On 05/12/2023 17:50, Han Dong wrote:
w.r.t. Benchmarks: I'm also curious in general how common optimizations people have done to improve network performance affects power? For example if we can support the same workload but with fewer instructions, then that automatically means lower power consumption right? Things that pop to my mind: bypassing some of the kernel, replacing TCP with UDP, impact of having a dedicated polling to reap packets for multiple workers (might not be a way around this for very low latency apps)
These sort of micro optimizations would only work after you have solved thermal pressure. In my own testing it seems like everything blows up (power wise) once the server fans kick in to keep the processor under the thermal threshold.