Any thought about trying Go 1.26 SIMD support here?

[0xsimulacra]

Hi,

Thanks for this package, I use it a lot.

I also use it as base for an "int256"-ish implementation here: https://github.com/0xsimulacra/int256

I saw that Go 1.26 now has experimental SIMD support on AMD64, and I was wondering if there had been any thought about trying to use that here. https://pkg.go.dev/simd/archsimd

I don’t know SIMD very well myself, so maybe for uint256 stuff it does not help much, or maybe it is not worth the added complexity, I’m not sure.

I just wanted to ask if this is something you already looked at, and whether you think there is any possible gain in some hot paths ?

I’d be happy to help experiment with it if you think it makes sense, but I would need guidance since this is not really my area.

Thanks again.

[holiman]

Someone did try to make use of simd instructions here: #199 I can't say I have looked into it. Theoretically it seems to me that it should be faster, so definitely an area where it makes sense to investigate, imo. You are welcome to make another stab at it!

…

-- Sent from /e/OS Mail.

[0xsimulacra]

Thank you very mush for the quick feedback.

I'll look into running my own experiments on the first occasion I have.

I will keep you updated of the results when there is any.