Skip to content

Does unitrace work for tensor parallel (TP) on multiple GPUs? #115

Description

@Xia-Weiwen

When I ran a model with SGLang on 4 B60 GPUs with tensor parallel (TP), the unitrace raised segfault errors. The command I ran is unitrace -d -h <my command to run model with SGLang>. So, does unitrace support such scenarios? If yes, how should I use it?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions