docs(device-supported): add a component column per device with links - #839
docs(device-supported): add a component column per device with links#839mesutoezdil wants to merge 8 commits into
Conversation
Signed-off-by: Aditya Kumar Nayak <adi.nayak45@gmail.com>
Signed-off-by: Aditya Kumar Nayak <adi.nayak45@gmail.com>
Signed-off-by: Aditya Kumar Nayak <adi.nayak45@gmail.com>
Signed-off-by: Aditya Kumar Nayak <adi.nayak45@gmail.com>
…components Signed-off-by: mesutoezdil <mesudozdil@gmail.com> # Conflicts: # docs/userguide/device-supported.md # i18n/zh/docusaurus-plugin-content-docs/current/userguide/device-supported.md
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: mesutoezdil The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
✅ Deploy Preview for project-hami ready!
To edit notification comments on pull requests, go to your Netlify project configuration. |
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Team Run ID: 📒 Files selected for processing (3)
Included review availability: Your plan provides up to 1 included review per hour; 0 remain after this review. 📝 WalkthroughWalkthroughThe device support documentation now includes support statuses, expanded capability definitions, AWS Neuron and Biren entries, component-specific guide links, and corresponding Chinese localization updates. ChangesDevice support documentation
Estimated code review effort: 1 (Trivial) | ~5 minutes Merge Risk: ⚪ Minimal · up to This change expands device-support documentation, component guide links, and Chinese localization. The supplied findings indicate the English and Chinese matrices and sidebar translation are aligned, with no current merge-blocking risk. Suggested reviewers: 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
The latest HAMi release is v2.10.0, not v2.9.0. Biren ships in it, so its status goes back to Stable. Signed-off-by: mesutoezdil <mesudozdil@gmail.com>
Kunlunxin, Vastai, Biren and AWS Neuron carried isolation values the code does not back. Checked the v2.10.0 chart device config and each device guide: - Kunlunxin exposes vxpu and vxpu-memory but no core resource, so CoreIsolation is No - Vastai exposes only va, so both are No - Biren exposes only birentech.com/gpu, so both are No - AWS Neuron exposes neuroncore, so CoreIsolation is Yes These values predate this branch. The page defines the columns as a hard per container limit, and without a resource name there is no way to request one. Signed-off-by: mesutoezdil <mesudozdil@gmail.com>
|
following your v2.10.0 catch, i audited the rest of the table against the same baseline and found four rows whose isolation values the code does not back. checked two independent sources, the v2.10.0 chart device config and each device guide:
the page defines these columns as a hard per container limit, and without a resource name there is no way to request one, so the mapping holds. these values predate this branch, they are on master today. i also re-checked MetaX and Enflame, which looked wrong at first because their keys are named differently, resourceVMemoryName and resourceNameGCUMemory. both are correct as Yes / Yes. this is the mismatch #689 describes, users finding out in production, so worth a second opinion from someone closer to those vendors before it merges. |
| | GPU | Biren | Biren166M | Yes | Yes | No | | ||
| | GPU | AMD | Instinct / ROCm | Yes | Yes | No | | ||
| | DPU | Teco | Checking | In progress | In progress | No | | ||
| | Type | Manufacturer | Models | Status | MemoryIsolation | CoreIsolation | MultiCard Support | |
There was a problem hiding this comment.
MultiCard Support is ambiguous here. It means multi-card partitioning, not whether a Pod can request multiple whole cards. Devices marked No can still request multiple whole cards, but should not set memory/core resources when requesting more than one card. Could we rename it to MultiCard Partitioning and update the Chinese version to 多卡切分支持 as well?
|
|
||
| - **MemoryIsolation** - Whether HAMi enforces a hard VRAM limit per container: workloads that exceed their requested memory are rejected instead of drawing on the full physical device memory. | ||
| - **CoreIsolation** - Whether HAMi enforces a hard compute usage limit per container: kernel execution is throttled to stay within the requested share instead of using the physical device's compute freely. | ||
| - **MultiCard Support** - Whether a single Pod can request and be scheduled onto more than one physical card of that type, with HAMi coordinating placement across the selected cards. |
There was a problem hiding this comment.
- MultiCard Partitioning - Whether a single Pod can use partitioned memory or compute resources across multiple physical cards.
Nomeans that multiple whole cards can still be requested, but memory or core resources must not be specified when requesting more than one card.
|
|
||
| - **显存隔离** - 是否为每个容器强制执行显存硬限制:超出请求显存的工作负载会被拒绝,而不能使用整块物理设备显存。 | ||
| - **核心隔离** - 是否为每个容器强制执行算力硬限制:内核执行会被限流以保持在请求的份额内,而不能自由使用物理设备的算力。 | ||
| - **多卡支持** - 单个 Pod 是否可以请求并调度到该类型的多张物理卡,由 HAMi 协调所选卡之间的分配。 |
There was a problem hiding this comment.
- 多卡切分支持 - 单个 Pod 是否可以同时使用多张物理卡的显存或算力切分资源。对于标记为
No的已支持设备,多卡请求必须使用整卡,且不能指定显存或算力资源。
What type of PR is this?
/kind documentation
What this PR does / why we need it:
Item 1.6 of the reorganization plan wants the supporting components and their links on the devices matrix. A second table answers which component supports which device, with every cell linking to that device and component guide: NVIDIA has all four, Huawei Ascend has HAMi and Volcano, the rest have HAMi today. A dash means there is no guide today rather than an impossible combination, which is stated above the table. It also baselines the matrix on v2.10.0 and corrects four rows whose isolation values the code does not back, checked against the v2.10.0 chart device config and each device guide.
Which issue(s) this PR fixes:
Part of #689, and #672
Checklist:
npm run lintandnpm run format:checkpassnpm run buildsucceeds for bothenandzhgit commit -s)