Vector pipeline performance monitoring

I’m working on the performance improvement for ML workloads on X280.

When writing a vector program, how can we make sure that the vector ALUs have been fully utilized.

Is there any performance monitoring utilities?

(For example, Andes provides the pipeline visualizer tool: AndesClarity)

You’re posting in a topic for the HiFive1 Rev B, a microcontroller board that doesn’t have vector instructions (or FPU, or MMU, or …)

In general, this whole forum is for supporting the general public who have purchased a SiFive eval board at retail.

If you have access to hardware that has not shipped to the public, such as the X280, then you should be getting support from your SiFive sales contact.

hentrikchen - Can we have a quick phone call. Please drop me a note at & I can share more details.