Present: Frank, Roland, Matt, Josh
Stampede slowdown in Noether release: * Jim reported on the mailing list that setting LoopControl::settle_after_iteration = 0 fixed the issue
Include link to archive email in mailman: * do not wait for mailman 3 which might be while before it is released * for einsteintoolkit.org we have root access and can modify mailman * adapt proposed scripts in ticket #719 * further discussion in ticket
using intel 14 compiler: * no one has yet successfully compiled the code with it or run a simulation with it
Misc: * call to add examples to the gallery http://einsteintoolkit.org/about/gallery/ . Previously Roland and Peter had promised examples, Philipp or Christian R. were volunteered for a collapse (to BH or NS) example * please review tickets: https://trac.einsteintoolkit.org/query?status=review&col=id&col=summ...
Yours, Roland
On 27 Jan 2014, at 19:19, Roland Haas roland.haas@physics.gatech.edu wrote:
using intel 14 compiler:
- no one has yet successfully compiled the code with it or run a
simulation with it
I successfully built with
icc (ICC) 14.0.0 20130728
on Datura and committed the optionlist to simfactory a few months ago. It is in simfactory/mdb/optionlists/datura-intel14.cfg. Since you also need to use mpirun compiled with Intel 14, you also need a new runscript, which I have just committed. The test results for the full ET thornlist are:
2 processes:
Tests failed:
waveinterp-2p (from CarpetInterp) refluxing-delayed (from Refluxing) refluxing-delayed-sources (from Refluxing) regression_test (from SphericalHarmonicRecon)
1 process:
Tests failed:
waveinterp-1p (from CarpetInterp) refluxing-delayed (from Refluxing) refluxing-delayed-sources (from Refluxing) regression_test (from SphericalHarmonicRecon)
The waveinterp tests are failing on all machines anyway, due to an unknown problem in init_3_timelevels. The other tests pass.
I don't think I needed to do anything special to get Intel 14 to work. What problems did other people encounter?
On Tue, Jan 28, 2014 at 03:05:16PM +0100, Ian Hinder wrote:
I don't think I needed to do anything special to get Intel 14 to work. What problems did other people encounter?
If you are interested in CTGamma, you might encounter this (at least with compiler options similar to the Intel13 options on stampede):
": internal error: ** segmentation violation signal raised ** Access violation or stack overflow. Please contact Support.
compilation aborted for /work/00920/tg459479/Cactus/configs/sim/build/CTGGauge/shift_gamma_driver.c (code 4)
Frank
users@lists.einsteintoolkit.org
-
Frank Loeffler -
Ian Hinder -
Roland Haas