[MINC-development] -O3 breaks N3

[Andrew Janke]

Hi Vlad,

I have known of this behaviour in N3 for some time, there are (or
were!) options set in the EBTKS and N3 autoconf make packages to take
care of this.

When I last looked into this (a long time back) it was to do with loop
unrolling, specifically the Greg Ward? initiated habits of using
for_less() style loops that resulted in a lot of index pre-crementing.

I did find at the time that rewriting (or removing) the for_less style
macros fixed the issues. At the time the decision was made not to
include these fixes as it made the code quite a bit slower and thus
the reason for the for_less() style macros in the first place.

Note that I have also seen this same behaviour in minctracc.


a

On 14 April 2012 01:49, Vladimir S. FONOV <vladimir.fonov at gmail.com> wrote:
> looks like compiling EBTKS and/or N3 with -O3 flag (default for release
> build in CMake) breaks nu_correct completely (i.e it doesn't converge).
> Any ideas why it happens and how to fix it, apart from the obvious
> suggestion of using -O2 flag of course.