[Docs] Add UCM configuration parameters reference page - #1293
Merged
harrisonyhq merged 8 commits intoSep 1, 2026
Merged
Conversation
ljccsu
requested review from
FangRun2,
Tarrei,
Wwwzff,
mag1c-h and
ygwpz
as code owners
August 27, 2026 08:48
qyh111
requested changes
Aug 28, 2026
| --- | ||
|
|
||
| ## pmr_config (PMR Configuration) | ||
|
|
|
|
||
| | Parameter | Required | Type | Value Range | Description | | ||
| |---|---|---|---|---| | ||
| | `use_layerwise` | Optional | bool | Default: `false` | Enable layer-wise (per-layer) load/save mode. Recommended `true`; DeepSeek V4 series recommends `false`. | |
| | Parameter | Required | Type | Value Range | Description | | ||
| |---|---|---|---|---| | ||
| | `store_pipeline` | Optional | string | See valid values below | Pipeline name. Default: `Cache\|Posix`. | | ||
| | `storage_backends` | **Required** | string | User-configured, multiple mount points separated by `:` | Storage mount path. See [Create Filesystem Mount](https://support.huawei.com/enterprise/zh/doc/EDOC1100582752/47f029f5#ZH-CN_TOPIC_0000002501883936). | |
Contributor
There was a problem hiding this comment.
链接不要了,只体现这里需要填本地目录或者挂载点,如果有多挂载点的话需要用冒号隔开
| | `storage_backends` | **Required** | string | User-configured, multiple mount points separated by `:` | Storage mount path. See [Create Filesystem Mount](https://support.huawei.com/enterprise/zh/doc/EDOC1100582752/47f029f5#ZH-CN_TOPIC_0000002501883936). | | ||
| | `io_direct` | Optional | bool | Default: `true` | Enable Direct I/O (bypass OS page cache). `false`: uses PageCache; `true`: skips PageCache. | | ||
| | `posix_io_engine` | Optional | string | Default: `psync` | File I/O mode. `psync`: synchronous; `aio`: asynchronous, requires `io_direct=true`. | | ||
| | `posix_data_trans_concurrency` | Optional | int | Default: `128` | Read/write threads per card in `psync` mode. NFS over RDMA: 128/card; DPC: max 256/node. Not used in `aio` mode. | |
…config, use_layerwise default true, remove Huawei links, remove DPC
…emove recommendation table, constraints block, fix cache_sdma_direct default, add GC conditional notes
harrisonyhq
reviewed
Sep 1, 2026
| | `enable_record_traces` | 选填 | bool | `false` | 用来记录请求信息(时间戳,输入长度,输出长度等信息)。 | | ||
| | `enable_metrics` | 选填 | bool | 默认 `true` | 是否开启 metrics 收集。 | | ||
| | `use_lite` | 选填 | bool | `false` | 是否启用 UCM Lite 功能。不对 KV Cache 数据进行保存和加载,仅对元数据进行保存和查询。仅可用于评估 KV Cache 命中率情况,无加速效果。 | | ||
| | `metrics_config_path` | 选填 | string | 自行配置 | 指定监控指标配置文件路径,启用后可通过 toolkit 进行 UCM 在线监控。 | |
Contributor
There was a problem hiding this comment.
metrics的配置在代码仓中的路径可以注明一下
| | `cache_load_backend_only` | 选填 | bool | 默认 `false` | 即使在 cache 层命中还是会强制从 SSD 上加载,仅供测试使用。 | | ||
| | `cache_io_aggregation` | 选填 | bool | 默认 `false`,仅在 `PLATFORM=ascend` 且模型为 V4 时自动开启 | 启用 IO 聚合 h2d 传输,仅在 A2 设备生效。 | | ||
| | `share_buffer_enable` | 选填 | bool | MLA 默认启用,GQA 默认不启用 | 是否启用共享内存。MLA 如果不用 shm 或 GQA 用 shm 都会导致性能下降。 | | ||
| | `posix_capacity_gb` | 选填 | int | 默认 `0`,表示不启用 GC;不可超过挂载文件系统可用容量 | 设置磁盘存储的最大容量(GB),当已用容量 >= `posix_capacity_gb * posix_gc_trigger_threshold_ratio` 时触发 GC。 | |
Contributor
There was a problem hiding this comment.
加一条注明,多实例部署,共享同一文件系统的时候,要求只能有一个实例开启GC,其他实例不要开启
…multi-instance note
… and troubleshooting entries
harrisonyhq
pushed a commit
that referenced
this pull request
Sep 4, 2026
## Modifications Add three reference pages to the docs-next MkDocs site based on the existing project documentation: - English project README - Chinese project README - Feature and model support matrix The PR targets the feature/docs-next branch and follows the page structure and table styling used in PR #1293.The previous PR is #1184
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Add comprehensive config parameters documentation with YAML hierarchy indication.