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


Just to confirm, `__attribute__(target(""))` is how PHP does it: =


  - https://github.com/php/php-src/blob/7ce531f2c28dcfe4aeed271b55b82de65c3=
bca8a/ext/standard/base64.c#L370-L371
  - https://github.com/php/php-src/blob/7ce531f2c28dcfe4aeed271b55b82de65c3=
bca8a/Zend/zend_portability.h#L592

In the end there's indeed quite a bit of ceremonial code with a significati=
ve amount of extra `#ifdef`, but IMHO it's tolerable, and could probably be=
 reduced with some macros.

> I'm more worried about code bloat, having many functions (c, sse2, avx2 v=
ersion etc...) to solve the same problem isn't DRY.

It would probably be limited to a few hotpaths though. I might be wrong, bu=
t I don't see it used beside a select few string functions.

----------------------------------------
Misc #16487: Potential for SIMD usage in ruby-core
https://bugs.ruby-lang.org/issues/16487#change-83760

* 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>