<muurkha>
jrtc27: has anybody shipped rv32e hardware?
CarlosEDP has joined #riscv
pho has joined #riscv
AhmedCharles[m] has joined #riscv
JosefHolzmayrThe has joined #riscv
<muurkha>
jimwilson: 1088 processors, wow. what's the interconnect like?
<muurkha>
Dave Ditzel founded Esperanto, eh?
rlittl01 has quit [Remote host closed the connection]
<muurkha>
I guess the other question is whether anybody's shipping RV32 hardware with an MMU
kailo has joined #riscv
davidlt has quit [Ping timeout: 252 seconds]
riff-IRC has quit [Remote host closed the connection]
riff-IRC has joined #riscv
EchelonX has quit [Quit: Leaving]
BOKALDO has joined #riscv
<geist>
muurkha: i haven't personally seen a rv32e, but someone on another channel was working on some little device and we ended up finding out it was rv32e
frost has quit [Remote host closed the connection]
frost has joined #riscv
<gordonDrogon>
boot to BASIC :-)
winterflaw has quit [Ping timeout: 276 seconds]
jwillikers has joined #riscv
<sorear>
muurkha: impossible to know, andes among others has a licensable core with a sv32 mmu
<muurkha>
hmm
jwillikers has quit [Remote host closed the connection]
jwillikers has joined #riscv
jwillikers has quit [Remote host closed the connection]
frost has quit [Ping timeout: 256 seconds]
jjido has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
jacklsw has joined #riscv
zjason` is now known as zjason
balrog has quit [Quit: Bye]
balrog has joined #riscv
mahmutov has joined #riscv
<jimwilson>
1088 minion + 4 maxion = 1092 processors total
<jimwilson>
esperanto web site says mesh style network
kailo has joined #riscv
smartin has joined #riscv
aburgess has quit [Ping timeout: 240 seconds]
nexgen has joined #riscv
nexgen2 has joined #riscv
nexgen has quit [Client Quit]
nexgen2 has quit [Remote host closed the connection]
nexgen has joined #riscv
nvmd has joined #riscv
nexgen has quit [Quit: Leaving]
nexgen has joined #riscv
nexgen has quit [Read error: Connection reset by peer]
nexgen has joined #riscv
nexgen has quit [Read error: Connection reset by peer]
nexgen has joined #riscv
jacklsw has quit [Quit: Back to the real life]
nexgen has quit [Remote host closed the connection]
nexgen has joined #riscv
Andre_H has joined #riscv
vagrantc has joined #riscv
marcj has quit [Ping timeout: 252 seconds]
marcj has joined #riscv
nexgen has quit [Quit: Leaving]
nexgen has joined #riscv
plantman has quit [Quit: Leaving]
davidlt has quit [Ping timeout: 250 seconds]
aburgess has joined #riscv
<dh`>
for my pet applications (teaching and experimental OSes) 32-bit is usually preferable
<muurkha>
also, for embedded
<muurkha>
though, why would 32-bit vs. 64-bit matter for teaching and experimental OSes?
<dh`>
that ends up meaning 32-bit with MMU and general-purpose-computer peripherals
<dh`>
it doesn't _matter_
<dh`>
but 64-bit has no advantages and the larger pointers are a measurable drawback when debugging kernel problems
<muurkha>
oh, because they're harder to read?
<dh`>
yeah
<dh`>
it is a very simple effect, but it's very real
<muurkha>
yeah, I've run into that problem in non-kernel stuff
<muurkha>
where I wasted 15 minutes trying to figure out why the same input was producing two different outputs until I finally realized it wasn't the same
<dh`>
it's much worse for kernel stuff when your virtual memory doesn't quite work
<sorear>
not a problem if you suppress leading zeros and have a s390x-style split address space :)
<muurkha>
heh, that makes sense, dh`. I hadn't thought about that
<muurkha>
and 16-bit is too small to be practical?
<dh`>
yeah
<dh`>
24 bits would probably be the best size but no useful hw comes in that size for obvious resaons
<sorear>
do you count 68000
<dh`>
(the 80286 does exist, but well.)
<dh`>
68000 is 32-bit in practice
<muurkha>
20-bit seems like a nice size
<dh`>
1M of space isn't really enough to give the feeling of the general case
<sorear>
80286 so students can get practice with memory segmentation and the Multics ring system
<muurkha>
haha
<muurkha>
20 bits with segmentation? 12 trits might also be reasonable
<muurkha>
in base 27 maybe
<dh`>
I suppose one could design a 24-bit architecture just ignoring the proposition that sizes are normally aligned
<dh`>
but then for both teaching and research you get hosed for your compiler and toolchain
<muurkha>
you could use the 68000 or S/360 approach
<muurkha>
15 trits in base 27 might have addresses like A!EEIZ
<jrtc27>
64-bit makes debugging kernel easier in a way
<sorear>
octal :p
<jrtc27>
if it's screaming fs at you it's a kernel address, if it's not it's a user one
<muurkha>
heh
<muurkha>
A!EIZ I guess
<jrtc27>
also easy to tell junk from real pointer
<dh`>
surely the digits for base 27 should be [0a-z]
<muurkha>
0 would work. @ (or `) would be more "logical"
<dh`>
32 bits is enough for those effects provided you're using small real sizes
<muurkha>
or you could use the Spanish alphabet: abcdefghijklmnñopqrstuvwxyz
<dh`>
using 0 for 0 is probably desirable
<muurkha>
if you were really going to build a ternary computer you'd probably want it to be balanced ternary
<muurkha>
so 0 comes in the middle of the digits
<dh`>
eh, it's all equivalent
Narrat has joined #riscv
<muurkha>
for a mechanical computer you might want to use a much higher base, like 15, 30, or 60
<muurkha>
because it's routine to machine one-centimeter machine parts to a tolerance of 10 microns, while with transistors and resistors on-chip we're lucky to get everything within a factor of 2
BOKALDO has quit [Quit: Leaving]
cousteau has joined #riscv
cousteau has quit [Client Quit]
Narrat has quit [Ping timeout: 252 seconds]
Narrat has joined #riscv
somlo has quit [Remote host closed the connection]
<gordonDrogon>
The 65c816 has a 24-bit adress bus, but it really never was usefull.
<gordonDrogon>
(although apple //gs owners might disagree)
loki_val has quit [Remote host closed the connection]
crabbedhaloablut has joined #riscv
somlo has joined #riscv
<muurkha>
well, it was capable of doing amazing things, but getting it to do them was a grinding puzzle-solving exercise
<gordonDrogon>
I currently have a retro hobby project using the 65c816 that I'm looking to move to risc-v..
<gordonDrogon>
I've written far too much '816 assembler - I don't find it enjoyable anymore )-:
<muurkha>
have you tried loading VexRiscV or something into an FPGA?
<dh`>
a 24-bit address space made out of 16-bit banks is definitely not the answer
<muurkha>
that's something we can all agree on
vagrantc has quit [Quit: leaving]
smartin has quit [Read error: Connection reset by peer]
smartin has joined #riscv
eigenform has joined #riscv
nexgen has quit [Quit: Leaving]
nexgen has joined #riscv
EchelonX has joined #riscv
lagash has quit [Ping timeout: 252 seconds]
cwebber has joined #riscv
smartin has quit [Ping timeout: 268 seconds]
Narrat has quit [Quit: They say a little knowledge is a dangerous thing, but it's not one half so bad as a lot of ignorance.]
cwebber has quit [Ping timeout: 250 seconds]
jjido has joined #riscv
lagash has joined #riscv
<gordonDrogon>
my retro system runs bcpl, so I'm in the process of slowly writing a risc-v emulator in bcpl (32IM) then use that to bootstrap my bcpl OS then find a nice hardware/fpga platform - esp32-c3 is a candidate - hard to find something with 0.5 -> 1MB of RAM as it seems to be "microcontroller" level or full-blown Linux...
<gordonDrogon>
(and yes, this does seem somewhat masochistic, but everyone needs a hobby ;-)
nexgen has quit [Read error: Connection reset by peer]
<dh`>
using bcpl is the sticking point there :-)
<dh`>
granted I've never understood many of the facets of retrocomputing
<drmpeg>
I think if you've lived through it, retrocomputing isn't that compelling.
<dh`>
maybe
<drmpeg>
RISC-V is more compelling to me. Uncharted waters.
<dh`>
one of the major strains seems to be running modern SW on archaic HW, and the people who do that seem to be older
<dh`>
as in, they used the HW when it was current
<dh`>
another is running archaic SW on modern HW, and that seems to mostly attract people who used the SW when it was current
<sorear>
aren't those the same group of people though
<dh`>
there's less overlap than you'd think (IME)
<sorear>
(and "archaic SW" gets fuzzy with some timeless SW)
<dh`>
well, I'm thinking of things like the people who maintain acorn riscos
<dh`>
(or its clone or whatever it is, not clear on the exact details)
<dh`>
or the many amigaos-related projects
<sorear>
how do you feel about Maxima
<dh`>
I have never had occasion to have an opinion about it
nexgen has joined #riscv
<dh`>
but for example while I might patch the existing SNOBOL tools if they stopped compiling, I can't imagine writing new software in SNOBOL