02:18 lowkicks: overall i can tell that i was probably wrong in everything that i tried or did try, was pretty wrong approach, but that means nothing that you come to my territory to drag me into things i am not interested in, i am interested to only hurt nasty people, that i was wrong with all my theories means nothing , every year science fails and projects with billions of dollars worth with no gain no results etc.
07:07 melodyofshootout: You do not even believe what you say or do and all you do is harass tourists! 2.5 years in a row they did it for me over the same slut whose specialty was shit hoses like you! The file system is pretty easy,i call my invention fusion trees cause the time domain which it operates under is the fastest possible, and compression is extraordinary, next things i move up to accelerator which uses that flow. So i guess you did not notice the
07:07 melodyofshootout: magic value, it was the magic number of the sample 624+724+931 where 524 was used for padding though more like bookkeeping this can not be accessed, and 931 is the end of bank which can be accessed with single 512, so fusion trees under file system umbrella or domain are in fact extremely simple.
07:15 melodyofshootout: fusion trees not as in AI literature and implementation, but stanford research from hypotetical model from 1991. AI in terms of memory is more like failure it's a memory hog.
07:16 melodyofshootout: but the banks can be as large as you want i demonstrated with 4 elements
10:04 fdobridge_: <t​om3026> @airlied hm prime setup with an external monitor those new fence patches might be causing a new regression for me, bit hard to pinpoint whos to blame here but after a while suddenly everything stops rendering but sound keeps on going in the background, swapping tty takes ~20 sec and when i return its back in action. no dmesg no journalctl error
10:07 fdobridge_: <t​om3026> didnt happend on 6.7.2 and doesnt happend on blob which im currently testing so im blaming those new fence patches i applied ontop of 6.7.3 but yeah :p
15:05 Sid127: oh boy people are bouncy today
15:05 Sid127: so many joins/leaves with connection timeouts ^^'
15:05 Sid127:moves the trampoline elsewhere
17:53 fdobridge_: <g​fxstrand> That's believable. That is living dangerously!
17:56 fdobridge_: <S​id> heheheh
20:35 fdobridge_: <i​ds1024> Hm, I still seem to be seeing compositor freezes with amdgpu + nouveau on mainline. Though at least magic sysreq + r lets me tty switch.
20:51 fdobridge_: <r​inlovesyou> yes i have observed the freezes happening on 6.8-rc1 & rc2
20:51 fdobridge_: <r​inlovesyou> yes i have observed the freezes happening on 6.8-rc1 & rc2 as well (edited)
20:54 fdobridge_: <i​ds1024> Seems it's blocking on a futex wait inside `nouveau_dri.so`
21:46 fdobridge_: <i​ds1024> Or, some things seem to be better with the current mainline kernel. While the hang I get starting sway happens with gentoo's media-libs/mesa-9999 but not a manually compiled mesa from the same commit? Hm.
22:04 fdobridge_: <i​ds1024> Oh, I forgot I had a bash alias for `sway` that sets WLR_DRM_DEVICES to work around issues I had with proprietary drivers when I didn't need the dGPU. And I wasn't using the alias when testing with the manual mesa install.
22:04 fdobridge_: <i​ds1024>
22:04 fdobridge_: <i​ds1024> So I guess things seem to be working now, on my system...