Re: [boost] Accelerating algorithms with SIMD - Segmented iterators and alternatives

Subject: Re: [boost] Accelerating algorithms with SIMD - Segmented iterators and alternatives
From: joel falcou (joel.falcou_at_[hidden])
Date: 2010-10-12 14:35:36

On 12/10/10 20:21, DE wrote:
> on 12.10.2010 at 21:38
>   joel falcou wrote :
>    
>> When no SSEx stuff are enabled, pack<T,N>  falls back to boost::array
>> emulation.
>> If you wan to ship code for various SSex flavor, just make various
>> compialtion fo your kernel code
>> then dynamically laod the correct one. There is no way you can check at
>> runtime before calling EVERY
>> function, this is far too costly.
>>      
> an idea immediately arises in my mind: what if it could be possible to
> force a particular backend e.g. sse2 and make it propagate down the
> call stack? this way you write generic code once and are able to
> (pre)compile for (expected) platforms/technologies
> then you just switch once depending on some circumstances and call the
> appropriate instance (say sse2 forced)
> at least this way you write the source once and possibly don't rewrite
> it for every circumstance
>
>    
It's like what we do already