<div dir="ltr">Matt,<br><div><br></div><div>The problem results with 1-4 cores. I have only observed the error when using GAMG. <br>I do not get any error with:<br><br>-ksp_type bcgsl -lspaint_pc_type hypre -lspaint_pc_hypre_type boomeramg <br>
</div><div><br></div><div>or<br><br></div><div>-ksp_type bcgsl<br><br></div><div>I do get an error with<br><br>-ksp_type bcgsl -pc_type gamg -pc_gamg_threshold 0.01
-pc_gamg_type agg -pc_gamg_agg_nsmooths 1
-pc_gamg_sym_graph true -ksp_monitor_true_residual<br><br></div><div>and<br></div><div><br></div><div>-ksp_type bcgsl -pc_type gamg -pc_gamg_threshold 0.01 -mg_levels_ksp_type richardson -mg_levels_pc_type sor -mg_coarse_ksp_type richardson -mg_coarse_pc_type sor -mg_coarse_pc_sor_its 5 -pc_gamg_type agg -pc_gamg_agg_nsmooths 1 -pc_gamg_sym_graph true -ksp_monitor_true_residual<br>
<br></div><div>John<br></div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Tue, Sep 24, 2013 at 9:58 AM, Jed Brown <span dir="ltr"><<a href="mailto:jedbrown@mcs.anl.gov" target="_blank">jedbrown@mcs.anl.gov</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div class="im">John Mousel <<a href="mailto:john.mousel@gmail.com">john.mousel@gmail.com</a>> writes:<br>
<br>
> I cloned a fresh repo and built from scratch. I'm seeing the same error I<br>
> previously reported with both intel and gnu compilers.<br>
<br>
</div>Can you reproduce with a PETSc example or create a test case so that we<br>
can reproduce?<br>
<div class="im"><br>
> Also, I get warnings in MatSOR_SeqAIJ_Inode during the build.<br>
><br>
> src/mat/impls/aij/seq/inode.c: In function ‘MatSOR_SeqAIJ_Inode’:<br>
> src/mat/impls/aij/seq/inode.c:2758: warning: ‘sum2’ may be used<br>
> uninitialized in this function<br>
> src/mat/impls/aij/seq/inode.c:2758: warning: ‘sum3’ may be used<br>
> uninitialized in this function<br>
> src/mat/impls/aij/seq/inode.c:2758: warning: ‘sum4’ may be used<br>
> uninitialized in this function<br>
> src/mat/impls/aij/seq/inode.c:2758: warning: ‘sum5’ may be used<br>
> uninitialized in this function<br>
> src/mat/impls/aij/seq/inode.c:2761: warning: ‘v2’ may be used uninitialized<br>
> in this function<br>
> src/mat/impls/aij/seq/inode.c:2761: warning: ‘v3’ may be used uninitialized<br>
> in this function<br>
> src/mat/impls/aij/seq/inode.c:2761: warning: ‘v4’ may be used uninitialized<br>
> in this function<br>
> src/mat/impls/aij/seq/inode.c:2761: warning: ‘v5’ may be used uninitialized<br>
> in this function<br>
> src/mat/impls/aij/seq/inode.c:2758: warning: ‘sum1’ may be used<br>
> uninitialized in this function<br>
> src/mat/impls/aij/seq/inode.c:2761: warning: ‘v1’ may be used uninitialized<br>
> in this function<br>
<br>
</div>I don't think these warnings are correct, but I've seen them too when<br>
building in optimized mode with gcc-4.8, so they should be eliminated.<br>
As we occasionally do elsewhere, we should work around such warnings.<br>
Mark?<br>
</blockquote></div><br></div>