I don't really speak for anyone on this but me (I'm even a core dev so it's perfectly fine to just ignore me :) ). Originally I was just asking a question about the downfalls in putting SIMD in an external library instead of trying to put work it into the compiler's optimizer. A similar issue is found in concurrency in where there seems to be no way to do it 'right'.
Someone must of just figured I had a strong argument for it... I really don't, but if the language can provide type safety for assembly functions I figure it would be more flexible than having fixed built-in Auto-vectorization. On Tue, Aug 17, 2010 at 3:36 PM, Jonathan S. Shapiro <[email protected]>wrote: > Tyler certainly doesn't speak for me on this, but the L4 guys aren't doing > much SIMD support in their OS, so I'm not sure what the point of this > exchange is, exactly. >
_______________________________________________ bitc-dev mailing list [email protected] http://www.coyotos.org/mailman/listinfo/bitc-dev
