Ivan Carnimeo
1a4a6e3038
arrays passed to ppcg* from c_bands are now passed as device arrays
2020-07-03 15:17:09 +02:00
Ivan Carnimeo
d9fe63b601
PPCG on GPU, Gamma and K-points (some minor fixes required)
...
Committer: Ivan Carnimeo <icarnime@r033c01s04.galileo.cineca.it>
modified: KS_Solvers/Makefile
modified: KS_Solvers/PPCG/Makefile
new file: KS_Solvers/PPCG/generic_cublas.f90
modified: KS_Solvers/PPCG/make.depend
new file: KS_Solvers/PPCG/ppcg_gamma_gpu.f90
new file: KS_Solvers/PPCG/ppcg_k_gpu.f90
modified: PW/src/c_bands.f90
2020-06-19 11:15:03 +02:00
Pietro Bonfa
4517cc1842
Fix zero dimensional allocations on the GPU for massively parallel simulations
2020-06-10 17:01:22 +02:00
Pietro Bonfa
1905fab836
Fixed CPU compilation and a couple of optimizations to recently added GPU code
2020-06-04 23:43:54 +02:00
Pietro Bonfa
30df69888f
Never use variables declared outside the subroutine in cuf kernels, it breaks PGI (at least old versions).
2020-06-04 18:25:02 +02:00
Pietro Bonfa
80e368bf30
Removed some code from cuf kernel (hopefully qe_erf is equivalent to Fortran's erf) and fixed #43 .
2020-06-04 14:29:24 +02:00
Pietro Bonfa
637552da0e
Fixed compilation on PGI < 19.10
2020-06-01 10:25:06 +02:00
Pietro Bonfa
d9af16b29c
Fix dependencies
2020-05-30 20:31:26 +02:00
Pietro
4885af13e3
Merge branch 'orthoatwfc-GPU' into 'gpu-develop'
...
orthoatwfc gpu
See merge request QEF/q-e-gpu!14
2020-05-29 08:01:37 +00:00
fabrizio22
7daf0ea809
orthoatwfc_gpu small fix
2020-05-28 17:06:42 +02:00
fabrizio22
caf15cc6a9
orthoatwfc_gpu
2020-05-28 17:06:42 +02:00
fabrizio22
74c992fe75
atomic_wfc_gpu complete
2020-05-28 17:06:42 +02:00
fabrizio22
07f3365449
atomic_wfc_gpu
2020-05-28 17:06:42 +02:00
fabrizio22
ac7c155c02
ortho_swfc_gpu complete
2020-05-28 17:06:42 +02:00
fabrizio22
473d89b5ca
ortho_swfc_gpu - scratch
2020-05-28 17:06:42 +02:00
fabrizio22
58da512622
orthoUwfc_gpu - scratch
2020-05-28 17:06:42 +02:00
Pietro Bonfa
cdd440a230
Merge branch 'develop' into gpu-develop
2020-05-28 15:19:12 +02:00
giannozz
3707e425b6
Merge branch 'newtest' into 'develop'
...
Added first test for PP
See merge request QEF/q-e!948
2020-05-27 15:43:20 +00:00
fabrizio22
3ac6f623ba
small change in coul_cut_2d for gpu alignment
2020-05-27 16:29:42 +02:00
fabrizio22
5c7d6e58de
additional argument removed in deriv_drhoc_gpu
2020-05-26 18:31:45 +02:00
Paolo Giannozzi
086af3c87d
Memory report updated
2020-05-26 16:06:29 +00:00
fabrizio22
2a29f075c4
additional argument removed in dvloc_ gpu routines
2020-05-26 17:12:57 +02:00
Paolo Giannozzi
b450381197
Work space in addusforce can be reduced
2020-05-26 09:52:59 +02:00
T. Thonhauser
22805feedd
Merge branch 'develop' of gitlab.com:QEF/q-e into vdW-DF3
2020-05-25 11:18:24 -04:00
Paolo Giannozzi
b2b2792f14
The recent chamge to allow restarting meta-GGA calculations from non-meta ones
...
did not work as expected because the file name has an additional ".dat" or
".hdf5" prefix. Check moved where files are opened.
2020-05-25 08:29:45 +02:00
T. Thonhauser
8168e5d7c2
Merge branch 'develop' of gitlab.com:QEF/q-e into vdW-DF3
2020-05-21 13:13:08 -04:00
T. Thonhauser
f047985b92
vdW-DF: Implementing vdW-DF3-opt1, vdW-DF3-opt2, and vdW-DF-C6 & inlc cleanup
...
We implemented the new vdW-DF type functionals vdW-DF3-opt1, vdW-DF3-opt2, and
vdW-DF-C6. For purposes of non-local functional implementation and evaluation,
we also grouped similar implementations into ranges for the "inlc" variable. To
this end, we have reserved inlc=1 to inlc=25 for vdW-DF development and we
moved rVV10 to inlc=26. This allows for a cleaner code when parsing through the
various non-local functionals. In addition, we have removed all instances of
"inlc" and "get_inlc" in several source files where they are no longer needed.
This work was performed by T. Thonhauser and is supported by NSF Grant No.
1712425.
2020-05-21 12:49:41 -04:00
Paolo Giannozzi
d6aaeae561
make.depend updated
2020-05-21 18:26:47 +02:00
Paolo Giannozzi
f8339aaa5b
Following a suggestion by Michal Krompiec: allow starting a meta-GGA
...
calculation from GGA data. Useful for cases of difficult convergence.
It wasn't possible until now because the code complained that the kinetic
energy density file was missing when reading the starting charge density.
2020-05-21 16:45:16 +02:00
Pietro Bonfa
add56f86b4
Fixed use of uninitialized variable in Hubbard potential contribution to Hpsi
2020-05-21 11:47:13 +02:00
Pietro Bonfa
d309fbd2ed
Fix forces for ESM and 2D cutoff. Closes #36 .
2020-05-20 18:27:24 +02:00
giannozz
06e55a689d
Merge branch 'requests' into 'develop'
...
New f90 code to find ibrav and celldm, starting from cell arbitrary axis
See merge request QEF/q-e!930
2020-05-19 19:20:08 +00:00
Lorenzo Paulatto
1a9fb59b3e
Removed obsoleted python scrip, replaced by native fortran code
2020-05-19 15:13:45 +02:00
Lorenzo Paulatto
91b9b37fd4
removed duplicated modified version of latgen
2020-05-19 12:09:20 +02:00
Lorenzo Paulatto
3796879853
latgen subroutin changed to latgen_lib subroutinw which does not stop on
...
error (it returns zero, an error code and message)
New wrapper routine "latgen" behaves exactly like the old "latgen"
scan_ibrav modified
added copyright note to powell
2020-05-19 12:06:36 +02:00
Pietro Bonfa
20818da2b7
Removed useless pointers, fixed CPU compilation
2020-05-18 21:58:39 +02:00
fabrizio22
6b0906dbd3
vhpsi-GPU input device
2020-05-15 13:34:22 +02:00
fabrizio22
e8aa3b8d0c
vhpsi-GPU cleanup1
2020-05-15 13:11:02 +02:00
fabrizio22
101e1622aa
vhpsi-gpu U alloc out
2020-05-14 13:49:18 +02:00
fabrizio22
6d50b66f49
vhpsi-gpu U bec_gpu
2020-05-14 13:44:20 +02:00
fabrizio22
d83fcd6a16
vhpsi-gpu U k
2020-05-14 13:40:40 +02:00
fabrizio22
a04532b4ee
vhpsi-gpu U gamma
2020-05-14 13:37:18 +02:00
Paolo Giannozzi
cba3fa696e
[skip-CI] The main program in "cell2ibrav.f90" was called "ibrav2cell"!
2020-05-13 11:54:54 +02:00
Lorenzo Paulatto
67cd466986
New f90 code to find ibrav and celldm, starting from arbitrary axis in
...
input. The code requires a modified version of "latgen" which returns
zero instead of crashing on error. Also, the tolerances of powell
minimization have been tightened a bit.
2020-05-12 18:36:07 +02:00
Pietro Bonfa
36915a4a6d
A few more checks for memory allocation failures
2020-05-12 15:19:21 +02:00
Pietro Bonfa
62bf041fc4
Fixed build dependencies and loosen test-suite check on number of iterations
2020-05-09 16:23:41 +02:00
Pietro Bonfa
d8645ecbff
Skip computing Q(G) for types with no atoms in ATOMIC_POSITIONS. Fixes #34
2020-05-09 11:25:21 +02:00
Pietro Bonfa
497ce579f1
Check scratch memory allocations
2020-05-08 00:03:46 +02:00
Pietro Bonfa
d5b4e8db6e
Oops, forgot a couple of cudafor
2020-05-03 23:52:31 +02:00
Pietro Bonfa
1cf057ce6b
Fixed CPU compilation
2020-05-03 23:36:47 +02:00