00:43 mareko: tjaalton: do you mean XWayland doesn't use xf86-video-amdgpu:
00:43 mareko: ?
00:49 karenw: XWayland does not need that part of the graphics driver stack as it delegates modesettings etc. to the wayland compositor.
01:35 damo22: does anyone think its a terrible idea to make a drm server for grub framebuffer?
01:35 damo22: i am on a platform with no linux kernel
01:37 damo22: ie no modesetting allowed, just passthrough the mode
01:47 Ermine: what does grub boot then?
02:16 damo22: Ermine: gnumach
02:20 damo22: via multiboot table
03:47 furiouselly: I don't think the society hates me, rather they try to help me from universities, but we now have the solvers data or manual guide to choosing two or one elements from the sum , this PDF deals with complex numbers which have imaginary unit included, logarithms of negative numbers as result have it where Pi get's calculated the i remains, you just read what it means, but i fluked in couple
03:47 furiouselly: of indices if not all of them, but if you leave enough gap of the indices, they are going to be uniform or contiguous so that logarithms can be taken upfront prestatically so they would not cause compile time nor runtime overhead. I mean sure the Cornell University as many others are many times over my math knowledge and experience as of today, but i did everything i could to get there
03:47 furiouselly: and close the gap so it's good to have a specification there luckily i found it, the formulas are derived from Euler, he was very long time ago a swizz mathematician. All in all i would expect the lemmas , theorems of the paper to be an exact match as to where we want to land but some might need a teachers help as you do not want me to talk and risk with human idiots lives in estonia and
03:47 furiouselly: elsewhere, just fucking abuse trolls they are with their sexual bragging, in my opinion they are all trash and should get a common massgrave in a war like this.
04:49 wickedrain: what impressed me the most is technology behind www.upnano.at it's clear that this computation that Cornell does to get natural number indexes and neat performance is not the almighty chip, but i have not found anything more performant than nuclear chips alike upnano, they are way beyond photonics or quantum computers etc. We are at the peak of science it's pretty sad situation that those
04:49 wickedrain: scams wanted to lead to my death and worldwide wars, as death to me was assigned and i already saw the accounts they laundered money at selling my genes to extend the life of jews, so all you talk and accuse is something that i never was involved in, that i got injured or could not keep up with fuckers after 14 years of age my libido started to sharply degrade cause of implants they
04:49 wickedrain: arranged on me. This grew into multimillion business yes, where i got almost fatally injured. But i do not collext money on this, i still have phenomenal organs, so i do not mind, hei at least jacob rothcshild is alive, and i have perspective to heal couple of things, overall you could find the dutch reports earlier to Cornells paper, i am aware it's military secret and this is how they
04:49 wickedrain: calculate the tanks aim of the barrel. Even though Europe does amazing things too, i think Americans are also often very good guys in the world, i have not gotten into conflict with them much when they visit. But worldwide terms, we have difficult times ahead, and NATO should remain together especially when us stepped off from UN regulators spot. US needs the nato too, in that sense Trump
04:49 wickedrain: as president of US should keep nato together. Europe is as good as us in science. just have a look at austrias science and swizz and germans uk etc. It's amazing dream but not it's where we are today. Estonia is very rich in certain resources, some might want to take it indeed if not all, same thing with me, i am very rich in positive permutes or favourable ones, but i know i heal up as
04:49 wickedrain: from now on, but i do not want those abusers anymore around me like thousands of estonians or mob dreamer fuckers gloria and laura, they deal with their own gangstas and get abused back whenever they show their faces or talk trash about real humans.
06:47 wickedrain: blob:https://www.facebook.com/deed62a7-b457-4619-a17a-1ae7a09d793d name is Mart Martin, it's a real name of mine. Joss Jossins was my latvian gamer name it was just a joke. I knew some latvians when i was young, great boys they were. But one of the algorithms seems to yield undefined results if first index value is 1 in the paper, cause 1-1 is 0 and logarithm of 0 is undefined. Have a look
06:47 wickedrain: at the binary probability.
06:59 wickedrain: oops they actually already say that in the next column, i > 1 so
07:11 wickedrain: i found that googline the internet day before yesterday, i can check for keywords, but i arrived to the same conclusion, science of numbers is very strong, hence not interested in bigger nonsense conflicts, cause i actually arrived to the same spot without that paper, but the paper seems very accurate and hence is way better guide than my talks have been.
07:12 wickedrain: the rest of the corpus of numbers is exactly like told in that paper, so that offset is kept is more like logics function, cause internal adder causes that to be evaluated.
07:14 wickedrain: so the internal adder yields a sum of two encodes as invariant sum and can filter based of logics all that does not match, and hence it reduces the entropy of hw
07:16 wickedrain: in 60s einstein did crazy good calculations and this year they find exact proof of this, it was about pulsar stars passing each other at 7cm distance
07:17 wickedrain: nasa determined that albert einstein was doing the calcs correctly
07:19 wickedrain: it might had been earlier yeah, but that is not the main point, numbers are done correctly, and logarithm is not taken hence it more resembles modulus or division based of logics at runtime but this is very fast then
07:19 Ermine: damo22: you can try to port simpledrm and whatever it needs then
07:29 damo22: Ermine: I am porting the ioctl interface but i cant mimic it exactly so i will need to hack into libdrm as well
07:31 wickedrain: We can make this paper entirely complete with very little modification, that is based of batched calculation selection that i developed, by adding base and those things i talked about and data access needs a shift or dma or unaligned access , cause i never figured out how to do it without those calcs went a bit to ashtray to me
07:40 Ermine: damo22: I guess you only need ioctls for DRM dumb buffers, since you don't modeset anyway
07:42 damo22: yeah but i probably need the basic ioctls as well for the driver, as there is no linux drm to handle them
07:43 damo22: even if most of them are stubs
15:38 alyssa: hrm
15:39 alyssa: if I do address mode optimizing early, I can handle gl/vk properly and that's all working nicely
15:40 alyssa: but then opt_load_store_vectorize does't work, because that's supposed to run late and it's very much unclear to me if I can teach that pass to vectorize the lowered hw load/stores
15:42 alyssa: although my hw thing looks like load_smem_amd so.. maybe?
15:45 alyssa: oh, maybe this does work, it's just not liking the particular pattern here
17:22 alyssa: reordered more passes. meh, good enough
21:02 alyssa: although apparently load_store_vectorize can't vectorize lowered globals resulting from lower_ssbo
21:02 alyssa: Hm.
21:04 dj-death: yeah
21:05 alyssa: dj-death: is load/store vectorization supposed to happen early then?
21:48 dj-death: alyssa: yeah that was my assumption
21:48 dj-death: alyssa: I'm guess it can't work on globals before of robustness
21:49 dj-death: alyssa: I'm guessin it can't work on globals because of robustness
21:59 redisaigner: So the time is mature to leave the IRC channels , i won't come back anymore! Looked as if none was listening not to mention participating, i must be crazy than. good bye.
22:14 alyssa: dj-death: maybe?
22:14 alyssa: it has code to handle globals and i've definitely seen it do stuff for opencl
22:14 alyssa: it just doesn't handle the mixed 32/64-bit stuff nir_lower_io generates
22:15 alyssa: which is.. incidentally the entire problem here
22:15 alyssa: s/here/for my address mode branch/
22:18 dj-death: :(
22:19 alyssa: yeah.....
22:19 alyssa: I have something that works nicely for gl at least
22:19 alyssa: still haven't done any shader work for vulkan
22:19 alyssa: idk how to use fossil-db with my driver
22:33 Shibe: I have a strange issue but I'm not sure where to report it. If I attempt to screenshare on applications running on my iGPU (Ryzen 7800x3d) with my RX580 being my primary GPU, I get corrupted output on all applications I've tried (firefox, chrome, discord, obs). I thought maybe this was a KDE portal issue but I tried with hyprland too. Could this be a mesa problem and should I report it there?
22:33 Shibe: Maybe this is a wrong modifiers being used problem especially since my rx580 doesn't seem to have modifier support at all
22:33 soreau: what version of mesa?
22:34 Shibe: This is what screenshare looks like, https://i.imgur.com/n1wAuz5.png
22:34 Shibe: soreau: mesa 24.2.5
22:34 Shibe: I haven't tried screensharing in this type of setup before, so don't know if this was always broken
22:35 Shibe: this only happens when an application is running on the integrated gpu, but I haven't tried running my entire desktop off of the iGPU display outputs and then attempting screenshare
22:35 soreau: the only thing I know of that might be possibly related is https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31725
22:37 Shibe: I assume to test this I'd have to run my entire environment off this (compositor included)? I can run applications using mesa git but not sure how to get my entire environment to use this on fedora
22:38 soreau: you can install mesa to a nonstandard prefix like /opt/mesa and then point LD_LIBRARY_PATH=/opt/mesa/lib/$libdir/ ./compositor
22:39 soreau: this way, you only use the mesa you built if that variable is set, otherwise it will use 'normal' system mesa
22:41 soreau: the other thing you might consider is trying older versions of mesa to see if it ever worked before the change that this patch fixes
22:53 dj-death: alyssa: fossilize-replay
22:56 alyssa: dj-death: yeah but like.. do I need to capture fossils myself? or will e.g. radv fossils replay on hk?
22:56 alyssa: allegedly I can grab them from steam's cache but that didn't work when I tried it
22:58 Sachiel: I wouldn't expect them to work
22:58 alyssa: OK
23:04 dj-death: alyssa: ah yeah fair point
23:37 DemiMarie: Why are Intel iGPUs so much slower than AMD iGPUs?
23:41 HdkR: DemiMarie: I'd blame die investment towards the GPU
23:42 DemiMarie: HdkR: why does AMD devote more die area to the GPU?
23:42 HdkR: DemiMarie: Seems like they care more about bigger numbers on the GPU :)
23:43 airlied: also pwr mgmt balance
23:44 HdkR: It's all a game of trade-offs
23:48 DemiMarie: One of the factors that made me want to get GPU acceleration into Qubes OS was that M1 die shots showed the GPU being bigger than the CPU.
23:48 DemiMarie: Another factor is that CPUs are terrible at math: the vast majority of their die area goes to control logic, not ALUs.
23:56 HdkR: Yea, GPU is bigger than the CPU clusters in most hardware. Intel iGPU has been historically anemic aside from their top end 48/72EU things that Apple used to ship
23:56 HdkR: https://x.com/Kurnalsalts/status/1848700612181168601/photo/1 Big GPU in the new sm8750 SoC
23:57 DemiMarie: I just looked at a Lunar Lake die shot and they could have doubled the GPU if they left out media, NPU, and IPU.
23:58 HdkR: That's one of the trade-offs people are making. Sticking wacking huge NPUs on the die even if no one wants it
23:58 DemiMarie: The GPU is about the same size as the CPU but has far more area allocated to math.