core-to-core-latency
core-to-core-latency copied to clipboard
Another AMD Threadripper 3960X results
Hey, leaving another sample for the system with the same CPU (SMT enabled) as from the #15 (run with 5000 sample iteration size like specified in the README), the results deviate more and are less uniform. I also have cores behaving like the ones from more-performant group in the worse performing one. output.csv.
Also, one time (attached screenshot, I sadly overwritten the results) the "better" performing cores were in 1-12 group instead of 13-24 so the CPU or OS is changing these.