[2025-05-19T00:50:18Z] huh, would you look at that. cachyOS, clear etc, boasts even less of a performance boost in phoronix's new test on a framework laptop. https://www.phoronix.com/review/framework-13-amd-linux-2025 [2025-05-19T00:50:53Z] tldr, pgo, bolt, lto seems to just be too much work, for not a whole lot of gains. ~5% increase is ridiculus [2025-05-19T00:51:09Z] actual performance gains come from superior software not aggressive compiler opts [2025-05-19T00:51:59Z] ive seen people spend actual hours tuning their overclocks for an extra 5% or less though so i guess there is a market for it [2025-05-19T00:59:43Z] right. some people just want to push to the absolute outer limit, no matter the cost. more or less [2025-05-19T01:00:26Z] and with all of this, I kinda just realised that, in order to have decent performance, its simple, just dont create a turd setup [2025-05-19T01:02:00Z] but yeah, as you also said, use better software, or just get better hardware really :p [2025-05-19T01:02:16Z] fun fact, -O3 hurts probably just as much as it helps [2025-05-19T01:02:24Z] ubuntu decided they're going to toggle it on a per package basis [2025-05-19T01:02:28Z] because for some it bombs perf [2025-05-19T01:03:09Z] correct. -O2 is a better compromise, or even -Os actually [2025-05-19T01:03:41Z] i usually just do -O2 for everything [2025-05-19T01:04:10Z] this also got me thinking, i wonder what situations zstd actually helps compared to something else like xz or gzip [2025-05-19T01:04:37Z] yeah ive been using -Os instead for a long time now. and I like to think it works decently enough [2025-05-19T01:04:42Z] with my initramfs (tinyramfs) at least, zstd was like twice as slow as gzip because i had to crank it to get it to spit out a file of the same size or smaller [2025-05-19T01:04:49Z] using gzip -9 [2025-05-19T01:05:09Z] xz is a turd. gzip is fast, but doesnt really compress much. zstd is a good compromise imo. fast compression and decompression, aswell as rate [2025-05-19T01:05:32Z] lz4 has supperior decompression speed [2025-05-19T01:05:35Z] also curious about it in the context of memory compression [2025-05-19T01:05:53Z] i have quite a banger CPU but haven't bothered testing zstd with zram, just been using lz4 [2025-05-19T01:06:33Z] I also use lz4 on kernel, but mostly just because its decompression speed is the best [2025-05-19T01:06:56Z] I use lz4 for zram, because of the speed. i think gentoo recommended it. iirc [2025-05-19T01:08:15Z] probably just really depends on what kind of cpu you're using [2025-05-19T01:08:46Z] lz4 seems quite effective though lol [2025-05-19T01:09:08Z] wait, not Im using zswap, not zram [2025-05-19T01:09:29Z] zswap is better, atleast thats what I came to understand [2025-05-19T01:10:08Z] lz4 is decent imo. atleast if speed is important. [2025-05-19T01:10:41Z] i mean the whole point of ram is speed so [2025-05-19T01:10:51Z] i don't on-disk swap at all anymore [2025-05-19T01:11:17Z] I have enough ram that I very rearly swap. even less so on my desktop [2025-05-19T01:11:30Z] and with zswap, I would use it even less really [2025-05-19T01:11:40Z] I could probably just not have a swap partition [2025-05-19T01:15:10Z] yeah same situation here [2025-05-19T01:16:40Z] despite that, I still opted for a 16G or something swap on my desktop for some reason though. just to be sure :p [2025-05-19T01:16:48Z] for good measure, I suppose [2025-05-19T04:35:46Z] rearly [2025-05-19T04:35:48Z] he he [2025-05-19T16:32:42Z] I'm back in the kiss linux world guys. Currently reinstalling everything on my macbook air. See you in 5 hours x) [2025-05-19T16:35:45Z] Any of you know if TDE would compile fine on kiss ? [2025-05-19T20:38:00Z] FDE on Kiss Linux? [2025-05-19T20:48:41Z] kris_: seems like the kernel is too a subject too the gcc 15.1 issues.. oof