bug-gsl
[Top][All Lists]
Advanced

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

Re: [Bug-gsl] make check fails on AIX 5.2.0 in 64 bit mode


From: Olaf Lenz
Subject: Re: [Bug-gsl] make check fails on AIX 5.2.0 in 64 bit mode
Date: Mon, 30 Aug 2004 17:26:18 +0200
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040114

Hello!

Brian Gough wrote:
 > However, "make check" seems to fail in the "rng" subdir, now. I've
 > attached the output to this mail. It seem that the ranlx-generators
 > do work differently on a 64 bit platform?

Would it be possible for you to look at this in the debugger with a
simple test program?.... a quick inspection of these files doesn't
show me anything obvious that would cause a problem on 64bit
platforms, so I am not sure why it is failing.

I've played around with different compiler switches, and apparently I've found the problem there. In higher optimization levels, the AIX compiler assumes "-qhot", i.e. "Higher Order Transformations". From the docs:
        
"Instructs the compiler to perform high-order transformations on loops and array language during optimization, and to to pad array dimensions and data objects to avoid cache misses.
[clip]
Both -qhot=arraypad and -qhot=arraypad=n are unsafe options; they do not perform any checking for reshaping or equivalences that may cause the code to break if padding takes place."

I suppose this may be the case here (even though I haven't looked into the code). I think I'm going to dig deeper into the problem later, but not now. I'll keep you informed.

To see if there are going to be other failures later you might want to
try "make -k check" also

Wow, that was the option I was looking for...

Attachment: signature.asc
Description: OpenPGP digital signature


reply via email to

[Prev in Thread] Current Thread [Next in Thread]