#
✕ nick:cgv[m]
#asahi · 2021-08-20 11:51
cgv[m]maz: i see. then i don't know whats going on, i dropped the barriers and can see a 1 cycle delay with data dependencies, the L1 miss should show up.
#asahi · 2021-08-20 11:37
cgv[m]LoC is 1, which means that only cleans/invalidates L1? if the latency is hidden by barrier that could explain that i do not observe the differences between L1 hit and L2 hit
#asahi · 2021-08-20 11:31
cgv[m]maz: ok, that makes sense. both LoUU and LoUIS are 0
#asahi · 2021-08-20 11:18
cgv[m]if I fill the cache to force evictions then I get timing differences, meaning that the measurement works
#asahi · 2021-08-20 11:16
cgv[m]CLIDR correctly shows separated L1, and unified L2; although CSSIDRs give a dummy value (0x70ffe07b)
#asahi · 2021-08-20 11:14
cgv[m]do CMOs work as expected? i'm not able to observe any timing effect of DC CVAU & co
#asahi · 2021-08-17 13:18
cgv[m]that would be super cool
#asahi · 2021-08-17 13:00
cgv[m]at some point i was tempted to port nanobench to arm, maybe m1n1 could be a good env for that?
#asahi · 2021-08-17 12:58
cgv[m]just exploring, i want to get a nice low-noise environment to quickly run snippets in different contexts, after that i guess some idea will popup
#asahi · 2021-08-17 11:56
cgv[m]event and PMCs where I observe changes (i'm not yet exercising all of them)
#asahi · 2021-08-17 11:55
cgv[m]< https://matrix.org/_matrix/media/r0/download/matrix.org/WflOPAkEoCVjlUTAajMVyegJ/message.txt >
#asahi · 2021-08-17 11:29
cgv[m]thanks, I'll start from that and see if I manage to count more events
#asahi · 2021-08-17 11:20
cgv[m]no, it's not, but i was trying to count the same event with another counter; ok, that make sense, i will check what-works-where, thx
#asahi · 2021-08-17 11:17
cgv[m]probably the later, since i don't seem to be able to observe many events :(
#asahi · 2021-08-17 11:16
cgv[m]dougall: i saw your repo :) did you experience anything like the above? just to be sure whether it's expected or if i'm doing something stupid
#asahi · 2021-08-17 11:02
cgv[m]has anyone played with perf counters in M1? I'm trying to use m1n1 to run microbenchmarks (setting up the counters and env from python), but I'm having some issues. e.g. evt=0x1 (retired uops) only seem to work in PMC1, I'm not able to set it up in PMC2, but evt=0x2 (cycle count) works well with any PMC