[petsc-dev] Experimental GNU make build system

[Jed Brown]

Satish Balay <balay at mcs.anl.gov> writes:

> On Wed, 5 Jun 2013, Satish Balay wrote:
>> Also - I removed all the compiler/tool autodetection code in
>> win32fe. [it wasn't working well - and was disabled by default for the
>> past few releases. And we've been insisting on users setting up
>> working compiler enviornment via 'compiler cmd' anyway]

Cool, that's a lot of code.

> All changes pushed to https://bitbucket.org/petsc/win32fe
>
> BTW: The time to build debug/c++/complex/sequential/64bit version of
> petsc on a 8core [16 HT] windows machine with cl/ifort [via win32fe]
> using gmakefile is:
>
>>>>>>>>
> balay at msnehalem2 ~/petsc.clone
> $ time make -f gmakefile PETSC_ARCH=arch-uni-cxx-cmplx -j 20 > gmake.log
>
> real    1m23.668s
> user    0m30.656s
> sys     1m14.297s

Not bad; might even be within a factor of two of a Linux machine.


Is the gmake build considered working now, once we add --download-gmake
and proper detection for -MMD -MP?