opened May 23, 2026 by
yeswanthk-26
Collaborator
Loading…
[NV] Update B300 DSV4 SGLang Pareto sweep
full-sweep-enabled
#1552
opened May 22, 2026 by
YAMY1234
Loading…
[Klaud Cold] minimaxm2.5-fp8-mi300x: add SHUFFLE_KV_CACHE_LAYOUT=1 + ROCM_AITER_FA backend
full-sweep-enabled
#1550
opened May 21, 2026 by
functionstackx
Collaborator
Loading…
1 task
[Klaud Cold] minimaxm2.5-fp8-mi325x: add SHUFFLE_KV_CACHE_LAYOUT=1 + ROCM_AITER_FA backend
full-sweep-enabled
#1549
opened May 21, 2026 by
functionstackx
Collaborator
Loading…
1 task
[codex] fix profile relay and add B300 DSv4 Flash profile config
#1547
opened May 21, 2026 by
Oseltamivir
Collaborator
•
Draft
[NV] Update H100 Qwen3.5 SGLang agg config
full-sweep-enabled
NVIDIA
#1544
opened May 21, 2026 by
anish-shanbhag
Collaborator
Loading…
[NV] H100 (Agg): migrate model path
sweep-enabled
#1537
opened May 20, 2026 by
Ankur-singh
Collaborator
Loading…
Previous Next
ProTip!
Follow long discussions with comments:>50.