Skip to content

FP6 Speed on A100 80g #1181

@shihaobai

Description

@shihaobai

ENV:
cuda: 12.1
torch: 2.5.0+cu121
python benchmark_fp6.py
image
Hello, have you tested the performance of the FP6 kernel on the A100? I found that the speed is much slower compared to FP16."

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