Re: [Math-atlas-devel] ATLAS/C++

[R C. W. <rw...@cs...>]

Maarten,

>If you include a solution with ATLAS, I suggest the first one. A
>different header file for C++ would confuse some people I think.

Thanks for the info.  What you say makes a lot of sense . . .

>Something else. I'm new on this mailing list, so I don't know exactly
>what I can post here. I compiled ATLAS on HP-UX 11 using HP aCC and HP
>f90. ATLAS is working fine now (solving dense complex matrices (ZGESV)
>works about 8% faster than with the HP tuned BLAS/LAPACK), but it took
>some time before I got all compiler and linker flags right. I also
>needed to change one source file a bit (removed a static identifier). Is
>there someone I should mail these changes and flags to? Do you know who?

I'm the guy, I guess.  I'm pleased to hear ATLAS doing well, and impressed
you managed to get the install through.  On the last supported HP platform,
I remember spending quite some time getting a good install . . .

I haven't had access to an HP machine in quite some time, so ATLAS presently
has no arch defaults or decent config support for the system.  If you want
to turn in what you've done, you might also consider providing your
architectural defaults, as described in ATLAS/doc/atlas_devel.ps . . .

Thanks,
Clint