Misha Komarov

Hey @wanghs09! It is feasible, you're right. This we had in our roadmap regarding the verification part implementation. At first we need to introduce all the algorithmic/proof system-level tricks we...

> The rows in the tables are locked incorrectly. So hardfix for that is as such: in doxygen.css comment the rule > > ``` > .directory tr.even { > padding-left:...

Hm. Good question. I was thinking of some kind of architecture-specific function selection macro for Boost.Multiprecision, Boost.Math and for one more library coming. I think lots of architecture-specific places (e.g....

No, I actually meant commenting out two lines commented as "black magic" (https://github.com/wbhart/mpir/blob/55fe6a9f52ca532a611a89f67186ed915bbf1123/mpn/x86_64/skylake/add_n.as#L56 and https://github.com/wbhart/mpir/blob/55fe6a9f52ca532a611a89f67186ed915bbf1123/mpn/x86_64/skylake/sub_n.as#L56), because they persist in non-AVX assemblies. All the AVX-enabled assemblies are inside the avx directory,...

I got these problems as well. Thinking of nop-ing same amount of bits. > On 13 Jun 2020, at 08:39, wjblanke wrote: > > > We had a G4400 user...

A fix worked partially, but unfortunately, still not fixed.