Jason Gross
Jason Gross
It would be nice to be able to verify that our Go code works. I'm not sure where to find Go tests, though. cc @andres-erbsen
> Optional/follow-up/whatever: if you're outputting a file that's meant to be #included somewhere (given the symbols are `static` anyway), there may be something to be said for the files being...
With `ulimit -S -s 1000000`, it completed in just over an hour: [fiat_32.txt](https://github.com/mit-plv/fiat-crypto/files/4834869/fiat_32.txt) Perhaps we should document in the README how to increase the stack size? _Originally posted by @JasonGross...
We should see if it's really slower or not. While we're at it, maybe we should add a CI test that drops the newest version of our code into BoringSSL...
Should we check on the CI that gcc and clang generate constant-time assembly code for our C somehow?
cc @andres-erbsen @davidben
There are a lot of Z.combine_at_bitwidth operations left over. Not sure why, I'll try to dig into this.
> > Thanks for setting this up for me. Our pipeline has come a long way since we first pushed stuff through it, I had an actually decent experience working...
Should we support karatsuba again? IIRC, the main issue was that we couldn't recognize that `x + y - x` had the same bounds as `y`. However, we should be...
Currently the analysis of what primitives we needs only looks at unsigned numbers. In implementing #679, we might want (for Java) the ability to emit, e.g., `cmovznz_64` rather than just...
This is definitely not ready for merge (though I believe it does build) I've left a bunch of `TODO`s for @davidben (or possible for me-in-the-future) to fill in the actual...