@FutureFractal you said you wanted to design a computer architecture… wanna help me with a safir one?
FutureFractal23-Mar-20 04:17 PM
Vāstra II23-Mar-20 04:25 PM
I've said before that safir computers are optical in nature and use balanced ternary (0 = off, +1 and -1 = opposing polarizations of light) but of course that needn't apply across space and time(edited)
I'm strangely fond of bit or trit widths that aren't a power of 2 (e.g. 18)
Vāstra II24-Mar-20 05:01 PM
FutureFractal24-Mar-20 05:15 PM
ah sorry, got distracted yesterday :o
that sounds really neat tbh!
Vāstra II24-Mar-20 05:30 PM
What would you like to see in a computer architecture?
Vāstra II24-Mar-20 07:56 PM
FutureFractal24-Mar-20 08:14 PM
ah sorry, got distracted again D:
Vāstra II24-Mar-20 08:15 PM
Pretty much all I do anymore is get distracted lol
FutureFractal24-Mar-20 08:15 PM
anyway, I'd love to see x86-style memory addressing in instructions! :D
x86's memory addressing is one of the only things about it that's unambiguously extremely good compared to other architectures
Vāstra II24-Mar-20 08:16 PM
It's pretty good
FutureFractal24-Mar-20 08:16 PM
Vāstra II24-Mar-20 08:20 PM
contemplates input surfaces. Definitely a telepathic one, maybe also a keyboard (though probably larger than ours because safir have four arms?), and he's also fond of control panels like old computers had
what would it be like to write a program that takes telepathic input tho I wonder
also agreed, old-school control panels are super aesthetic and awesome
Vāstra II25-Mar-20 12:38 AM
Good question :o
I have no idea lol
What input surfaces would you suggest?
Vāstra II25-Mar-20 01:12 AM
FutureFractal25-Mar-20 01:48 AM
input surfaces? :o
Vāstra II25-Mar-20 03:13 PM
@FutureFractal@Shyriath for that matter I'd be eager to hear y'all infodump about y'all's ideal con-computer architectures!
Shyriath25-Mar-20 03:27 PM
My experience with computers is more on the user side - I admit, pondering architectures is something I haven't yet had to do, especially since Avishraa hasn't gotten anywhere near that tech level yet.
They're still smacking each other around with swords and axes and spears.
Vāstra II25-Mar-20 03:31 PM
You're still welcome to offer input
e.g. what would you like to see out of this computer as a user?
Shyriath25-Mar-20 09:12 PM
...for some reason, I find myself wondering about the telepathic interface. Is it the equivalent of a desktop? Does it have something that would be the equivalent of "wallpaper" - something like a background with a particular pattern of emotion/thought-pattern, chosen for aesthetics, against which the representations of the programs stand out?
Or would safir computers be bright enough to be more interactive - something you can communicate with? You find a program by asking it "what do you have that does X?" maybe?
Vāstra II26-Mar-20 04:49 PM
@Shyriath There's definitely a wallpaper-equivalent in the form of the computer's “surface thoughts”, as you described, though I'm not really jumping toward the desktop metaphor; the latter also exists, in the form, on most systems, of metadata attached to programs.(edited)
Shyriath26-Mar-20 07:52 PM
If there's not a desktop, presumably there's instead a "most recent" or "most frequent" option to bring up the software that's most likely to be needed.
Step up your game with a modern voice & text chat app. Crystal clear voice, multiple server and channel support, mobile apps, and more. Get your free server now!
Shyriath26-Mar-20 07:55 PM
You could, in fact, imagine more general software searches bearing more of a resemblance to modern music-player functions. Don't know exactly what you want? You could "shuffle" through a particular metadata tag.
Vāstra II26-Mar-20 07:56 PM
Definitely, and doubly so when searching through a software repository
Shyriath26-Mar-20 08:00 PM
...it just occurred to me that virtual reality applications, too, would probably take on a whole new level of realism if there's a telepathic connection possible. No headset, just project sensory illusions directly into the brain.
Vāstra II26-Mar-20 08:00 PM
Shyriath26-Mar-20 08:01 PM
Aside from the amazing gaming experiences possible, that would probably play a huge role in learning and training.
Vāstra II26-Mar-20 08:01 PM
I think I've talked on this server before about an idea I had called the lakuna, or dreamworld: basically a literal MUSH shared by all the sleeping safir in/on a given country or planet, which often takes a form similar to a MMORPG
Shyriath26-Mar-20 08:03 PM
Vāstra II26-Mar-20 08:04 PM
Vāstra II27-Mar-20 04:53 PM
FutureFractal27-Mar-20 09:16 PM
ah sorry, been distracted with stuff :o
lemme take care of some stuff and then we can talk architectures :3
Vāstra II27-Mar-20 10:50 PM
FutureFractal27-Mar-20 10:58 PM
:3 back now
I've come up with a few con-computer architectures
one of them I came up with in high school, which is basically the architeture for an immensely powerful alien artifact which is a reality altering supercomputer called the Ω Capsule from a work-in-progress scenario of mine(edited)
it can run parallel threads of execution on different timelines and then reconvene them afterward
and it can programmatically read and write to the positions and attributes of subatomic particles within a huge radius around itself(edited)
Vāstra II27-Mar-20 11:11 PM
Vāstra II27-Mar-20 11:20 PM
Meanwhile, I'm reading up on computer architecture models other than the most common one, von Neumann
(also watching LGR lol)
Vāstra II27-Mar-20 11:41 PM
@FutureFractal maybe safir computers focus on parallelism over raw speed, reflecting a design philosophy that on Earth is usually only seen in GPUs and other specialized applications?
so stuff like cellular architectures, SIMD/SIMT, etc.
Vāstra II27-Mar-20 11:49 PM
Neuromorphic computing also seems interesting
as does fabric computing
Vāstra II28-Mar-20 12:28 AM
@FutureFractal I'm also looking at minimal instruction set computers and dataflow architectures
The transputer is a series of pioneering microprocessors from the 1980s, featuring integrated memory and serial communication links, intended for parallel computing. They were designed and produced by Inmos, a semiconductor company based in Bristol, United Kingdom.For some tim...
What do you think?
FutureFractal28-Mar-20 01:11 AM
agreed, parallelism sounds super cool :3
I wish VLIW and EPIC architectures managed to take off, those sounded really cool
Vāstra II28-Mar-20 01:12 AM
Still trying to figure out a good basic instruction set, meanwhile
FutureFractal28-Mar-20 02:03 AM
I've been meaning to learn about GPU architectures for awhile, that'd probably be pretty relevant here :o
Vāstra II28-Mar-20 02:05 AM
Sorry for the radio silence, I've been trying to chase down a deleted reddit comment so I could attribute a piece of art I requested and am using in a project
FutureFractal28-Mar-20 02:06 AM
ah it's all cool! I'm sorry for my radio silence too
Vāstra II28-Mar-20 02:07 AM
Vāstra II28-Mar-20 02:48 AM
Here have a quick and dirty chart showing various word lengths
Vāstra II28-Mar-20 02:56 AM
and one showing the number of bits/trits necessary to hold various powers of five different number bases in use in the Alliance
I think a byte should be standardized as six trits, with the most common word length being 36 trits; that said, however, I kinda want to go with 12 or 18 trits for the architecture I have in my head, even if it'd make more sense to design a modern than an old one? but at the same time, the weight of history etc.? @Chip Dipper pls advise
(ideally while I sleep. g'night y'all )
Vāstra II28-Mar-20 06:06 PM
@Chip Dipper evening
Vāstra II30-Mar-20 02:51 AM
@Chip Dipper hi
FutureFractal30-Mar-20 04:17 AM
sorry about that
Vāstra II30-Mar-20 04:20 AM
\pokes his above ramblings**
Vāstra II30-Mar-20 04:50 AM
Vāstra II30-Mar-20 05:10 AM
FutureFractal30-Mar-20 05:10 AM
that byte idea sounds nice to me! :o
Vāstra II30-Mar-20 05:10 AM
FutureFractal30-Mar-20 05:11 AM
sorry, I'm rly absorbed in writing my Gen 1 ROM dumping library rn :3
Vāstra II30-Mar-20 05:11 AM
@FutureFractal make sure to leave or, ideally, infodump any ideas/etc, no matter how ill-formed, for alien computers in the very near future
(namely because I'm going to sleep. g'night )
FutureFractal30-Mar-20 05:15 AM
sure thing! c:
Vāstra II30-Mar-20 03:50 PM
Vāstra II31-Mar-20 01:37 AM
FutureFractal31-Mar-20 01:38 AM
why the sad?
Vāstra II31-Mar-20 01:42 AM
Because I was looking forward to your infodump :o
FutureFractal31-Mar-20 01:42 AM
ah sorry D:
I've been asleep for most of today >_<
Vāstra II31-Mar-20 01:56 AM
FutureFractal31-Mar-20 02:01 AM
sorry if I seem distant, I'm currently extremely absorbed in this gen 1 glitch datamining project of mine
like, to a degree of focus I haven't really had in my personal projects for awhile :o
Vāstra II31-Mar-20 02:03 AM
I wish I could have that degree of focus on anything
FutureFractal31-Mar-20 02:04 AM
same, I wish I could have this degree of focus more often
just figured out how glitch cities work by analyzing the map loading code :3
glitch cities are caused by a destination warp ID that doesn't exist in the map you're traveling to, so they index the map header's destination warp table out of bounds and get a garbage map pointer
I knew that already but what I just figured out is that apparently the pointers are precomputed in the table
so garbage pointers out of bounds can pretty much be anything
Vāstra II31-Mar-20 02:07 AM
FutureFractal31-Mar-20 02:08 AM
also the reason Cut doesn't work in most glitch cities is because that pointer lands in the ROM for them instead of the map buffer in RAM that normal maps use
and Cut tries to modify the block data to remove the tree, but that doesn't work if the block data is in ROM
so armed with this knowledge, now I should be able to dump the block data for any glitch city :3
and all I need to do after that is to write code to load the right overworld tileset from the game and render the block data graphically
Vāstra II31-Mar-20 02:11 AM
FutureFractal31-Mar-20 02:15 AM
also thankfully overworld tilesets aren't compressed like Pokémon and trainer sprites are
Vāstra II31-Mar-20 02:27 AM
Meanwhile I'm still bashing my head against the instruction set lol
FutureFractal31-Mar-20 02:42 AM
ooh, for the trinary CPU? :o
Vāstra II31-Mar-20 02:42 AM
FutureFractal31-Mar-20 02:43 AM
hmm, I wonder how negatives would work in base-3
is there a two's complement equivalent?
Vāstra II31-Mar-20 02:44 AM
In balanced ternary, the leading nonzero digit has the same sign as the number itself
FutureFractal31-Mar-20 02:48 AM
Vāstra II22-Apr-20 12:25 PM
@Chip Dipper Summary of above ramblings:
- Optical computer
- Uses balanced ternary, with -1 and +1 being opposing linear polarizations
- Six-trit trytes, 36-trit words
- x86-style memory addressing
- Telepathic input/output, with “wallpaper” in the form of a package of surface thoughts
- Maybe control panel à la old mainframes/minicomputers? Definitely a keyboard, albeit big because safir have four arms?
- Parallelism over raw speed — cellular architectures, SIMD/SIMT, etc.
- VLIW? EPIC? Neuromorphic computing? Fabric same? MISC? Dataflow architecture? Transputer?
Hmm, I think safir might use base 9 or 27 for working with computers the same way we use base 8 or 16, but I can't decide on collation order for trits; that is, whether 0₉ should be -- and 0₂₇ --- or 00 & 000 respectively
@Vāstra II I didn't really get too far in designing it, aside from deciding that there will be both 16- and 32-bit opcodes(edited)
designing a RISC instruction set is a bit trickier than designing a bytecode instruction set, so I kinda got sidetracked by other stuff
Vāstra II27-Jan-21 07:31 AM
Though I'm not fully certain of the difference myself(edited)
FutureFractal27-Jan-21 07:32 AM
basically RISC opcodes are constant-width, as opposed to CISC/bytecode which are generally one-byte opcodes with optional operand bytes
so you need to pack both the opcode bits and the operand bits into the opcode, and often different instructions use different amounts of bits for each
as an example, here are ARM's THUMB opcodes
Vāstra II27-Jan-21 07:33 AM
FutureFractal27-Jan-21 07:34 AM
the fixed bits on the left side specify the opcode, and the named chunks on the right are the operands
I had an idea to basically get the best of both worlds vis-a-vis ARM and THUMB where I'd have a singular instruction set where some instructions are 16-bit and some are 32-bit(edited)
IIRC 64-bit ARM's THUMB is like this too(edited)
Vāstra II27-Jan-21 07:36 AM
Sorry I can't be more attentive; I'm playing mahjong right now
FutureFractal27-Jan-21 07:36 AM
ah it's all cool c:
incidentally, since I've been thinking about instruction sets a lot over the past half decade or so, I've realized that RISClike instruction sets are really good for executing on hardware but less optimal for interpreting in software, whereas bytecode/CISClike instruction sets are much more suited for interpreting in software
same goes for register vs stack machines: register architectures work really well for hardware execution but are suboptimal for software interpreters, while stack architectures work really well for software interpreters
Vāstra II27-Jan-21 07:42 AM
I haven't thought about whether this architecture would be register, stack, or something else entirely if possible, because lolaliens
FutureFractal27-Jan-21 07:42 AM
KaliFyre27-Jan-21 04:06 PM
How about FILO stack registers
Registers that also work as First-In-Last-Out stacks?
Something something psionic memory?
K: granted I've just woken up so mind is still semi disassociated
Vāstra II28-Jan-21 05:56 AM
I think FILO stacks would be synonymous with LIFO ones
In logic, a three-valued logic (also trinary logic, trivalent, ternary, or trilean, sometimes abbreviated 3VL) is any of several many-valued logic systems in which there are three truth values indicating true, false and some indeterminate third value. This is contrasted with the more commonly known bivalent logics (such as classical sentential o...
i'm happy to announce the release of my latest project, the Graphics Gremlin! it is an 8-bit retro video card for ISA bus computers, and it emulates the MDA and CGA graphics cards. scan conversion is built in so it works with modern LCD monitors!