mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-14 06:27:45 +08:00
cbca6505ff
This issue is found for tp=ep=8 on the multi-node machine due to the inconsistent PP sizes. * Reform the workspace allocation implementation to avoid the list-out-of-range issues. * Disable min_latency_mode under the multi-node scenario to avoid the illegal memory access issue. Signed-off-by: Yukun He <23156053+hyukn@users.noreply.github.com> |
||
|---|---|---|
| .. | ||
| attention_backend | ||
| auto_deploy | ||
| compilation | ||
| custom_ops | ||
| distributed | ||
| models | ||
| modules | ||
| peft | ||
| pyexecutor | ||
| speculative | ||
| __init__.py | ||
| autotuner.py | ||
| llm.py | ||
| metadata.py | ||
| model_config.py | ||
| pipeline_interface.py | ||
| utils.py | ||