00:46karolherbst[d]: okay.. I fixed the cross tests 🙃
00:47karolherbst[d]: had to fetch `nir_intrinsic_cmat_signed_mask` and overwrite the sign of the operands
00:48karolherbst[d]: I have some weirdo convert fails.. mhhh
00:55karolherbst[d]: mhhh.. it's just the b matrix being broken...
00:56karolherbst[d]: probably some layout mess up
01:00karolherbst[d]: ehhh... why isn't this working...
05:48troubledpony: what is clear is this dwfreed is not mentally ill by genetics, cause such thing does not exist, he is a birth handycap just like mohamexiety[d] dave airlie and such, cause you had no results anywhere in the history, you had no results in sports nor programming, such guys i call abortion leftovers (what counts is their physical illness that contributes to such chaos type of incapability),
05:48troubledpony: they are soon under heavy shelling due to their abuse done to others. If you do not understand anything i talk about like sequences where 58+127+71=198 and 140 is 127+71 what bitcount, why do you waste time on the bitcount, if you do not understand it's an honor of mine, cause you are not brain equipped people, as i have been telling.
05:58troubledpony: And where is that testosterone or Hgh or IGF-h coming from for you to bully like this, so where does sigma aldrich extract those human carrier steroids if not from human donors?
05:58troubledpony: humanoid or what, from aliens so to speak?
06:10troubledpony: So i give a little pointer, several biochips have been identified from my circulation after i got jammed and terrored under blacklisted doctors, they take human steroids from certain carriers like me, to get very pure substance, and they sell it to others with very high prices, and this is also seen from my shape of the head easily.
06:11troubledpony: there were something wrong in my genes and this is not ancient thing, cause several nations test their mass destruction weapons and i was a twin dominant on top.
06:11troubledpony: so it's known that all they are after are those substances.
06:12troubledpony: that is why i call myself a god, cause earlier times such giants were mistakenly perhaps called as ones.
06:19HdkR: 🎉
10:54snowycoder[d]: Does nvk have somethjing similar to `nouveau's mip_tree->layout_3d`?
10:54snowycoder[d]: I don't see how the layer count might change with mip level
12:43falconm[m]: is the freedesktop blog not being hosted or was it just not updated?
12:48falconm[m]: I know this is a silly question but can I ask the question of how can I contribute?
13:04snowycoder[d]: falconm[m]: If you're new, welcome!
13:04snowycoder[d]: I can help, I'm quite new myself.
13:04snowycoder[d]: (if you're not I'm sorry I'm really bad at remembering names)
13:15falconm[m]: <snowycoder[d]> "falconm: If you're new, welcome!" <- thank you! I am new, yes
13:17falconm[m]: <snowycoder[d]> "I can help, I'm quite new myself..." <- the only real thing I can help with is frontend
13:17falconm[m]: dunno if it is needed though that is why I am asking
13:18snowycoder[d]: falconm[m]: Help is always appreciated, do you have some hardware to work on?
13:24falconm[m]: <snowycoder[d]> "falconm: Help is always apprecia..." <- may you specify what do you mean by hardware to work on?
13:27karolherbst[d]: uhh.. cmat_convert is going to be a pain to implement I think...
13:28karolherbst[d]: the layout of the source and dest matrix cna be different, so I think we might have to move values between threads to fix it all up...
13:29karolherbst[d]: I wonder if some smart shuffle can help here... kinda don't want to do it via temporary storage
13:30snowycoder[d]: falconm[m]: Do you have a specific graphics card?
13:31snowycoder[d]: I wouldn't mind a help with kelpler but that is old hw
13:32snowycoder[d]: There are also some issues about nvk bugs on gitlab (e.g. games not working)
13:40falconm[m]: <snowycoder[d]> "falconm: Do you have a specific..." <- I have an rtx 4080 laptop gpu
13:41falconm[m]: can I run tests or something on it?
13:45snowycoder[d]: I think so, you could run dEQP to see if something's broken but it shouldn't be
13:47snowycoder[d]: There's also DOOM 2016 that's having issue starting, a good starting point would be to bisect what commit broke it: https://gitlab.freedesktop.org/mesa/mesa/-/issues/12549
13:47snowycoder[d]: Otherwise ask the experts, I kinda stole a lot of easy tasks, sorry
13:47falconm[m]: I use nvidia drivers dunno if running dEQP will help
13:48falconm[m]: but we can compare results of me running it on nvidia drivers compared to nouveau
13:48falconm[m]: to see the difference
13:48falconm[m]: not sure if it will help
13:48falconm[m]: all of this is kind of new to me
13:49snowycoder[d]: Reverse Engineering is possible, but it's not something I would begin with.
13:49snowycoder[d]: (belive me, I tried and failed😂 )
13:50falconm[m]: makes sense
14:21plonkateveryone: mohamexiety[d]: they are not developers, all the code and ideas i gave to them, they are the biggest shithoses of the history of human species LGBT fecalists, as soon as i saw them being endorsed in Europe along with other retards, i knew where this is going. Gladly enough just like Laura Tornado and Gloria Terreur and ones gangs you are walking dead jack dedman's name already refers to
14:22plonkateveryone: his state with bunch of other deluded english shithoses. My ideas have always worked because i have brain. am saying that, It's a matter of diplomacy and other agreements to be put on the pipeline and you are under heaviest shelling and slaughter houses all in all for good. so next index works the same way 121-130=-9-16-32+63=6+63=69 121-16-32=73 surprise surprise 256-121-72=63 cause
14:22plonkateveryone: the interpolant was coming from index 116 and 120 virtual index having 136 divided by two 68 divided by two 34, so hence 116-24+34=126 twice of 63. and 73-69=4 not 71-69 aka 69 coming from 127-32-26=69 so 68+4=72 where as index 115 was odd and hence 71-69=-2 72-2+1=71, trust me your case will be understood and you might even live ahead of time, if you fuck off from around real men,
14:22plonkateveryone: never stalk them again, you might even keep your genitals and anus this way, if you trash my type of people again and again, it's one hundred percent sure you end up tortured to death.
15:13mohamexiety[d]: mohamexiety[d]: yes and no. so it seems with Hopper and Blackwell we get 57 bit addresses. 32+24 is 56 which is also what the total program address is in QMD v4. 32+20 is tricky; however thanks to Faith I realized that the Hopper cbuf addresses are shifted by 6. and indeed checking out QMD v4, the total cbuf address is 32+19..
15:13mohamexiety[d]: so now the mystery is solved. all that remains is figuring out which of the 32 bit packets is the lower program address and which is the lower cbuf address :KEKW:
16:03snowycoder[d]: Was following a segfault for 10 minutes for a misuse of `nir_builder`, NIR rewrite in rust when? (\s)
16:24ramblereric: if you'd look into mirror and say, i am a retarded monster, that would help you to understand or plead your guilt and case would be better for you. We do not need to deal with retards punishing , we'll go straight to the heads of European monster rulers and force them to change their actions. And this 16 and 26 was put deliberately it does not care what that is, until it is in the correct
16:24ramblereric: range to extract the delta, but to see that the index is indeed odd, you are preferred to do 69+68=137 and 69+72=141 so it would look consistent to your retarded eyes, however i said before and you noticed not, you are not supposed to be starting below 72, if 16 subtracted before was 18, but it never matters, until you do not overshoot it and have same interpolant on both of the
16:24ramblereric: artihmetics. You are the biggest trash known in the world, you will have sophistated life with your head fecalists, like dog ryan and airlied, due to many sided abuse you committed on my vacation, as well as on irc. Do not even doubt it. If you plead guilty and understand you are shithoses and liquid shit producers in worlds climate , things go better.
16:31HdkR: 🎉
16:46f_: :(
17:19karolherbst[d]: Passed: 2764/31414 (8.8%)
17:19karolherbst[d]: Failed: 36/31414 (0.1%)
17:19karolherbst[d]: Not supported: 28614/31414 (91.1%)
17:19falconm[m]: are those tests of some sort
17:20karolherbst[d]: cooperate matrix
17:22snowycoder[d]: falconm[m]: They are (a subset of) Vulkan conformance suite: https://github.com/KhronosGroup/VK-GL-CTS
17:23falconm[m]: snowycoder[d]: thank you for clarification!
17:25karolherbst[d]: I wonder if I want to wire up the 8x8x4 ones...
17:25karolherbst[d]: should be "simple" (tm)
17:26karolherbst[d]: but I don't know if they are even supported on other gens besides turing
18:00gfxstrand[d]: snowycoder[d]: I so wish...
18:01gfxstrand[d]: There are so many corner cases with nir_builder. That thing is very unsafe.
21:32snowycoder[d]: Wait, can I even add custom modifier to a new NIR intrinsic?
21:33snowycoder[d]: Should we really do the sucalc dance in NIR? it seems more of a backend task
21:37leftmostcat[d]: Since https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34372 is a bunch of only loosely-related commits, would it be helpful to break it up?
21:41technomination: now the last part is the most interesting as to where those come from: we do not want to divide by two hence 58+58+18=134 134-144=-10 if you want to divide by two 58+72=130 130-140=-10 you clearly see this is on the minus side where 24 added to -10 goes to 14+58=72, so you are safe if you always put 24 to the 126-140-24+58+58=78 and now 126-24-24=78 more exotic ones as
21:41technomination: 126-140-29-22+58+58=51 126-29-22-24=51 you see the 24 in the second as obligatory, now second index range limits are virtually already unconstrained 63+63+17 =143-136=7 on the positive that is why, and btw. 63+68=131-129=2 again positive. And to the walking dead retards, this technology was brought to you by Mart Martin like all your previous bigger ones that ever made sense faggots.
21:42technomination: And Mart is the toughest smartest and yet most abused guy in the world.
22:08mangodev[d]: mangodev[d]: bumping this issue
22:11mhenning[d]: mangodev[d]: please file a bug report. It's hard to keep track of issues as irc/discord threads
23:09mangodev[d]: mhenning[d]: i'll do so later today when i have time
23:48reallymadpuppets: This is not any type of restriction if you have some procedure done on the hashes indexes, you could then go below or above anyhow if you want. You are mad puppets, it's not like the first time i see retarded puppet scrubs. We have your funeral started upfront. There are big list of executions to be done for Estonian fraud wank spammers, they are as dead as you are. But all restlessly
23:48reallymadpuppets: tried to show they were not born as dead which obviously was not the case. So Ryan we will meet if you want, you go crawl infront of me to apologise maybe you clear your sick mind that way, either way the code you write is entire garbage and it applies to all of you. Bwidawsk was profilic contributor for intel, and smart men alike are silent when they know that another most abused man
23:48reallymadpuppets: was always correct. So if anything to pick would be a man like this. Definitely his fpga works if not before, but after my contributions will carry even bigger success. Ben Skeggs is profilic and Jerome Glisse. And all they are quiet surely cause they understand a bit too, instead of morons like most of you here.
23:48reallymadpuppets: Alex Deucher, one of my favourites.