8 Pull requests merged by 7 people

• docs: Update MAINTAINERS file with current contributor information

  #918 merged Mar 7, 2025

• feat: Add kube-qps and kube-burst parameters.

  #769 merged Mar 7, 2025

• typo: fix typos in log and comment

  #917 merged Mar 6, 2025

• Fix UT not be properly executed during CI phase

  #911 merged Mar 4, 2025

• docs: add syntax highlighting to various code blocks

  #906 merged Mar 3, 2025

• feat: support scheduler replicas > 1

  #898 merged Mar 3, 2025

• add config gpu core isolation policy for webhook

  #901 merged Mar 3, 2025

• docs: fix broken commmunity links

  #907 merged Mar 3, 2025

4 Pull requests opened by 4 people

• Optimize the handling of the return messages of the extender scheduler

  #908 opened Mar 3, 2025

• Fix the wrong describe of device registry in protocol.md

  #910 opened Mar 3, 2025

• fix: hygon dcu concurrent creation conflict

  #921 opened Mar 7, 2025

• chore: add namespace override for multi-namespace deployments

  #924 opened Mar 9, 2025

6 Issues closed by 3 people

• V100-SXM3-32GB和V100-PCIE-32GB支持MIG切分吗？

  #862 closed Mar 7, 2025

• scheduler: client-side throttling, not priority and fairness

  #745 closed Mar 7, 2025

• HAMi调度策略配置是否支持除nvidia外的其它型号的卡？

  #821 closed Mar 6, 2025

• 容器内执行nvidia-smi出现multiprocess_memory_limit.c:499)]: Calling exit handler 30

  #916 closed Mar 5, 2025

• 为什么调度器的实例个数被固定成1个？

  #877 closed Mar 3, 2025

• 两个pod，gpucoores分别为20,40。两个pod同时执行模型推理，为什么显卡的利用会超过60%，而不是60%以内？

  #875 closed Mar 3, 2025

8 Issues opened by 7 people

• Can dlsym Hijacking Fully Intercept libcuda.so

  #925 opened Mar 9, 2025

• HAMi RoapMap

  #923 opened Mar 7, 2025

• 如何让gpumem等资源名也注册到node

  #922 opened Mar 7, 2025

• [BUG] hygon dcu will bind wrong vdev when concurrent creation

  #920 opened Mar 7, 2025

• 分配了虚拟gpu但pod提示不足(Virtual GPU is allocated but pod prompts insufficient)

  #919 opened Mar 6, 2025

• deployment多副本同时启动异常且副本数不符合预期

  #915 opened Mar 5, 2025

• 在容器内使用nvidia-smi命令一直卡在Initializing

  #913 opened Mar 4, 2025

• Wrong go module version

  #912 opened Mar 4, 2025

11 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• 部署HAMi-core后，执行测试案例，一直提示unified_lock locked, waiting 1 second...

  #696 commented on Mar 5, 2025 • 0 new comments

• Apply for reviewer permission

  #699 commented on Mar 5, 2025 • 0 new comments

• Who's using HAMi / 您在使用 HAMi 吗 ?

  #4 commented on Mar 6, 2025 • 0 new comments

• this pr will lead pod schedule hang very long time. https://github.com/Project-HAMi/HAMi/pull/714

  #810 commented on Mar 7, 2025 • 0 new comments

• 使用此插件调度出现pud状态 RunContainerError，事件显示：nvidia-container-cli: device error，使用docker run是正常起来的

  #411 commented on Mar 8, 2025 • 0 new comments

• Does HAMi-Core have any gpu overhead in theory? If it has, how to eveluate it?

  #904 commented on Mar 9, 2025 • 0 new comments

• Bump github.com/NVIDIA/nvidia-container-toolkit from 1.15.0 to 1.17.4

  #831 commented on Mar 4, 2025 • 0 new comments

• feat: Integrate NVIDIA/k8s-device-plugin into Project-HAMi/HAMi

  #855 commented on Mar 4, 2025 • 0 new comments

• Bump github.com/NVIDIA/go-nvlib from 0.2.0 to 0.7.1

  #868 commented on Mar 4, 2025 • 0 new comments

• Bump sigs.k8s.io/controller-runtime from 0.16.3 to 0.20.2

  #870 commented on Mar 4, 2025 • 0 new comments

• enhancement: use placeholder for node name in device plugin ConfigMap template

  #873 commented on Mar 5, 2025 • 0 new comments