- old variables substituted by transition variables,
eventually to be renamed as the old one, when transition will be over
- All array now are distributed across one or more processor groups
Note that ** NOW CPV IS almost MEMORY BOTTLENECK FREE **
Warning, not all functionalites have been tested, CPV with band distribution
should be considered in ALPHA testing.
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@7376 c92efa57-630b-4861-b058-cf58834340f0
__OPENMP in such a way that compilation with openmp enabled but without
-D__OPENMP doesn't yield a compilation error (for testing purposes)
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6890 c92efa57-630b-4861-b058-cf58834340f0
- Parallelization of some beta.dot.c arrays to reduce the
memory requirement of the code when using many processors
- cleanup in wannier functions related subroutines
- various cleanup
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@5346 c92efa57-630b-4861-b058-cf58834340f0
in the restart subroutine related to the possibility of "correclty"
restart a simulation setting to zero the velocities of some degree
of freedom.
Simple tests shows that if a simulation is restarted setting to
zero the velocities of electrons without setting to zero the velocities
of ions and vice versa, the generated trajectories of ions contain
unphysical fast oscillations.
- Some improvement in the parallelism when cell dynamics is active.
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@5122 c92efa57-630b-4861-b058-cf58834340f0
allocated memory now scale with the number of processor
in the ortho group. To be exact memory scale like the
square root of the number of processor used in the ortho
group. With 4 proc the allocated memory is 1/2 of the memory
required with a single proc.
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@4058 c92efa57-630b-4861-b058-cf58834340f0
potentials and energy is ok, on the forces there is still
some small problems. Further clean ups required.
Contributed by
Patrick Sit
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3605 c92efa57-630b-4861-b058-cf58834340f0
- correct an unquoted string (iosys) in PW/input.f90:483
(this was the cause of tonight compilation failere)
- correct a few incorrect format strings
- make more use of the constants module and thus
provide more consistent units. NOTE, this has some
numerical changes in the outputs, as in some places
rather low precision and inconsistent numbers were
used for unit conversion.
- convert all(?) single precision constants to double
using the attached little perl program.
exceptions: efermi.f90 (as it is supposed to be rewritten
anyways), plotbands.f90 (it uses single precision everywhere,
which may result in saving a significant amount of memory,
so i converted the two double precision constants to single).
Unused routine 'set_fft_grid' removed
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3602 c92efa57-630b-4861-b058-cf58834340f0
- BGL porting of CP almost complete. New command
line argument "-ntask_groups" to specify the
number of task group used in parallelization.
to generate an executable with BGL extensions,
in the file make.sys the macro __BGL should
be specified.
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3069 c92efa57-630b-4861-b058-cf58834340f0
first step needed to parallelize neb over images also for CP.
Next we need to add the right communicator to all communications
- subroutine reduce substituted everywhere with mp_sum
- mp_sum for array with 4dims added in mp.f90
- workaround for xlf compiler, it has problems compiling file with
initialization of large array in the definition line,
see Modules/input_parameters.f90 , initialization moved to
Modules/read_cards.f90 .
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@2946 c92efa57-630b-4861-b058-cf58834340f0
- the logic of the combined index for US PP is now the same everywhere
(i.e in PW, CP, and in the pseudopotential format):
do iv=1,N
do jv=iv,N
ijv=jv*(jv-1)/2+iv
(in PW the indices are called nb, mb). In order to get ijv from (iv,jv):
if (iv > jv) then
ijv=iv*(iv-1)/2+jv
else
ijv=jv*(jv-1)/2+iv
end if
- the above change also fixes a serious bug affecting Vanderbilt US PP
in UPF format (only half of the qfcoef array was present, but not the
good half)
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@2916 c92efa57-630b-4861-b058-cf58834340f0
PLEASE NOTE: the logic for packed index is as follows everywhere
do i=1,N
do j=1,i
ij = (i-1)*i/2 + j
This is equivalent to
ij = 0
do i=1,N
do j=1,i
ij = ij + 1
This is not (yet) the same as used in PW, though
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@2871 c92efa57-630b-4861-b058-cf58834340f0
- Many ortho auxiliary functions (tauset, rhoset, sigset, calphi, updatc)
are now in common between FPMD/CP, and moved to module ortho_base.f90
- In FPMD, three index vectors, related to real space like charge and potential
have been substituted with single index vector like in CP, for compatibility
and efficiency.
- Bug fix in pwtools/matdyn.f90 a logical variable was used in place of a
character variable
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@2694 c92efa57-630b-4861-b058-cf58834340f0
conversion to real => DBLE
(including real part of a complex number)
conversion to complex => CMPLX
complex conjugate => CONJG
imaginary part => AIMAG
All functions are uppercase.
CMPLX is preprocessed by f_defs.h and performs an explicit cast:
#define CMPLX(a,b) cmplx(a,b,kind=DP)
This implies that 1) f_defs.h must be included whenever a CMPLX is present,
2) CMPLX should stay in a single line, 3) DP must be defined.
All occurrences of real, float, dreal, dfloat, dconjg, dimag, dcmplx
removed - please do not reintroduce any of them.
Tested only with ifc7 and g95 - beware unintended side effects
Maybe not the best solution (explicit casts everywhere would be better)
but it can be easily changed with a script if the need arises.
The following code might be used to test for possible trouble:
program test_intrinsic
implicit none
integer, parameter :: dp = selected_real_kind(14,200)
real (kind=dp) :: a = 0.123456789012345_dp
real (kind=dp) :: b = 0.987654321098765_dp
complex (kind=dp) :: c = ( 0.123456789012345_dp, 0.987654321098765_dp)
print *, ' A = ', a
print *, ' DBLE(A)= ', DBLE(a)
print *, ' C = ', c
print *, 'CONJG(C)= ', CONJG(c)
print *, 'DBLE(c),AIMAG(C) = ', DBLE(c), AIMAG(c)
print *, 'CMPLX(A,B,kind=dp)= ', CMPLX( a, b, kind=dp)
end program test_intrinsic
Note that CMPLX and REAL without a cast yield single precision numbers on
ifc7 and g95 !!!
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@2133 c92efa57-630b-4861-b058-cf58834340f0
dion, beta, bec everyware.
- subroutines formfn, compute_beta, nlsm1, nlsm2, ecc ... now are common
between FPMD and CPV, a lot of clean ups!
- Changes in stdout: relevant physical quantities ( positions velocities an cell )
are now printed with the seme format of the corresponding input card,
like in PW, as was suggested by SdG.
- exemple23 updated to reflect the new input namelist "wannier"
- Subroutine init_run now is used in FPMD too.
- WARNING in the stress computed with CP, for a pseudo with core-corrections,
a contribution is missing! Not yet fixed, I need to talk with PG for the
box staff.
- WARNING the examples reference are not updated, I'm on the IBM sp, and
I prefer to update them from a linux machine.
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@2110 c92efa57-630b-4861-b058-cf58834340f0