[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ATLAS



Hi Clint!

1) Thanks for the developer's release.
2) The sse gemv/ger work great.  I noticed you included only the cases
   that compiled best on your hardware.  Is it plausible that some of
   the other unrolling options would be better on different
   incarnations of the PIII, and that the timer should try them all
   out when building the library?
3) prefetcht0 -> prefetchnta = +20% !  I'll be forwarding you some new
   headers soon.
4) The complex case is about done, and looks very good, as you
   expected.  I'm having trouble tuning these as the timer results
   jump around *a lot*, even when I use -DWALL on time.c
5) Next step on dgemv is to try to unravel your _mm.c and add
   prefetch.  

Take care,

R Clint Whaley <rwhaley@cs.utk.edu> writes:

> Hi,
> 
> Just thought you might want to know I have just posted detailed instructions
> on how to make DF (Digital Visual Fortran) and CL (MSVC++) linkable libraries
> using ATLAS.  It's in the errata file,
>    http://www.cs.utk.edu/~rwhaley/ATLAS/errata.html#Wincclib
> 
> Cheers,
> Clint
> 
> 

-- 
Camm Maguire			     			camm@enhanced.com
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah