Issue #16487 has been updated by byroot (Jean Boussier).


> My concern here is who can read, understand and maintain that code?
> Such complexity is likely to have bugs or need tweaks over time.

It's a totally understandable concern. A few responses to that:

  - In that specific case, the UTF8 spec probably won't change before a whi=
le, so that limits the amount of maintenance required
  - In a more general case, if for some reason the SIMD version of a functi=
on is too hard to maintain, it can simply be removed.

> It's unfortunate the C compiler cannot do this on its own.

As @ahorek showed, they can do it to some extent. I doubt they can auto-vec=
torize `coderange_scan`, but maybe they can do it to `search_nonascii`, if =
so it could yield a decent speedup with very little code to maintain. It's =
likely worth a try.

----------------------------------------
Misc #16487: Potential for SIMD usage in ruby-core
https://bugs.ruby-lang.org/issues/16487#change-83758

* Author: byroot (Jean Boussier)
* Status: Open
* Priority: Normal
* Assignee: =

----------------------------------------
### Context

There are several ruby core methods that could be optimized with the use of=
 SIMD instructions.

I experimented a bit on `coderange_scan` https://github.com/Shopify/ruby/pu=
ll/2, and Pavel Rosick=FD experimented on `String#strip` https://github.com=
/ruby/ruby/pull/2815.

### Problem

The downside of SIMD instructions is that they are not universally availabl=
e.
So it means maintaining several versions of the same code, and switching th=
em either statically or dynamically.

And since most Ruby users use precompiled binaries from repositories and su=
ch, it would need to be dynamic if we want most users to benefit from it.

So it's not exactly "free speed", as it means a complexified codebase.

### Question

So the question is to know wether ruby-core is open to patches using SIMD i=
nstructions ? And if so under which conditions.

cc @shyouhei





-- =

https://bugs.ruby-lang.org/

Unsubscribe: <mailto:ruby-core-request / ruby-lang.org?subject=3Dunsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>