mysterymath

7 ms·

31.
by mysterymath 4y ago
I ran the C version of this benchmark using llvm-mos's Clang for the 6502. The results: 21.4 seconds 5793 bytes Which is middle of thepack for the Z80 benchmarks, but well below the 6502 ones. We're also using a slightly tweaked
32.
LLVM-MOS 6502 Toolchain Initial Release(github.com)
13 points by mysterymath 5y ago|0 comments
33.
by mysterymath 5y ago
Consider the input as a big space of points, where each point is a possible input. These points aren't unrelated; speaking loosely, foo(3.000000000001) is usually more likely to behave like foo(3) than foo(4) is. The different behavior
34.
by mysterymath 5y ago
A Z80 backend could probably reuse our static stack allocation, but that's about it. It's definitely useful, but the time-consuming part of writing LLVM-MOS's code generator has ended up being... writing LLVM-MOS's code
35.
by mysterymath 5y ago
Some of it might benefit AVR, which being 8-bit, shares some of the same problem space. But most of the changes we've made so far are of the kind where LLVM says "this doesn't happen", or "when it does, it's no
36.
by mysterymath 5y ago
Most of the inoptimality in the article isn't due to the issues you've raised, but rather due to us just starting to optimize LLVM-MOS. First, I have utterly no idea why there are so many calls to memset; it looks like it's u
37.
by mysterymath 5y ago
Up until just a few weeks ago, 100% of the codegen work we've put into LLVM-MOS has been to get it feature-complete and rock-solid. It's awesome to see that that work has paid off! We're just now starting to really optimize t
38.
by mysterymath 6y ago
Because the objects are probably either relatively deeply encoded inside hundreds of plain-C stack and heap locations, or they're not even fully resident in RAM anymore by the time output occurs. It's not that uncommon for ol&#x27
39.
by mysterymath 6y ago
Shouldn't be too hard to add later, since they're mostly just swizzles onto the existing instructions, and LLVM already has good support for registering slightly different versions of a target (subtargeting). For now, I'm tar
40.
by mysterymath 6y ago
I'm porting LLVM/Clang to the 6502. Notoriously difficult project, attempted dozens of times by as many folks. https://github.com/mysterymath/clang6502 My take generates pretty darn good assembly for the case
41.
Hacker’s Manifesto (1986)(phrack.org)
13 points by mysterymath 7y ago|1 comments
42.
A Simple CPU(mysterymath.github.io)
3 points by mysterymath 7y ago|0 comments