[QE-users] [SUSPECT ATTACHMENT REMOVED] Floating underflow in QE6.4.1 under Linux when running examples

[Matthew Marcus]

First, I'd like to apologize in advance for any violations of
mailing-list etiquette.  I'm new to the mailing list and to QE.
I've  been running through the examples in PW and XSpectra.  In the
PW/examples folder, the following examples result in sets of warnings
like these:
Note: The following floating-point exceptions are signalling:
IEEE_UNDERFLOW_FLAG IEEE_DENORMAL
Note: The following floating-point exceptions are signalling:
IEEE_UNDERFLOW_FLAG IEEE_DENORMAL
Note: The following floating-point exceptions are signalling:
IEEE_UNDERFLOW_FLAG IEEE_DENORMAL
Note: The following floating-point exceptions are signalling:
IEEE_UNDERFLOW_FLAG IEEE_DENORMAL

All of the tests in run_tests executed with success.

I don't know how to compare the outputs with those in the reference
directories, but they all seem to be populated with numbers instead of
NaN.  The examples in PW displaying this behavior are
example01, example04, example06, example07, example09,
example11,example12, ESM_example, EXX_example, gatefield and
vdwDF_example.  The others ran to completion with no complaint.

In XSpectra, of the four examples, the two that die this way are
run_example_Cu_L23 and run_example_SiO2_USPP.  The latter produces
.dat files which look like those in the reference folder except for a
4eV shift, even though they displayed the underflow warning. The
former example dies at the point indicated in the attached shell file.
While it produces an output XANES spectrum (xanes.dat), that spectrum
doesn't look like any of the .dat files in the reference folder.
The Cu.wfc file looks like a concatenation of wavefunctions for
1s,2s,2p, 3s and 3p, as I think it should.
            Matthew Marcus
            Advance Light Source
            Lawrence Berkeley Lab