Re: [Libmesh-devel] [Libmesh-users] about conjugate heat transfer problem

[Benjamin K. <ben...@na...>]

> Specifically, look for the statement "global_ids.clear();" in
> mesh_communication_global_indices.C.  The for loop immediately
> following is where all the CPU time is being sucked up.  But I don't
> see what the problem is.  Nothing inside a libmesh_assert() there is
> expensive.  Putting std::lower_bound inside an O(N) loop isn't great,
> but I don't see why it would be so much worse with METHOD=dbg on.
> push_back() should be fine since the proper amount of memory has been
> preallocated outside the loop

'dbg' turns on some serious gcc standard library error checking...  You can
remove the lines "-D_GLIBCXX_DEBUG -D_GLIBCXX_DEBUG_PEDANTIC"  from
Make.common and see what you get.  Should be *way* faster.

Sorry, my uselessness is around 90% today -- we had to shut down work and
get out to prepare for an impending storm.

-Ben