| 28 Jul 2025 |
 jade_ | i have it on my work computer! | 01:32:29 |
 raitobezarius (DECT: 7248) | :D | 01:32:32 |
| raitobezarius (DECT: 7248) | In reply to @emilazy:matrix.org
some people will do anything to avoid clicking three buttons in a nice GUI :p i use macOS exclusively over SSH | 01:32:45 |
| raitobezarius (DECT: 7248) | I have my excuses | 01:32:47 |
| jade_ | i don't want to click three buttons in a nice GUI and i want to be able to dtrace it if i start having a bad time | 01:32:53 |
 emily | this is a good way to hate macOS | 01:33:01 |
| jade_ | well we're all very good at that :) | 01:33:11 |
| jade_ | I just happen to now have enough macs to be forced to care :P | 01:33:23 |
| emily | the Instruments UI is basically a very refined UI around the DTrace scripts you'd end up writing for perf analysis already | 01:33:32 |
| jade_ | mm | 01:33:42 |
| emily | it can do system-wide profiling and stuff too | 01:33:57 |
| jade_ | anyway i have run into an old stupid bug where you can't nix store delete a self referencing store path | 01:34:08 |
| jade_ | i hit that one basically every time i am trying to do benchmarking of nix builds or similarly wanting to clean build a certain path | 01:34:57 |
| jade_ | https://eclecticlight.co/2022/01/13/scheduling-of-processes-on-m1-series-chips-first-draft/ related info to prev discussion | 01:37:34 |
| emily | I think I probably already linked to https://blog.xoria.org/macos-tips-threading/ but it has a useful brief overview re: scheduling/QoS | 01:40:17 |
| emily | though not as detailed of course | 01:40:29 |
 hexa | loving the priority QoS has | 01:40:40 |
| hexa | * loving the priority QoS has in this flow | 01:40:42 |
| hexa | wondering how the Linux schedulers deal with P/E core scheduling | 01:41:09 |
| emily | pretty badly I think | 01:41:26 |
| emily | Linux definitely does not have comparable to the Darwin systemwide QoS handling | 01:41:51 |
| jade_ | android likely invented one, but it sounds very plausible that it never made its way into desktops | 01:42:38 |
| emily | there is https://docs.kernel.org/scheduler/sched-energy.html | 01:42:40 |
| emily | but Darwin fundamentally has more information (since everything in the OS is setting QoS information that passes through IPC and they have a global QoS-aware work scheduling mechanism and etc.) | 01:43:56 |
| emily | the fun part of efficiency/perf cores is when the two support different ISAs | 01:44:50 |
| emily | iirc Intel have chips where you can use AVX512 on the P cores but not the E ones or something | 01:45:11 |
| emily | and if the scheduler messes it up you just die | 01:45:17 |
| jade_ | intel did this on my machine where only p cores have the perf counters required to run rr | 01:45:22 |
| hexa |
EAS in its current form is SMT unaware and is not able to leverage multithreaded hardware to save energy.
| 01:45:23 |
| jade_ | i have a stupid script to workaround it | 01:45:27 |
