Naoki Shibata
Naoki Shibata
>Computers with avx512 (like the ones I've been benchmarking on) have 32 registers. They can hold quite a bit of numbers in their registers across loop iterations. But even with...
@fpetrogalli How do you think?
@colesbury Could you give me an e-mail?
Actually I haven't started working on this. It may require a dramatic change in the structure of the source code. So, I have no idea at this point.
I think it is easier to replace those generators with python scripts. Is there any reason that you are trying to do everything with cmake?
Please go on. Make an issue first.
What @fpetrogalli-arm means is that we should discuss detailed design of your change first. Python vs. cmake is one thing. We need to know how cmakelists will look like after...
@fpetrogalli How about this patch? I think this patch is safe to merge.
The generic vector ABI is only for generic use, and there are additional requirements for vectorized math functions. Since we are working on implemenenting vectorized math library, I think it...
> It is also not only that the won't work on the wrong machine, but that callers must uphold a particular ABI: that vectors like __m128d and __m256d are passed...