liu21yd comments

Repositories
Issues
Comments

Results 3 comments of


                                            liu21yd

TransformerEngine v1.2.1 throws CuDNN frontend error on H100 GPU (AWS p5.48xlarge instance)

@sirutBuasai Was your problem solved? Could you tell me the solution. I meet the same problem.

Encountered an error: peer access is not supported between these two devices in tensorrt_llm::runtime::IpcMemory::allocateIpcMemory()

> Could you try adding `--use_custom_all_reduce disable` when you build the engine? It works , thank you very much.

Result is out of order when using http stream mode

> +1 I am facing the same issue. Are you using >1 for `postprocessing_instance_count` in your config.pbtxt? Yes. When I use the v0.11.0.dev2024051400 version's config, it works. It seems that...