Commit 02f72e37 authored by Alexander Neukirchen's avatar Alexander Neukirchen

No fi outside main.

parent 7873cae7
......@@ -138,7 +138,7 @@ CONTAINS
CALL subPotDen(vxcForPlotting,vTot,vCoul)
CALL makeplots(stars, atoms, sphhar, vacuum, input, mpi,oneD, sym, cell, &
noco,nococonv, vxcForPlotting, PLOT_POT_VXC, sliceplot)
IF ((mpi%irank.EQ.0).AND.(sliceplot%iplot.LT.32).AND.(MODULO(fi%sliceplot%iplot,2).NE.1)) THEN
IF ((mpi%irank.EQ.0).AND.(sliceplot%iplot.LT.32).AND.(MODULO(sliceplot%iplot,2).NE.1)) THEN
CALL juDFT_end("Stopped self consistency loop after plots have been generated.")
  • @wortmann: With this commit, or rather because of the one before it, the PlotOnlyMT test succeeds again. It failed when checking values, that were taken after a successful iteration or even a second one. Values resulting from the starting density were ok before too. Now, the iterations do not even happen anymore, which is way more sensible, and the values stay the same for all three plot test runs (though some are set to 0, which we want).

    That means: the only test left failing is the relax one. I suppose that is because a) it runs multiple iterations and b) it is fully fully noco. The only other FFN test meeting the requirement of grepping a value after the diagonalization is my SF test Fe_bcc_SF_LDA, grepping for the magnetic moment after one iteration. That test is not terribly sensitive and I might rethink it.

    So the problem that lead to test failures after the openacc merge can be pinpointed to the hsmt part for the FFN case I guess. I hope this helps you find what exactly went wrong. :)

    @hilgers: The plot tests are now way quicker. PlotOnlyMT went from 54s to 17s. Great success!

Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment