Re: [boost] SIMD libraries

Subject: Re: [boost] SIMD libraries
From: joel falcou (joel.falcou_at_[hidden])
Date: 2010-06-14 13:32:34

Simonson, Lucanus J wrote:
> There are people who make their livelyhood getting that 10-20% and look down on us who accept less than 100.  For my part I'd rather write 10X more code that runs at 90% maximum-theoretic so I'm still on board.
>   
As I said in my boost'con presentation, we target a different 
demographic: people who don't care about the details and just want to go 
"faster".
See slides 2->4
> Interesting.  That's a good use of proto
>   
We thought so ;)
> I opt for 1 and 2.  I read the EVE publication, which seems to describe only a optimized valarray type interface.  Do you have a reference for the generic vector register interface you mentioned?
>   
EVE is pretty old, it predates proto by 2-3 years. Back in the day it 
was all made by hand, perusing Veldhuizen seminal work.
I don't think we have any publciations on the new versions except for 
the Boost'Con slides. Time to make one I think even if I
see it harder to get published as it seems it's not interesting for a 
lot of people (read reviewers).
> I made some minor contributions to writing a DSEL for wrapping intrinsics.  My main design input was to push for making values be models of our expression concept so that everything composes well.  Since we weren't using proto we had to figure out a lot of things for ourselves that proto would have done for us.
>   
Proto changed our life basically from a mess of hand written ET mess to 
a clean stuff especially on the specialization front.