This is the mail archive of the cygwin mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Parallelization


On 7/18/2016 1:38 AM, Tony Kelman wrote:
Eliot Moss <moss <at> cs.umass.edu> writes:

True ... it also made me think of Python, which is designed to use
parallelized numpy (etc.) libraries, optimized for your platform.
Can use all the hardware threads on your machine, as well as make
good use of vector extensions such as AVX.  A 64-bit (x86-64)
version will give best use of vector processing, in my
experience.

Regards -- Eliot Moss

numpy is only as parallel as the underlying BLAS/LAPACK library that
it uses is. So if you're using Cygwin's openblas then you're in
decent shape. But I don't think cv_adams spends much time (if any?)
in BLAS/LAPACK dense linear algebra functions, I think it's mostly
dominated by function evaluation time.

Ok -- I'm not sure what to suggest then ...

Regards -- EM

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]