track performance

[jordens]

Track the long term performance of ARTIQ components and alogrithms. Some are more critical than others but potential metrics could be:

• TTL, DDS, SPI, SAWG event throughput, input, outputs.
• FP performance.
• memory performance, latency, throughput.
• connection latency, RPC latency, RPC throughput.
• worker startup, kernel compilation, linking, kernel upload. Windows vs Linux.
• use PCUs

https://irclog.whitequark.org/m-labs/2017-03-17#1489752484-1489752048;

[jbqubit]

Thank you for advocating for this @jordens. Much needed. Add TCP throughput to the list #685.

[whitequark]

That's RPC throughput in the list.

[jordens]

@whitequark Now that we have PCUs, would it be quick to have another stab at trying #667?

[whitequark]

It certainly won't be quick as I don't have an easy way to reproduce the issue. But it will be possible.

[lriesebos]

I randomly stumbled on this issue. We currently do not track performance with steady time intervals, though we might start doing that later this summer. We do sometimes take performance samples with benchmarks we wrote ourselves. See https://gitlab.com/duke-artiq/dax/-/wikis/ARTIQ/Core-device-hardware