Re: [multiprecision] Getting the best performance on windows for uint128_t etc

From: degski (degski_at_[hidden])
Date: 2019-07-12 11:01:39

On Fri, 12 Jul 2019 at 13:21, John Maddock via Boost-users <
boost-users_at_[hidden]> wrote:

> That's only possible with assembly level add/multiply/divide.
>

Right.

Ah OK, just looked at MPIR on win64 with the mpir_gc backend, and they
> do use 64-bit limbs, but rely on GNU's longlong.h for
> add/subtract/multiply which appears not to use inline assembly for msvc,
> but instead breaks the inputs into high/low parts and then does
> schoolbook multplication.

The mpir_gc ([g]eneric [c]) backend will by definition not use assembly,
one needs to build something a bit closer to one's hardware (run the
mpir_config.py script [in the msvc sub-folder] to configure, generate
solution files and build) to get any performance. It requires yasm as
assembler [installed and properly integrated into VS20XX]. Altogether it's
neither obvious nor trivial to get something [a binary] that works well.

degski
-- 
@realdegski
https://edition.cnn.com/interactive/2019/06/middleeast/saudi-teen-death-penalty-intl/
"Anyone who believes that exponential growth can go on forever in a finite
world is either a madman or an economist" - Kenneth E. Boulding