-
Notifications
You must be signed in to change notification settings - Fork 292
Pull requests: SemiAnalysisAI/InferenceX
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[AgentX GB300] enable Kimi-K3 disaggregated power / 启用分离部署功耗
#3047
opened Sep 12, 2026 by
edwingao28
Collaborator
•
Draft
7 of 9 tasks
[AgentX GB300] enable Kimi-K3 aggregate power / 启用聚合部署功耗
#3046
opened Sep 12, 2026 by
edwingao28
Collaborator
•
Draft
6 of 9 tasks
[AgentX GB200] enable Kimi-K3 aggregate power / 启用聚合部署功耗
#3045
opened Sep 12, 2026 by
edwingao28
Collaborator
•
Draft
6 of 9 tasks
[AgentX H200] enable Kimi-K3 measured power / 启用实测功耗
#3044
opened Sep 12, 2026 by
edwingao28
Collaborator
•
Draft
6 of 9 tasks
[AgentX B200] enable Kimi-K3 power and correct DCP metadata / 启用功耗并修正元数据
#3043
opened Sep 12, 2026 by
edwingao28
Collaborator
•
Draft
6 of 9 tasks
[AgentX Power] collect Slurm power results and retain failure evidence / 保留失败证据
#3042
opened Sep 12, 2026 by
edwingao28
Collaborator
•
Draft
4 of 9 tasks
[PowerX AMD] preserve telemetry through the measurement end / 保留完整窗口遥测
#3041
opened Sep 12, 2026 by
edwingao28
Collaborator
•
Draft
6 of 9 tasks
[AgentX B300] recover the Kimi-K3 image and runner / 恢复镜像与集群路由
#3040
opened Sep 12, 2026 by
edwingao28
Collaborator
•
Draft
5 of 9 tasks
Add B200 DeepSeek V4.1 Flash P/D without KV offload / 新增无 KV 卸载的 B200 P/D
full-sweep-fail-fast
#3039
opened Sep 12, 2026 by
Oseltamivir
Collaborator
Loading…
Add B300 DeepSeek V4.1 Flash P/D without KV offload / 新增无 KV 卸载的 B300 P/D
full-sweep-fail-fast
#3038
opened Sep 12, 2026 by
Oseltamivir
Collaborator
Loading…
Add GB200 DeepSeek V4.1 Flash P/D without KV offload / 新增无 KV 卸载的 GB200 P/D
full-sweep-fail-fast
#3037
opened Sep 12, 2026 by
Oseltamivir
Collaborator
Loading…
Add GB300 DeepSeek V4.1 Flash P/D without KV offload / 新增无 KV 卸载的 GB300 P/D
full-sweep-fail-fast
#3036
opened Sep 12, 2026 by
Oseltamivir
Collaborator
Loading…
config: switch MI355X DSv4.1 Flash to the ROCm nightly image / MI355X DSv4.1 Flash 切换到 ROCm nightly 镜像
#3035
opened Sep 12, 2026 by
Fangzhou-Ai
Collaborator
•
Draft
[Klaud Cold] Update dsr1-fp4-b200-dynamo-sglang-mtp SGLang image to v0.5.19 / 将 dsr1-fp4-b200-dynamo-sglang-mtp 的 SGLang 镜像更新至 v0.5.19
#3034
opened Sep 12, 2026 by
Klaud-Cold
Collaborator
•
Draft
[DNM] Validate MI355X AITER + vLLM backend fixes / 联合验证 MI355X 后端修复
full-sweep-fail-fast
#3031
opened Sep 12, 2026 by
functionstackx
Collaborator
•
Draft
[PowerX] enable Kimi-K3 AgentX power backfills / 启用功耗补测
#3030
opened Sep 11, 2026 by
edwingao28
Collaborator
•
Draft
7 of 9 tasks
[Klaud Cold] Try TEP4 and DEP4 for DeepSeek V4.1 Flash AgentX on B300 / 在 B300 上尝试 DeepSeek V4.1 Flash AgentX 的 TEP4 与 DEP4
full-sweep-fail-fast
#3028
opened Sep 11, 2026 by
functionstackx
Collaborator
Loading…
[PowerX] require measured power for normal 8k1k runs / 为常规 8k1k 运行启用必需实测功耗
#3027
opened Sep 11, 2026 by
edwingao28
Collaborator
Loading…
7 of 9 tasks
[PowerX] preserve outcomes and collect AMD worker power / 保留基准状态并采集 AMD 节点功耗
#3026
opened Sep 11, 2026 by
edwingao28
Collaborator
Loading…
6 of 9 tasks
feat(config): add GLM-5.2 FP8 H200 AgentX 1P1D HiCache / 新增 GLM-5.2 FP8 H200 AgentX 1P1D HiCache 配置
full-sweep-enabled
#3021
opened Sep 11, 2026 by
RohitNagraj
Collaborator
Loading…
4 of 9 tasks
[WIP][AMD][AGENTX] GLM-5.2 FP4 MI355X SGLang v0.5.19: full Pareto re-sweep
full-sweep-fail-fast
#3014
opened Sep 11, 2026 by
giovanniguastiamd
Collaborator
Loading…
4 tasks
[Klaud Cold] Update dsv4-fp4-b200-sglang-agentic-hicache-mtp SGLang image to v0.5.19-cu130 / 将 dsv4-fp4-b200-sglang-agentic-hicache-mtp 的 SGLang 镜像更新至 v0.5.19-cu130
full-sweep-enabled
#3012
opened Sep 11, 2026 by
Klaud-Cold
Collaborator
Loading…
[Klaud Cold] Update qwen3.5-fp8-mi300x-sglang SGLang ROCm image to v0.5.19-rocm720-mi30x / 将 qwen3.5-fp8-mi300x-sglang 的 SGLang ROCm 镜像更新至 v0.5.19-rocm720-mi30x
full-sweep-enabled
#3011
opened Sep 11, 2026 by
Klaud-Cold
Collaborator
Loading…
[Klaud Cold] Update dsr1-fp8-h200-sglang SGLang image to v0.5.19-cu130 / 将 dsr1-fp8-h200-sglang 的 SGLang 镜像更新至 v0.5.19-cu130
full-sweep-enabled
#3010
opened Sep 11, 2026 by
Klaud-Cold
Collaborator
Loading…
[AMD] Enable triton DSA backends and top-k v2 path for glm5.2-fp4 in MI355X
full-sweep-fail-fast
#3009
opened Sep 11, 2026 by
EricKing626
Collaborator
Loading…
1 of 4 tasks
Previous Next
ProTip!
no:milestone will show everything without a milestone.