AI-Hypercomputer / AI-Hypercomputer/cloud-accelerator-diagnostics
How to understand the buffer size column?
Open
- Dominant language
- Python
- Stars
- 32
- Forks
- 15
- PR merge metrics
- No merged PRs in 30d
Description
Below is an example with `tpu-info@v0.5.0`:
`tpu-info --metric collective_e2e_latency`
There are two rows with `1KB+`, why?
Also, can we order the rows by buffer size?
Contributor guide
Assessment
This issue has not been assessed yet.