Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Gpurational merge #282

Closed
wants to merge 124 commits into from
Closed

Gpurational merge #282

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
124 commits
Select commit Hold shift + click to select a range
4ddef6c
Here the latest GPU related files from my local code are taken over,
Feb 8, 2013
573483a
fixed new allocation of solver fields in all outer solvers
Feb 13, 2013
56b8393
added GPU support in det and detratio
Feb 13, 2013
5d2db32
adapted function arguments to allow for hamiltonian field
Feb 15, 2013
0df793a
fixed all compile errors due to c99
Feb 18, 2013
5440583
both degenerate and ND solvers in EO running again.
Feb 20, 2013
a1d7898
renamed ND Matrix to match name in cpu code
Feb 20, 2013
3884064
complex kappa now possible in HoppingMatrix
Feb 20, 2013
7f1e65c
worked on:
Mar 11, 2013
924b5e1
removed finalize temporal gauge routine in invert_doublet which caused
Mar 13, 2013
f7b84bb
made MPI version compile again
Mar 15, 2013
9989d3e
some small changes
Apr 3, 2013
75693b0
removed all REAL data type define related stuff and changed to float
Apr 26, 2013
405259b
removed subsequent mem alloc in init_temporalgauge_trafo which lead to
Apr 30, 2013
e6a478a
added first version of mixed clover tm inversion
Aug 13, 2013
ed3454b
Merge branch 'master' into gpurational
Aug 13, 2013
f01b09c
added mpi support for clover
Aug 14, 2013
262e6d7
Number of gpus per node can now be specified in gpu input section
Aug 19, 2013
424d485
fixed issue with gpu shift solver and added max EV normalization in n…
Aug 19, 2013
2fa0747
improved non-EO version of degenerate matrix substantially,
Aug 19, 2013
1022e3c
added some functionality to limit inner solver iterations in 1+1
Sep 17, 2013
69c76e3
first trials for mixed prec cg with ONLY 32bit opertions in inner solver
Sep 20, 2013
6c6255d
first working float/double mixed solver
Sep 27, 2013
1c4b9b1
started adding 32 bit gauge field
Sep 30, 2013
4e70534
finished adding 32 bit gauge field.
Oct 1, 2013
3986564
removed old solver version
Oct 3, 2013
4279626
bgq version added, not checked
Oct 7, 2013
fd0220d
added xlc hopping part
Oct 9, 2013
4a2ff32
fixed forgotten #endif
Oct 9, 2013
0aa2b4c
working BG/Q version
Oct 29, 2013
47e0e45
removed some testing stuff again
Oct 29, 2013
85de341
added su3 times 1 vector qpx
Oct 31, 2013
07f50c9
included overlapping communication from InterleavedNDTwistedClover
Nov 1, 2013
702c734
omp orphaning of matrix is working and gives speedup on BGQ
Nov 2, 2013
a64c230
adde a NOCOMM single prec Hopping Matrix and corresponding benchmark
Nov 5, 2013
997b120
fixed a potential bug in mixed_cg that allowed the application of the
Nov 7, 2013
5d78413
fixed warnings due to implicit declaration of _32 init and free
Nov 7, 2013
294b5ea
added mixed solver to monomials DET and DETRATIO
Nov 7, 2013
3d06f47
added timing in mixed solver
Nov 12, 2013
c11e78c
fixed issue with sloppy precision
Nov 12, 2013
cda6636
brought xchange_2fields and mpi_init into the state of
Nov 15, 2013
8589b26
GPU input parameter "DeviceNum" can now also be used to set the number
Nov 27, 2013
c374865
added first working version of ND nonEO matrix.
Dec 4, 2013
daa1570
fixes a bug that prevented code from compiling when HALFSPINOR was
Jan 23, 2014
5d1c187
nd non-eo operators and solvers
Jan 23, 2014
d3d337e
getting acceptance with non-eo NDRAT and NDRATCOR monomials
Jan 23, 2014
2fc6cde
cg-flops for non-eo version
Jan 29, 2014
d8c8ce7
working double sequential mms solver on gpu
Jan 31, 2014
1a19130
fixed incompatibility with cpu deriv_SB that was introduced by changes
Jan 31, 2014
6077b08
fixed a bug:
Jan 31, 2014
f8fb681
su3 matrix mult with +=
Feb 5, 2014
afb5a8f
added basic double2 tex support via int4 detour
Feb 5, 2014
91e2fab
added a double Hopping Matrix aequivalent
Feb 5, 2014
50e869c
bug fix in polynomial initial guess generation -> currently disabled
Feb 5, 2014
9e03504
wrapped a test_double_operator by a #ifdef MATRIX_DEBUG
Feb 5, 2014
f790aef
working double inverter for nd doublet called sequentially from gpu m…
Feb 5, 2014
d4cd522
some restructuring to make transition to double2 easier
Feb 5, 2014
e10a564
working ndrat ndratcor monomials
Feb 10, 2014
798d4c0
added working TRUE mms solver in pure double
Feb 12, 2014
ec2fc3e
adapted debug_level for printing out pre/post gauge-fixing data
Feb 12, 2014
43ba67c
fewer output by addign #ifdef LOWOUTPUT
Feb 12, 2014
624facb
changed some comments
Feb 12, 2014
b027e76
removed some further output
Feb 12, 2014
b6d2be5
fixed a small bug when benchmark is on
Feb 12, 2014
60a6518
fixes a bug that lead mms solver to break too early
Feb 13, 2014
7c2a8f4
fixed a wrong pointer type
Feb 18, 2014
36c5455
added relativistic basis in double nd operator
Feb 19, 2014
ed64afc
moved tm and nd-tm operators from the mixedsolver files to a new
Feb 19, 2014
f88cd4c
forcing less iterations in mixed solver
Feb 20, 2014
1f5c162
Fixed cudaMemcpyToSymbol which did not work on cuda version > 3.2 as
Feb 20, 2014
410d105
unified the way grid and block sizes are set for solver/matrix related
Feb 28, 2014
359bdf5
Added double mpi light and nd matrix for with mpi
Mar 5, 2014
ddf5aeb
fixes issue with temporalgauge and mpi
Mar 6, 2014
fbfbb5d
adds relativistic basis to double light matrix
Mar 6, 2014
b11365d
code smoothening
Mar 6, 2014
e6c353a
code rearrangement for more transparency
Mar 11, 2014
18b49aa
fixed deriv_SB with MPI
Mar 11, 2014
c994751
added missing xchange + g_debug_level adapted
Mar 11, 2014
7f11f1e
added debug_level
Mar 15, 2014
a941dc8
fixed a potential bug in dotprod with MPI - abtracted blas
Mar 15, 2014
e9100be
started working on clover
Mar 15, 2014
d81ca62
default device_num depending on MPI
Mar 17, 2014
d22c691
fixed a bug related to normalization of max. ev
Mar 17, 2014
ab03ec0
eigenvalue calculation for non-eo
Mar 18, 2014
2505f4f
*made compile without warnings (apart from one not coming out of GPU/)
Mar 28, 2014
eca04a7
Merge remote-tracking branch 'upstream/master' into gpurational_merge
Mar 28, 2014
e3aac4b
changing measure_gauge_action to measure_plaquette in temporalgauge.c
Mar 29, 2014
3ecf36b
uncommented double matrix check which should be the default
Apr 4, 2014
cd0a63b
first attempt to get rid of if(usegpu_flag) in monomials
Apr 8, 2014
25688e6
working mms reliable update solver
Apr 15, 2014
0d3f35c
replaced all ifdef MPI and similar statements with ifdef _USE_MPI
urbach Apr 28, 2014
b122279
Merge pull request #1 from urbach/replaceMPIdef
florian-burger Apr 28, 2014
a0e4e16
Merge branch 'gpurational_merge' of github.com:florian-burger/tmLQCD …
Apr 28, 2014
8ee6b9b
removed #include's in mixed_solve_eo_nd.cuh that already exist in mix…
Apr 30, 2014
3fae26a
This fixes an issue that when compiled without GPU_DOUBLE
May 2, 2014
b83dbf1
Merge branch 'hmc_interleaved_mixed' into gpurational_merge
May 2, 2014
cba2b22
merging changes of detached commit 8630f7a0862 into merge branch
Nov 13, 2014
f940cff
fixes interchanged gpu <-> cpu calls in derivSB
Nov 13, 2014
e58919b
added a monomial solver that wraps the inversions in monomials
Nov 13, 2014
a7a9e4e
Merge branch 'noneo_nd_matrix' into gpurational_merge
Nov 13, 2014
d1fae5e
Changes:
Dec 9, 2014
c20b148
some first work on nd+clover
Dec 11, 2014
51164ad
fixed MPI -> _USE_MPI
Dec 12, 2014
949cbd7
fixed MPI -> _USE_MPI in mixed_solve_eo_nd
Dec 12, 2014
81c9bd0
fixed a bug for MPI set_global_sizes
Dec 30, 2014
2d713cc
undoing overlapping in xchange
Jan 2, 2015
4db656f
added mixed_cg in monomial_solve
Jan 3, 2015
6d27d7f
added D_psi_32
Jan 5, 2015
9f2979e
added mixed cg ND mms solver and 32bit Matrices (w/o clover)
Jan 9, 2015
6fd402a
bicg in det and detratio now wrapped via solve_degenerate
Jan 13, 2015
920af80
fixed bug with non-eo+bicgstab related to g_mu sign swap before
Jan 14, 2015
8cba40e
solves mpi problem with infiniband
Mar 17, 2015
6c4d232
fixed double allocation of host exchange fields and a missing free
Apr 1, 2015
7193ad0
non_eo inversions, source generation and io fixed, dropped calls to e…
May 16, 2015
a1d95ef
resolved merging conflicts from merging the current tmLQCD master
urbach Jun 9, 2015
9ac5e97
residual ifdef MPI replace by ifdef USE_MPI
urbach Jun 9, 2015
1a859f9
defined ALIGN32 and ALIGN_BASE32 in all cases, not neccessarily to th…
urbach Jun 9, 2015
d224dcf
removed BG/L and BG/P related code
urbach Jun 9, 2015
ed8f07b
use exclusively the defined solver types and include SUMR in the list
urbach Jun 9, 2015
b9a5437
added missing version of update_backward_gauge_32 in case of disable-…
urbach Jun 9, 2015
c83a164
corrected a printf format mistake
urbach Jun 9, 2015
7dac82b
removed warning due to a re-definition
urbach Jun 9, 2015
d30ade8
32 bit clover working with mixedcg
Jun 20, 2015
e375e2f
missing Qm added
Jun 23, 2015
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
2 changes: 1 addition & 1 deletion DirectPut.c
View file
Open in desktop
Expand Up @@ -24,7 +24,7 @@
#include <stdio.h>
#include <sys/types.h>
#include <stdint.h>
#ifdef MPI
#ifdef _USE_MPI
# include <mpi.h>
#endif
#ifdef OMP
Expand Down