01:57 agrisis: so some more info on why elogind isn't cutting it for chvt w/ drm/kms app running: Elogind does not manage virtual terminals.
11:00 EdB: airlied: may be it could be nicer to have a struct like "struct arg_info" in module.hpp to store printf info and the format kind
13:11 pcercuei: drm-misc-next doesn't include v5.10-rc1?
13:30 nipung271: Hi. I had a question - is it possible to use the llvmpipe rasterizer with vulkan?
13:31 kisak: versus what lavapipe has?
13:31 kisak: (I'm indirectly asking if you're aware of lavapipe existing)
13:34 nipung271: No, I'm not aware of lavapipe
13:35 kisak: well good news, the next release series has a not-yet-conformant vulkan on the cpu implementation
13:36 nipung271: Okay. Is the dev branch for that release available?
13:36 kisak: mainline mesa git master
13:36 nipung271: I see.
13:38 nipung271: Is the process for building this similar to llvmpipe?
13:38 nipung271: There is `src/gallium/targets/lavapipe` in the main repo. So, if I build this, will this give me something that I can drop in place of libvulkan.so then?
13:41 kisak: sorry, I haven't tinkered with it myself to know more details, but I doubt that you want to override the vulkan loader
13:44 nipung271: This isn't for running production. What I'm looking for is to run small headlessCI tests on the CPU side.
13:44 nipung271: So, yeah, I do want to overload the vulkan loader for this? Could that cause problems?
13:54 sravn: pcercuei: They are waiting for -rc2 to surface, so expect something next week
15:31 pcercuei: sravn: alright, thanks
15:51 Lightkey: nipung271: Are you familiar with Hyperspace Delivery Boy!? ;-)
16:12 nopleasure1: You are just high end idiots (you that are attending those channels), and the code speaks for itself also, red hat and collabora and khronos projects are all retarded btw. And you seem to be messing with opencl driver for more than ten years without any success stories present, that is awful when such hve the courage to tell things to me. All major simulators have opencl 2.0 drivers ontop of your kernel, you can get very easily with even generic
16:12 nopleasure1: driver of opengl as backend to gem5-aladdin and smaug and tvm, not to mention m2s-cgm and the likes.
16:16 emersion: daniels: ^
16:16 nopleasure1: such drivers are not meant to be built with ten years of work, but a trained programmer authors them with couple of months work.
16:20 nopleasure1: yes of course i took the latest amd opencl 1.2 driver from closed userpsace too, this ocra one, the only one they offered, it is just incredibly weak work, this is nuts.
16:34 nopleasure1: how can you even not write nowhere near sane code is beyond me, go apply for place in paraolympics karolherbst, they did not accept me, you have a reasonable brain injury with that other tranny.
16:52 nopleasure1: Here i boot out of irc with ignoring my possible further rage too. What kind of company is google also anyways or cisco, earning billions of dollars from entirely broken code, the likes of h264/h265 most obviously have been done by very broken minds too.
17:11 nopleasure1: As UPWORK programmers are total nutters as well, manage to fuck up the scheduling and interrupt premptions, because they do not know how to take screenshot after every 10seconds, very decent atom in order processor just freezes on its hands hence.
17:11 nopleasure1: this is incredible bogus really.
17:13 nopleasure1: so i rolled a gauzy modifications and chromium devtools support from master, luckily this one is open source.
17:13 nopleasure1: electron javascript engine
17:14 nopleasure1: actually now that i remember her computer is not even atom it is pentium microarchitecture and ivy bridge gpu, this flavor.
17:27 nopleasure1: you know dudes, compute shader of opengl isn't anything special or opencl as to how it works, i know everything about how it is handled, you can lift clang6 opencl 2.0 to opengl using fragmend or vertex shaders on the backend , glViewport is so old thing, every hw supports this pseudo-op on the ring
17:29 nopleasure1: so for instance smaug allows any of the language front-end that can generate X86 LLVM ir to be using gpu backends, you know all it needs are the alus really
17:29 nopleasure1: everything else the tensor kits generate i chose TVM this time instead of MLIR or something else.
17:31 nopleasure1: there is no need for another and another IR basically, whatever really why do not you read the hw instead.
17:32 nopleasure1: metal is allready LLVM IR, and vulkan has cpu backends for X86 basic LLVM IR too
17:34 nopleasure1: why do you even see a delusion that program binaries need SPIR-V to be dumped, this dutch guy had wladimir van der laan to dump program binaries in machine code , they must have assemblers and all do really only.
17:35 nopleasure1: you can also dump the old LLVM IR no spir/V needed.
17:37 nopleasure1: why the fuck would someone want to use multiple cores to access memory over IO bus behind the bridge and belonging to GPU anyhow, are those crazy really?
17:38 nopleasure1: #osdev guys also big bullies but yeah all their text is so called camel SALIVA where they tried to make a point albeit failed entirely each and fucking every time dudes.
17:40 nopleasure1: trying to pull me into their troubles and get me into major conflict, and not much to wish but for them to score a miserable loss as my supporters all do wish. IF this mental shit goes on a war will be declared basically to them anyways.
17:41 nopleasure1: have you even noticed what is happening in the world, soon you see what happens to conspirers in my country also.
17:43 nopleasure1: it is not even much the things which my direct supporters wish, agressor and violator persistent one to be beaten, really all wish that tbh.
17:44 nopleasure1: and they will be crushed so bad if it goes down this way really, so like Borat says a lizards wont even breathe in their areas.
18:18 airlied: overnight marting
18:18 airlied: EdB_: you mean instaed of the flag to buffer type?
18:18 airlied: EdB_: I'm have a struct in there for the the format strings already