#1464: Dissipation test cases get stuck on Bethe ----------------------+----------------------------------------------------- Reporter: eschnett | Owner: Type: defect | Status: new Priority: major | Milestone: ET_2013_11 Component: Other | Version: development version Keywords: | ----------------------+----------------------------------------------------- I was running the Dissipation test cases on Bethe, and found that they got stuck on Bethe. That is, they kept running for many hours, and I had to kill them manually. This is for test_ah.par and test_ob.par, when running on 1 process.
The last output I see is: {{{ INFO (IDAxiBrillBH): Calling axisymmetric solver cyc= 2 max(res)=0.00E+00 tolerance condition tol=1.00E-08 satisfied INFO (IDAxiBrillBH): Solve complete rmax = 2.211409083630045E-009 axibheps = 1.000000000000000E-008 psi2d = 1.00000000000000 -0.130757983129253 Resulting eps = 1.243881126811175E-005 INFO (IDAxiBrillBH): interpolating solution to xyz grid points INFO (IDAxiBrillBH): converting to physical metric INFO (IDAxiBrillBH): computing ADM mass ADM mass: 1.92127863511242 }}}
When running on 2 processes, the test case QuasiLocalMeasures/qlm-bl.par got stuck in a similar manner. The last output is {{{ INFO (QuasiLocalMeasures): Initialising Quasi-Local Measure calculations }}}
#1464: Dissipation test cases get stuck on Bethe -----------------------+---------------------------------------------------- Reporter: eschnett | Owner: Type: defect | Status: new Priority: major | Milestone: ET_2013_11 Component: Other | Version: development version Resolution: | Keywords: -----------------------+----------------------------------------------------
Comment (by rhaas):
Is this still happening? I just tried this with the current proposed for release code and find no hang when using a 12 OMP threads and 1 process. The tests are in ~rhaas/simulations/?p?t/ and the source tree in ~rhaas/ET_trunk.
The single modification is my code changes to make the reconstruction work for all intel versions when using curvilinear coordinates.
#1464: Dissipation test cases get stuck on Bethe -----------------------+---------------------------------------------------- Reporter: eschnett | Owner: Type: defect | Status: closed Priority: major | Milestone: ET_2013_11 Component: Other | Version: development version Resolution: fixed | Keywords: -----------------------+---------------------------------------------------- Changes (by eschnett):
* status: new => closed * resolution: => fixed
Comment:
I believe this was caused by a bad interaction with other tests that were running at the same time, and which used too much memory.
trac@lists.einsteintoolkit.org