On Fri, 20 Feb 2015, Xujun Zhao wrote:
I currently only have opt mode, the errors are as follows.
Error messages in opt mode are nearly useless. Both libMesh internal
checks and libMesh checks of application usage are nearly all disabled
for speed, so flaws that could get caught by a nice assertion instead
go unchecked until there's a segfault.
How large a problem can you reproduce the failure on? If it's too
large for dbg mode to run in reasonable time, devel mode is a good
intermediate choice - it leaves most libMesh assertions on, but
disables the very-slow libstdc++ assertions.
It can pass those lines, but stopped and output error at equation_systems.init
();,
which is the next line of the above codes.
If I use serial mesh: SerialMesh mesh(init.comm()); Then there is no problem.
Your first email left me with the impression that serial runs were
working but parallel runs were failing. This email gives me the
impression that parallel runs with SerialMesh are working but parallel
runs with ParallelMesh are failing. Is my second interpretation correct?
PeriodicBoundaries are also supported and tested with ParallelMesh,
but it's much trickier code, and it's possible that there's a buggy
corner case or a recent regression we haven't caught. If you could
help us track down the problem (ideally with a small test case, if the
dbg-mode error information isn't enlightening enough) I'd appreciate
it.
Thanks,
---
Roy
------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users
