[NIXL][OOT platform] support nixl_connector with oot platform and other nixl_backend (#25121)

Signed-off-by: Chendi Xue <Chendi.Xue@intel.com>
This commit is contained in:
Chendi.Xue
2025-09-22 23:17:42 -05:00
committed by GitHub
parent e8db44f883
commit 5774b0a1da
5 changed files with 99 additions and 9 deletions

View File

@@ -31,6 +31,12 @@ Now supports 5 types of connectors:
--kv-transfer-config '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both"},{"kv_connector":"SharedStorageConnector","kv_role":"kv_both","kv_connector_extra_config":{"shared_storage_path":"local_storage"}}]}}'
```
For NixlConnector, you may also specify one or multiple NIXL_Backend. Such as:
```bash
--kv-transfer-config '{"kv_connector":"NixlConnector","kv_role":"kv_both", "kv_buffer_device":"cuda", "kv_connector_extra_config":{"backend":["UCX", "GDS"]}'
```
- **OffloadingConnector**: enable offloading of KV data to CPU memory, customizing the CPU block size (in tokens) and number of blocks to allocate (per worker):
```bash