#avr Logs

Jun 26 2022

07:00 AM specing_ is now known as specing
08:01 AM josuah: hello rue_mohr, did you happen to be the author of https://paste.josuah.net/LhbQ4a64HJvEPpk0ybab/article.txt ?
08:05 AM qu1j0t3: lol the author seems not to realise supercomputers existed in 1967
08:05 AM qu1j0t3: they were using it for many of the same things we use it for today
08:06 AM qu1j0t3: weirdly phrased article, wait until they find out about the 30 years of computer science that predated 1967
08:08 AM qu1j0t3: even fortran was 10 years old by then
08:10 AM josuah: maybe it was hilighting one particular implementation used at many places, rather than the one algorythm that everything relies/relied on?
08:19 AM qu1j0t3: i know of cooley-tukey, my uncle sent me the original paper in the late 80s
08:19 AM qu1j0t3: i used it to generate fractals.
08:19 AM qu1j0t3: tukey is a cool dude e kills fourier and doesn't afraid of any scalar
08:20 AM qu1j0t3: it was an advance on earlier methods iirc
08:32 AM josuah: qu1j0t3: early supercomputer, does it mean the CRAY among others?
08:43 AM qu1j0t3: josuah: e.g. cdc 6600
08:44 AM qu1j0t3: before cray
08:46 AM josuah: is that not an awful amount of wire-wrap to eventually debug if anything goes wrong?
08:47 AM josuah: hmm, there might be just as much with PCB tracks after all...
08:49 AM qu1j0t3: well, the idea is that the circuit must be correct. you're only fighting mechanical failure which is quite infrequent.
08:50 AM qu1j0t3: 6600 would have been subject to discrete transistor failure but at a much lower rate than vacuum tubes
08:54 AM qu1j0t3: josuah: most of those interconnections are in silicon now and extremely reliable. the pcb carries mostly power and obviously interface lines
08:56 AM qu1j0t3: josuah: incredible to think 6600 had no integration, all discretes :)
09:01 AM josuah: I had no idea that supercomputers could be made of discretes! :O
09:03 AM qu1j0t3: state of the art at the time
09:03 AM josuah: and still an art as of today I'd say
09:04 AM qu1j0t3: 6600 modules are floating around ebay
09:38 AM qu1j0t3: josuah: anyway for examples of what we were doing with the fastest computers then, standout ones would be seismic (oil/gas), nuclear sim, signal processing, etc. military always has first dibs on the fastest new products
09:38 AM qu1j0t3: josuah: that was as true in the 50s as since then
09:44 AM * josuah imagine being a little mice in the reasearch labs of militaries
11:42 AM Cracki: goose - geese. moose - meese. shoop - sheep. mouse - mouses. mice - mices.
12:01 PM rue_mohr: josuah, I think so, yea
12:01 PM rue_mohr: well I re-digitized it
12:02 PM rue_mohr: sounds like my comment
12:02 PM rue_mohr: oh yea, with the lower comments, definitly me
12:02 PM rue_mohr: the orig is right...
12:02 PM rue_mohr: http://ruemohr.org/~ircjunk/programming/FOUR1.txt
12:03 PM rue_mohr: dont remmeber where I posted the comments
12:03 PM rue_mohr: cool, where did it come up?
12:03 PM Cracki: fortran doesn't appreciate spaces between identifiers, eh
12:04 PM qu1j0t3: right
12:04 PM Cracki: *tokens
12:04 PM qu1j0t3: does not use spaces
12:04 PM Cracki: the infamous do5i bug
12:04 PM qu1j0t3: which makes sense when you consider punch cards
12:04 PM rue_mohr: oh its probably from github..
12:04 PM Cracki: or was it do3k or something like that
12:04 PM Cracki: do15i
12:05 PM rue_mohr: are you running it on an orig fortran thing?!
12:05 PM rue_mohr: :)
12:05 PM rue_mohr: https://github.com/ruenahcmohr/max7219fft
12:06 PM rue_mohr: still cant find where I posted *that* text with it
03:58 PM josuah: rue_mohr: it is to be put in the Opust 5 of The Gopher Times here: gopher://bitreich.org/1/tgtimes/ - https://gopherproxy.net/bitreich.org/1/tgtimes/
03:58 PM josuah: if you feel like it, I can publish it with your name, or anonymously, or not at all
04:30 PM rue_mohr: sure stick my name on it! I had a great adventure reworking that code!
04:31 PM rue_mohr: I properly converted it to C, as per the code on github (which is an 8 bit implementation, but its easy enough to do more)
04:31 PM rue_mohr: ** at 8 bits, you only get 3 bits of output amplitude
04:33 PM rue_mohr: what astonished me most, is that it was a hack from the start in 1967
04:33 PM rue_mohr: hacking other peoples code has been there right from the start of computers
04:35 PM rue_mohr: congrads on finding me to pin that down to!
05:13 PM josuah: rue_mohr: I had help from others in ##stm32
05:15 PM josuah: thank you, it is in the box! :) I will let you know when the whole opus is published
05:17 PM josuah: on another topic: what do you think of the 8051-based microcontrollers?
05:18 PM josuah: it look similar to AVR in it is there since long, and 8-bit, with 16-bit address and pure harvard arch
05:18 PM josuah: they are used in the EFM8BB
06:04 PM rue_shop2: 8051 never went away, so it must have not been as bad a thing as it always seemed
06:04 PM rue_shop2: I have a rule about not jumping on the boat of each new arch that comes along,
06:05 PM rue_shop2: which is a rule I cant hold against 8051 cause its as old as time
06:05 PM rue_shop2: (for all intents and purposes)