docs(DeepSeek-V4): mark gb300|{small,big}|{cp,pd-disagg} verified + GB300-specific fixes (#23691)

This commit is contained in:
fzyzcjy
2026-04-25 12:21:57 +08:00
committed by GitHub
parent d2c61acf25
commit 8a395994ed
2 changed files with 56 additions and 5 deletions
@@ -158,6 +158,13 @@ TCP, which can lead to garbled KV transfer on large checkpoints.
In order to use base models, please enable `SGLANG_FIX_DSV4_BASE_MODEL_LOAD=1` and use latest code, before the next round of testing matrix is finished.
**GB300 PD-Disagg cross-pod MNNVL**
On some GB300 clusters with cross-pod KV transfer over NVLink, mooncake may
fail with `nvlink_transport.cpp:497 Requested address ... not found!`. If
this happens, prepend `MC_FORCE_MNNVL=1 NCCL_MNNVL_ENABLE=1 NCCL_CUMEM_ENABLE=1`
to both prefill and decode `sglang serve` commands.
## 4. Model Invocation
### 4.1 Basic Usage