Re: [boost] Boost.Pipeline -- scheduling of segments

Subject: Re: [boost] Boost.Pipeline -- scheduling of segments
From: Hartmut Kaiser (hartmut.kaiser_at_[hidden])
Date: 2014-07-10 08:29:51

> > The overhead of one such thread is in the range of 700-900ns
> 
> Do you have numbers for specific architectures (ABIs)?

These numbers are for Sandybridge x64, depending on the number of cores used
concurrently (e.g. the difference is caused by contention effects in the
scheduler). We have numbers for XeonPhi, which are worse, but that's because
this architecture runs on a slower clock rate, and other architecture
specific things. We have it running on ARM32 and BlueGene/Q (Power) as well,
but I don't have any numbers available at this point

Regards Hartmut
---------------
http://boost-spirit.com
http://stellar.cct.lsu.edu