The Powder Toy

R3A1016 by LBPHacker

LBPHacker
LBPHacker
125 / 0
1st May
3rd May
I made a computer again. It can do a lot of stuff the previous one couldn't and can do most stuff that I'd ever wanted a TPT computer to do. It definitely can churn out Fibonacci numbers and primes. Check the relevant forum thread too.
subframe processor technology mandelbrot dayofweek fibonacci prime r316 computer magic

Comments

  • LBPHacker
    LBPHacker     8th May
    This is off FP now, nice run, thank you all.
  • LBPHacker
    LBPHacker     8th May
    The one exception is fetching an instruction freshly written by the previous instruction, which the manual documents as yielding the value before the write. This is not due to the nature of subframe or multi-instruction-per-frame computing in TPT, this is just because this specific computer was easier to build that way.
  • LBPHacker
    LBPHacker     8th May
    This is not in fact in the manual because it's not a problem in general. Subframe enforces a total order on the update code of all particles, and you can (and this computer does) transmute that into total order on memory accesses.
  • kerbal2346
    kerbal2346     8th May
    Sorry if this question could be answered by rtfm-ing, but I'm curious how data hazzards are handled in comparison to a "real" architecture. I'd have to imagine writing to and reading from one address on the same frame is difficult.
  • deathhawk
    deathhawk     7th May
    @LBPHacker will there ever be audio added to this game?
  • LBPHacker
    LBPHacker     7th May
    The ALU is 16b but I rolled my own bignum primitive for fibonacci. It uses base 10000 digits to make printing easier. Fibonacci happens to be one of the demos that I've already documented, see https://github.com/LBPHacker/R316/blob/v2/examples/demo.asm
  • penguinrules
    penguinrules     7th May
    What's the integer limit on this computer? Seems like fibonacci can run indefinitely with sufficient ram?
  • deathhawk
    deathhawk     7th May
    this is one of the fastest computers in the powder soy
  • LukeTheAwesomePro
    LukeTheAwesomePro     6th May
    @RandomGoober And from my experience, writing and maintaining assembly code is significantly harder and less productive than working with higher-level languages like C/C++. Nearly all modern software development relies on higher-level languages because they are more accessible, maintainable, and portable.
  • LukeTheAwesomePro
    LukeTheAwesomePro     6th May
    @RandomGoober Using a transpiler to convert between different assembly languages (such as MIPS or ARM32 to another architecture) is generally not a practical approach though. Programming directly in modern ASM is notoriously difficult and time-consuming.