10:18 kawasaki: i could not get the fft alike scala idea to work reliably, but only two of my own creations do work reliably from past logs. Old dog does not learn new tricks.
10:43 kawasaki: transition graph is the best of them which has couple of ways to get implemented other one is where i do not remember where the logs went, it was twice the value and inverse magic which worked in descending order unless indexes inverted, that while ago i represented. but the transition graph is type of universal for data and code, i like this the best, babylonian march 5.
10:43 kawasaki: so i start to do the banks with that similar logic as in logs in march 5 by babylonian.
13:18 tursulin: mlankhorst: If I got it right you seem to be handling the next drm-misc-next feature pull? If so could you please merge https://lists.freedesktop.org/archives/dri-devel/2024-April/448402.html?
13:20 tursulin: ah not online.. should have checked first
13:36 kawasaki: I am finding the logs soon cause they where interesting, they demonstrated how the access of lists happens in ascending sry not descending order, but it might matter if the operation on them is subtract or add. So if there are duplicates it goes smallest first.
13:38 kawasaki: i am testing the whole week from today, but i see that it might be possible, cause i did a few set of tests before too on my half bake invented methods, and they were promising.
13:44 kawasaki: the meaning of this sentence is that when transition values are offered as banks, you offset the slot through subtraction leaving a remainder that increments the value added to the hash, and then it processes out of all banks smallest first and out of all transitions smallest first.
13:47 SolarAquarion: what are the mesa tools?
13:48 SolarAquarion: their optional for testing and stuff?
13:48 SolarAquarion: they're
14:06 kawasaki: but the way i understand duplicates can not happen if one increseas every bank of data or sets of elements to maximal value of presentable number per every upcoming bank + such offset is added, collision is impossible then
14:06 kawasaki: and it would and up or basically has to function that way according to my tests
14:16 kawasaki: so if this hash or set of subsets has 1024 alu bank sets accepting operands at different offsets, the 1024th bank elements are all 1024 times bigger and you later subtract 1024times1024 and that 1024 comes from maximal compressed representable value, if the value is max 4096 then the step is 4096
14:17 kawasaki: so that much testing i did, and that should be how the hardware would be built down in systems hearts, or is bit similar tbh it has to work like this
15:44 kawasaki: deuchland deuchland über alles, cannabis legalized there now. But i have to inspect one more thing, it might be possible to only use one selected element offset to increased value, than it saves space, cause others get eliminated, it seems like worth idea, allowing to stash more alus.
15:46 kawasaki: cause logics say it does not care if the banks get confused on transition values, cause after all that is the same value.
16:49 kawasaki: whoa, technically that might work too even, so increasing remainder, ascending order processed since no duplicates, descending order output, so 3 banks 300+200+100 at top operand, and 50+100+150 bottom of the op subtract, that is followed by three banks of max range 1to50 answers
16:56 kawasaki: now you add 30 to the register marking an incoming operand to the pc1, 300-50+30 is 280...so you do the babylonian trick now 280 is downconverted to a transition value
17:04 kawasaki: i am getting letters from austria tomorrow gonna pick it up, i do not know what this is about though, but austrians made bioprinter at upnano.at which uses quite similar looking technique, it's nuclear tech triple triple photon annihillation upconversion, it's similar to some ideas of w3c and my labs too, prolly the neat trick is possible to implement anywhere.
17:07 kawasaki: i am also getting letters how i took studen loans and if it was not you do not respond, and i did not respond, cause i never took any of such, so the criminals who bothered me, did much harm it seems.
17:11 kawasaki: it is a criminal syndicate which where i am not involved in, i spotted 50 accounts in skrill system that i had never made, and there is no way that my memory do not serve.
19:01 Ntemis: @marcan you around?
19:02 Ntemis: am looking for hdmi audio out for asahi kernel 6.6.x
19:02 Ntemis: *patches
19:02 Ntemis: anyone that can help me out please?