03:52 garyleibniz: Well someone fixed intel's microcode and touchpad issue, quite quickly after i reported. I am also done with my modifications and will get a new job and bring my startup into play, i am also quite experienced in programming, so that caveat for x86/arm/risc-v/sparc/mips etc. asics and accelerators like gpu in fw's that we can not report timing based of very high resolution timing, is not
03:52 garyleibniz: dramatic issue, it can get to approximate high resolution timers , I do not know electric vechicles architecture entirely if that is better off with only using FPGA's based which can do it easily, anyways i am starting to look at vienna's last XDC tomorrow. I do not know why 2ps resolution timer is needed there. AUTOSAR is europes car manufacturers specifications, you can find a lot of
03:52 garyleibniz: pdf's about it, the one i looked stated that those vehicles come with 80microcontrollers on board integrated. Those car's are heavy due to battery weighting lot, but at the same time very fast on roads. American tesla's are similar. Elon Musk. Drones are also such popular vehicles. And irreplaceable in modern war, i am soon doing also a little investigation if those very high resolution
03:52 garyleibniz: timers hold anything back, i doubt my own. Also how accurate and close can we report timing to ps range, i stated i have known for years how to jailbreak ios and android phones, but i do not do illegal activies, inspired by one old american man randomly passing by me, saying i fight mdm -- mobile device management, if there is a very compelling need i do it, otherwise to spam channels that
03:52 garyleibniz: is not compelling enough of a reason. the vulnerability is in the ethernet layer designed with hazards, it gets in currently to anything, as said i never distribute illegal activities, iphone has integrated firewall which blocks stash proxy like app stores app Fair socks5/http rndis gateway. I use iphone for the past year or so now.
04:30 garyleibniz: I want to say something small about execution of such magic, i was trolling you with combinations when connor commented, it's that maybe you also know, that packed bitcount yields invariance as modifier, and all asic hardwares i know for cpus and also what i have can do bitcount instructions indeed, if you sideload 1to64 as two operands bitcount, yes it becomes very easy to run the code,
04:30 garyleibniz: but we'd agree we do not commit this in open source code, but land a smaller version in fw, i am not paranoid about dying at all, however i do not wantt to still, but once something happened with backteria multiplication and iran had a report of that one that 1tesla energy was needed , to cause huge problems to human being organs, and such opportunity general public should not commonly
04:30 garyleibniz: have, i underestimated by a mile of other envious or angry hackers.
04:58 hummercarhumvey: I think those were estonian fraudsters, who tried to kill me, i was resting in home and prepared to die, 4+bacteria poisoning later detected, people cried and yelled outside, mart wtf. go to a doctor, i just had so enough of that shit, what my family and estonians arrange that i was just sleeping at home and drinking, i could feel the bacteria easily and heart was already messed up and
04:58 hummercarhumvey: painful, then finally i called to police who never arrived someone took me to hospital and i got surgery and treatment , if you thought i was kidding, they killed Edgar Savisaar alike. In my case it was one of the most vicious fraud this estonian human trash presented to world. But they have given up, investements into military have vanished already corruption , Estonia is not going to
04:58 hummercarhumvey: fight with Russia. And imo they deserve to die, i finally went over to russian side, cause i want to make sure they will.
05:03 HdkR: 💃
05:41 santaclausew: Those were jewish estonians in most part leading that terror, all the way through in life of mine, dad , granny all jews and their friends, they speak fluent estonian and terror others to go against me, on the day i had beamer on the lightbulb , and other additional beamers as jewish crooknoses outside visibly seen, hence it's easy fix to go to putin's side who has a program against jews
05:41 santaclausew: and the other amoral estonians, but those all methods including the last chip that tries to finish me, are known jewish worldwide terror methods, iran just reported that and the fun or not fun part is, i never did anything bad to those, hah to deserve that 35years victimized state. So we start to kill all the relevant ones again. Me first time doing any for of killing all in all, but needs
05:41 santaclausew: to be done. Good night and life, i am getting close to Fidel castros record of assassination attempt , after joining Russian side.
05:42 HdkR: 💃
11:52 zmike[d]: has anyone tried the piglit sample locations tests with zink yet?
12:04 rhed0x[d]: hummercarhumvey: wtf is going on here
12:55 karolherbst[d]: seems like blocking those VPS IP ranges does start to work
12:56 karolherbst[d]: already saw 2 or 3 auto kicks
16:26 mhenning[d]: zmike[d]: Not that I'm aware of
21:00 mhenning[d]: marysaka[d]: do you have any ideas why envyhooks might fail with dEQP? eg. I can trace vkcube just fine but if I try to run dEQP-VK.api.smoke.triangle in envyhooks I get "Fail (vk.createDevice(physicalDevice, pCreateInfo, pAllocator, &object): VK_ERROR_INITIALIZATION_FAILED at vkRefUtilImpl.inl:417)"
21:01 marysaka[d]: I'm not too sure I saw something similar on my end recently with new blobs and still haven't found time to debug it
21:01 marysaka[d]: it might be my dup fd tracking changes maybe
21:02 mhenning[d]: okay, thanks. I'll try to hack at it
21:42 chomwitt: HI. i have a gtx 970. Can nouveau set the clock for that model ?
21:42 gfxstrand[d]: Unlikely
21:42 chomwitt: gfxstrand[d]: thank you.
21:43 chomwitt: why nvidia wouldnt allow that for 10+ old model ?
21:44 gfxstrand[d]: It's less that they won't allow it and more that they're not willing to put in the work to help us enable it.
21:45 chomwitt: sorry to hear that.
21:46 chomwitt: but if it's not a matter of policy but just finding time then there is hope i guess.
21:51 orowith2os[d]: Aren't the 9 and 10 series limited by firmware signing?
21:52 orowith2os[d]: which is something only NVIDIA can do; meaning they would have to hand over firmware that Nouveau can redistribute and utilize, similar to GSP.
21:52 gfxstrand[d]: Finding time for 10-year-old GPUs isn't going to happen
21:53 gfxstrand[d]: orowith2os[d]: They've already done that. Those are the firmwares we run today. But they don't support reclocking.
21:53 orowith2os[d]: oh, interesting.
21:53 orowith2os[d]: didn't know that.
21:54 orowith2os[d]: it just allows them to boot enough to run some slow graphics, and display?
21:54 gfxstrand[d]: If someone did the work to strip firmwares out of the latest legacy blob driver, R/E the interfaces, and getting nouveau running on them, we could maybe figure out how to ship that. Might not be able to legally put them in linux-firmware but I'm sure someone could bundle it in rpm fusion or whatever.
21:55 gfxstrand[d]: But if someone's counting on NVIDIA to do that work and/or to release new FOSS-friendly firmwares that do reclocking, they're going to be waiting indefinitely.
21:55 steel01[d]: Is there a tl;dr on why the separate firmware exists in the first place? Like, why didn't nouveau back in the day just use the proprietary firmware api?
21:56 gfxstrand[d]: licensing is part of it
21:56 steel01[d]: Like, nvidia has stuff in the proprietary firmware that they can't sub-license?
21:57 steel01[d]: Why would that require breaking api changes, though?
21:59 gfxstrand[d]: <a:shrug_anim:1096500513106841673>
22:00 gfxstrand[d]: Nah, we could probably keep the same uAPI
22:03 steel01[d]: But yeah, I can't even get nvidia to release gv11b firmware, which is more recent. And tegra, which has better oss support than desktop gpus. ><
22:04 frankyomega: writing from my phone, if my memory serves about all the baseband chipset, all the duty cycles packet handling is normally inside the firmware it’s my final investigation and research, I suppose you think I was wrong on dangers since I can not control the clock, what I think its irrelevant , the code can not be accepted due to dangers around baseband firmware is my bet, it has stm32 and
22:04 frankyomega: such microcontrollers and the beamer chipsets but I apologise I can not remember it clearly maybe I am mistaken there , spectrum is large and duty cycle routines are different ones, I think my code is dangerous but its so far a guess that is not educated, if it turns out different case I can upload it. some reading
22:04 frankyomega: https://indico.cern.ch/event/813103/contributions/3389997/attachments/1843559/3023740/190513_NIR_Physiological_Effects.pdf so all duty cycles have constant 20ns latency for my code. that could be door to out of spec beamer.
22:04 frankyomega: so give me a day I am out today
22:06 HdkR: :thonking:
22:08 orowith2os[d]: I'm quite amused that the linux-firmware repo allows you to view the firmware contents on-site in a hexdump.