ATLAS 3.8.1

[Clint Whaley]

ATLAS 3.8.1 has been released.  It fixes all the known bugs in 3.8.0.  There
are also several other improvements, but probably the most important is that
I have finally fixed the Level 1 timing, so that it doesn't die so much with
"unable to get timings within tolerance" when you don't have architectural
defaults.

The ChangeLog entry is below.

Cheers,
Clint
ATLAS 3.8.1 released 02/21/08, Changes from 3.8.0:
   * Fixed bug in slvtst that counted complex flops same as real
   * Fixed bug causing wrong answer for row-major gemm C=A*A' or A'A
   * Fixed bug in configure causing Pentium-M to be IDed as CoreDuo
   * Fixed bug in tfc.c causing memory overwrite when too many samples taken
   * Improved L1 BLAS timers so they work like the rest of the package, and
     thus don't die all the time on tolerance failures
   * Improved ATLAS/tune/blas/gemm/mmsearch.c:
     - for x86, tried more registers, since smart compiler can reduce A & B
       regs to 2 (and possibly even 1)
     - Made it so search tries both load-C-at-top and load-C-at-bottom of
       M loop.  Bottom is superior for error, and ATLAS originally defaulted
       to load-C-at-top.
   * Added configure support for new K10h platform from AMD, as well as
     basic architectural defaults (no new kernels, just good search)

**************************************************************************
** R. Clint Whaley, PhD ** Assist Prof, UTSA ** www.cs.utsa.edu/~whaley **
**************************************************************************

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Math-atlas-devel mailing list
Math-atlas-devel@...
https://lists.sourceforge.net/lists/listinfo/math-atlas-devel