ChanServ changed the topic of ##yamahasynths to: Channel dedicated to questions and discussion of Yamaha FM Synthesizer internals and corresponding REing. Discussion of synthesis methods similar to the Yamaha line of chips, Sound Blasters + clones, PCM chips like RF5C68, and CD theory of operation are also on-topic. Channel logs: https://freenode.irclog.whitequark.org/~h~yamahasynths
* ej5
awkwardly finds out he's not following her. wtf, thought i was. just fixed that problem ;)
<whitequark>
i mostly shitpost i think
<ej5>
hey everyone does it, sometimes shitposts really get some mileage
<TD-Linux>
whitequark, your shitposts are of high, and consistent quality.
* ej5
thinks back to some particularly stupid pooh bear memes
<whitequark>
TD-Linux: damn! failed again!
<cr1901_modern>
I mostly... RT. B/c I've very much fallen prey to the outrage machine.
<cr1901_modern>
once in a blue moon I remind ppl that I actually know something about a technical topic lol
<ej5>
it's really hard not to. i try to keep my twitter apolitical. but it's so hard
<TD-Linux>
whitequark, oh also your high ratio of OC to low effort dunking is incredibly high for twitter, which helps
<ej5>
btw if you follow someone new today, make it @MLE_Online. she's very cool, has lots of neat projects.
<cr1901_modern>
Looks like I viewed them recently
<cr1901_modern>
ej5: Follow count of mine is currently locked
<cr1901_modern>
it hovers around my Dunbar's Number of 286
<whitequark>
i don't keep my twitter apolitical but before i dunk on someone or rt something like it, i ask myself a simple question: "what am i trying to achieve here?"
<ej5>
hmm, dunbar's number is apparently about 150.
<whitequark>
so if i want to drag some asshat with 15 followers, i swear at a wall instead.
<cr1901_modern>
(It's not my actual Dunbar's number)
<whitequark>
if the said asshat is an ex-CISO of Facebook, into dirt he goes.
<ej5>
^ nice
<cr1901_modern>
FB needs to be turned into compost
<ej5>
oh it's so bad. so, so bad.
<whitequark>
in particular i assume everyone who follows me already knows about the nazis in 2019 america and they don't need reminders every five goddamn minutes
<whitequark>
(in fact i block people for doing that so they don't pollute my feed)
<ej5>
very familiar. have a brother who's antifa, so i get to hear all the things
<cr1901_modern>
hehe
<whitequark>
keeping political shit *actionable* seems to work well enough for me, is all
<cr1901_modern>
ej5: I'll buy him a milkshake if I ever meet him
<cr1901_modern>
on me
<cr1901_modern>
to be used for whatever purpose
<ej5>
well, if you're ever in portland lol
<TD-Linux>
whitequark, I think you're doing a great job, btw.
<cr1901_modern>
ej5: The ugly side re: RTs, is that _in general_, I don't suffer from outrage fatigue
<cr1901_modern>
and I fail to appreciate that lots of ppl aren't like that
<ej5>
wow that gravislizard post really hits home. i've read it several times and it's really good.
<cr1901_modern>
Yea, it sums up my feelings better than I ever could
<cr1901_modern>
and it's not like a post most ppl can make w/o being called out for being arrogant
<cr1901_modern>
or looking for attention
<cr1901_modern>
or some other ulterior motive. Sometimes it just stinks to miss the boat
<TD-Linux>
I've had to unfollow/mute some followers because they are anxiety factories (thanks twitter incentives) but not a problem for your posts
<ej5>
the first comment from narunetto is actually really good, about LGR's lack of technical depth
<TD-Linux>
(tbh gravislizard is in the first category for me)
<cr1901_modern>
gravislizard's serious posts need to be said IMO
<cr1901_modern>
but I understand
<whitequark>
TD-Linux: good.
<ej5>
LGR is a consumer, but he doesn't understand the technical underpinnings which are often REALLY fascinating
<cr1901_modern>
If you watch his Mario Teaches Typing vid, you will hear him discuss how the 8259 works in decent detail as part of a skit
<cr1901_modern>
that makes me think he actually _does_ understand some of the technical underpinnings
<cr1901_modern>
but I still think it's mostly cash flow at this point for him :P
<ej5>
yeah that was 6 years ago for him lol
<TD-Linux>
solid technical underpinnings + excellent production value is very hard and expensive sadly
<cr1901_modern>
ej5: When I'm better, or if/when I can get a suitable CPU board, I think my next retro project is gonna be a custom ISA bus system using one of the FOSS 8-bit backplanes
<cr1901_modern>
w/ even a custom CPU maybe (20 bits of I/O should be enough for anyone)
<ej5>
cool
<whitequark>
boneless in hard logic wen
<whitequark>
(you could probably microcode most of it, tbh, especially if you're fine with using 8/16 CPI)
<cr1901_modern>
Wasn't even thinking that hardcore lol. Boneless w/ that 74xxx ALU type, huh?
<whitequark>
yeah, sure, why not?
<whitequark>
if you're willing to lay out the board i can help out with that
<whitequark>
i know how i'd design that, anyway, so that you only need one ALU and it's used for address calculation too
<cr1901_modern>
I will keep that on the backburner, tbh :). Right now I def couldn't undertake the project. My custom CPU would've either been a glasgow plus GlasgISA adapter or just a plain old ISA board w/ FPGA breakout
<whitequark>
myeah
<cr1901_modern>
or third option I just thought of- the ISA backplane is FOSS. I could put the FPGA plus some hyperRAM on it directly.
<cr1901_modern>
(or SDRAM, whatever)
<cr1901_modern>
whitequark: Depending on how much you want to "cheat", for want of a better phrase and/or save costs, parts of Boneless could be done w/ EEPROMs to implement microcode program
<whitequark>
cr1901_modern: that's the idea yes
<whitequark>
well, not necessarily EEPROMs. could use EPROMs. or just PROMs.
<cr1901_modern>
right, whatever medium-scale-integration solution works best for you :)
<whitequark>
cr1901_modern: most boneless instructions can be decoded using 7 bits plus some muxes
<whitequark>
make it 8 bits indexing into microcode and you've got every instruction
<whitequark>
make it 9 bits and you've got multicycle instructions too
<whitequark>
you could almost certainly pack it tighter if you really tried
<whitequark>
so it doesn't even have to be a large ROM
<whitequark>
something like 16x512?
<cr1901_modern>
The few TTL impls I've heard of (Magic-1, NANDputer, gigatron) already have multicycle insns.
<cr1901_modern>
Rob Baruch's riscv impl might be multicycle as well
<whitequark>
so in the FPGA impl, the only multicycles are shifts, window instructions, and complex jumps
<cr1901_modern>
If one sticks to single cycle... well, tbh I don't think anyone's gonna care about the speed loss in 2019 :P
<whitequark>
you can't easily make a single cycle boneless, i mean you can but it becomes much less compact
<cr1901_modern>
oh nevermind then. Just thinking out loud
<whitequark>
you could make a boneless where memory accesses are multicycle too
<whitequark>
then the address calculation would be done in the ALU too
<cr1901_modern>
might be beneficial on an ISA system (wait states for memory were in fact common even back then, so nobody reliving nostalgia- clearly the target audience ;)- is gonna notice/care)
<whitequark>
you really don't wanna wait states for boneless, given that your register file is in ram
<whitequark>
in the FSM implementation, each single cycle is a load or a store
<whitequark>
(with true dual port RAM, the 4th cycle is a load AND a store)
l_oliveira has quit [Quit: ChatZilla 0.9.92-rdmsoft [XULRunner 35.0.1/20150122214805]]
<cr1901_modern>
Until semi recently, Cypress _made_ DIP dual port RAM. I have a specimen; it's meant for video memory
<cr1901_modern>
(okay, the point is less that it's DIP, and more that there was a market for jellybean dual port RAM)
<whitequark>
cypress makes tons of weird shit
<cr1901_modern>
now if only I remember where I put it to get a part number
andlabs has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
andlabs has joined ##yamahasynths
andlabs has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
andlabs has joined ##yamahasynths
_whitelogger has joined ##yamahasynths
Xyz_39808 has joined ##yamahasynths
cr1901_modern1 has joined ##yamahasynths
cr1901_modern has quit [Ping timeout: 258 seconds]
ej5 has quit [Read error: Connection reset by peer]