[cig-commits] [commit] master's head updated: Merge remote-tracking branch 'origin/devel' (32685b6)
cig_noreply at geodynamics.org
cig_noreply at geodynamics.org
Thu Jul 10 13:48:04 PDT 2014
Repository : https://github.com/geodynamics/specfem3d_globe
Branch 'master' now includes:
b173684 fixed a bug in Matthias Meschede's crustal model smoothing routines: latitude was used instead of longitude
bed4deb Merge pull request #69 from komatits/devel
9a28612 added a remark about ellipticity to the users manual.
1e48348 Merge pull request #70 from komatits/devel
cdbee0f switched the default to FORCE_VECTORIZATION = .true.
a2b53e8 Merge pull request #71 from komatits/devel
47332f7 switched from a single observation point to a whole surface around the Earth for Roland_Sylvain integrals
7be81d6 Merge pull request #72 from komatits/devel
25bd1b9 add OpenCL support to Specfem3D-globe
32268ae add OpenCL configuration backend
47badd5 point m4 module to our repository
a9b5683 fix bugs
c6d2aa9 improve speed with clFlush instead of clFinish and fix bugs
d321da5 fix bugs
6895f34 cosmetics
8934464 satellite clipart added to the cover of the manual to prepare for the release of Roland_Sylvain gravity integrals
b399bb5 Merge pull request #73 from komatits/devel
c33bf42 better format for the output of Roland_Sylvain gravity integrals, added output of the norm of the g vector, and added an exit message if the solver is run with ROLAND_SYLVAIN = .true. (only the mesher is needed in that case, since we compute some integrals on the mesh and then we are done, no need for a time-domain solver)
1fce2e1 Merge pull request #75 from komatits/devel
68fd046 put http:// at the beginning of all \url{} addresses. It seems many had been missing for months. Although the result in the PDF is really ugly, it seems to be the only way of making the clickable URL work.
c45b60b Merge pull request #76 from komatits/devel
c38411c fix bugs
5997858 better TOPO_MINIMUM and TOPO_MAXIMUM
6509705 Merge pull request #77 from komatits/devel
bde5c50 added more precise comments and doc about ellipticity calculations
90f8a21 Merge pull request #78 from komatits/devel
49bb85f added a comment referring to eq (14.21) in Dahlen and Tromp (1998) regarding ellipticity
7cbcde8 Merge pull request #79 from komatits/devel
8e77f34 improved a variable declaration
d2496e9 Merge pull request #80 from komatits/devel
5b48368 moved some old files from DATA to utils/oldstuff
63d2961 Merge pull request #81 from komatits/devel
385bfa7 cleaned DATA/topo_bathy
b8d6a16 Merge pull request #82 from komatits/devel
06dd1c2 updated DATA/topo_bathy
0e84408 Merge pull request #83 from komatits/devel
85124e8 compressed the large topo files that are unused by default and added some conversion and display scripts
97da690 Merge pull request #84 from komatits/devel
04ab1bc replaced the old version of ETOPO with the latest one (etopo2v2c), which fixes several issues, and also added ETOPO1 (etopo1_ice_c)
ba3f67c Merge pull request #85 from komatits/devel
2614dd9 done cleaning and adding all the topography files, in particular adding ETOPO1
e63245c Merge pull request #86 from komatits/devel
ffd536a renamed model_crust to model_crust_2_0 to be consistent with existing model_crust_1_0
f2de2cd Merge pull request #87 from komatits/devel
a5dde6d added an email from Daniel Peter about ellipticity as a comment in src/shared/rthetaphi_xyz.f90
f218984 Merge pull request #88 from komatits/devel
5cfc0d4 use ocl events for async dependencies
b2c7b97 add events around copy queue operations
589f129 Merge branch 'devel' into opencl
3bcafda add restrict to boast kernels
d08fd18 enable MANUALLY UNROLL LOOPS by default
62b2186 Merge branch 'opencl' of github.com:Nanosim-LIG/specfem3d_globe into opencl
66a0592 import last version of the kernel references from boast
32bcdea pass launch_bounds macros to OCL processor
9f9ceca add launch bounds qualifiers in gpu kernels
e45b2ec remove reqd_work_group_size from OpenCL kernels (inefficient)
6acd7dc Merge pull request #89 from Nanosim-LIG/opencl
78dfa42 switched back to the right version of DATA (it seems Kevin Pouget's commit of the OpenCL modifications had changed it to an older version)
d561fb7 Merge pull request #90 from komatits/devel
99c106b added Kevin Pouget's description of the new OpenCL routines to the manual
e51b41c Merge pull request #91 from komatits/devel
ec1a33c update git submodules
54fc581 Merge branch 'devel' of https://github.com/geodynamics/specfem3d_globe into opencl
466344f translate OpenCL note in English and regenerate the PDF
b2c09ee Merge pull request #92 from Nanosim-LIG/opencl
3730a2f added the two suggestions made by Elliott
d696d58 Merge pull request #93 from komatits/devel
7582c07 fixed a compilation problem introduced by Kevin Pouget's OpenCL modification: the solver did not compile without GPUs any more. My modification works fine for CPUs, regarding GPUs I am not sure (I am not familiar enough with how the Makefile works for GPU routines)
69ca609 Merge pull request #94 from komatits/devel
198182e updated some comments
cd5bc89 Merge pull request #95 from komatits/devel
615d1f7 tried to use the same notations in model_crust_1_0.f90 and model_crust_2_0.f90; also cleaned useless white spaces in all source files using a script
e1ce7fe Merge pull request #96 from komatits/devel
d03666b renamed INCLUDE_SEDIMENTS_CRUST to INCLUDE_SEDIMENTS_IN_CRUST
a4fbfb9 Merge pull request #97 from komatits/devel
1260874 made source code of model_crust_2_0.f90 closer to that of model_crust_1_0.f90
6e4a1ea Merge pull request #98 from komatits/devel
80bfee7 improved a comment
5ad7f5f Merge pull request #99 from komatits/devel
2e95185 added support (off by default) for the ice layer in crustal models model_crust_1_0.f90 and model_crust_2_0.f90
6df1df1 Merge pull request #100 from komatits/devel
412b039 got rid of MAXIMUM_HEIGHT_TOPO and MAXIMUM_DEPTH_BATHY, which had become useless
9d3a086 Merge pull request #101 from komatits/devel
178f39e also output a timestamp file for ispec == nspec
eb990f2 Merge pull request #102 from komatits/devel
918596c added a test to check that color palette file GMTglobe.cpt exists
d736762 Merge pull request #103 from komatits/devel
8b12737 better cleaning at the end of the GMT script
fdeb613 Merge pull request #104 from komatits/devel
b0ce31f convert final output file from PS to PDF
0fddd2b Merge pull request #105 from komatits/devel
f4e8641 added a color scale to the GMT script
7ccf1c6 Merge pull request #106 from komatits/devel
3cae917 Fix out-of-bounds array access warning.
a5db7e8 Merge pull request #107 from QuLogic/warning-fix
b316cb7 fixed a minor range problem when Roland_Sylvain conditions are off
b119f58 added Julien Derouillat's (from Maison de la Simulation, France) OpenMP support
eb90220 Merge pull request #108 from komatits/devel
be02546 more complete gfortran options
d45fd44 Merge pull request #110 from komatits/devel
a378188 added flag REUSE_EXISTING_OBSERVATION_SURF to be able to reuse an existing observation surface from a previous run. Also added directory utils/Roland_Sylvain_gravity.
ae22573 Merge pull request #111 from komatits/devel
293d10f added a comment
d462ac4 Merge pull request #112 from komatits/devel
86b9b4d folded OpenMP lines that were longer than the standard
c45144d Merge pull request #113 from komatits/devel
ac73846 cosmetic change to an OpenMP statement
349757e Merge pull request #114 from komatits/devel
6fcc510 added the calculation of longitude and latitude of the observation grid for GMT display purposes when the observation grid is read back from a saved file on the disk
e96c78a Merge pull request #115 from komatits/devel
1cb1024 added "blockmedian" to GMT script; added output of time stamp file for ispec == 1
4c2fe25 Merge pull request #116 from komatits/devel
121972c uncommented a line that was commented out by mistake
cbd301f Merge pull request #117 from komatits/devel
ebc9e2c changed blockmedian to blockmean (according to the GMT manual)
f11fa7f Merge pull request #118 from komatits/devel
e6f8bd7 added the calculation of the location of the center of mass of the Earth model and mesh
9c54f7d Merge pull request #119 from komatits/devel
d02f73b removed extra white spaces
c5007a6 Merge pull request #120 from komatits/devel
3027b17 removed "blockmean", which was smoothing out the results way too much
0c658eb Merge pull request #121 from komatits/devel
76b1524 fixed a typo in convert_all_result_files_to_GRD_format.sh and improved the color scale in display_g_vector_norm_SPECFEM3D_GLOBE_with_GMT.sh
f16203a Merge pull request #122 from komatits/devel
70b3411 added Vittorio Ruggiero's additional OpenMP support, and in particular added a new compiler option -DUSE_OPENMP_ATOMIC_INSTEAD_OF_CRITICAL to use OpenMP ATOMIC statements for some critical loops instead of OpenMP CRITICAL regions.
44a503f Merge pull request #123 from komatits/devel
66411e2 initialize moho to zero
dd7f9ad Merge pull request #124 from hejunzhu/devel
72f95ad fixed the problem introduced by Hejun Zhu's commit: the version of DATA and m4 had been changed
b1e1a79 Merge pull request #125 from komatits/devel
dc6d421 removed the "goto 777" that was used to avoid the s362ani + attenuation bug in Hejun Zhu's routine
046ab6a Merge pull request #126 from komatits/devel
b5fe28d fixed a bug I had introduced in OMP END CRITICAL
3c9d5b6 Merge pull request #127 from komatits/devel
02ccaa1 added flags SHIFT_TO_THIS_CENTER_OF_MASS and ONLY_COMPUTE_CENTER_OF_MASS for Roland_Sylvain gravity integrals
1610204 Merge pull request #128 from komatits/devel
f6f55ce left a comment about the fact that it is crucial to leave "moho = 0" to avoid the s362ani + attenuation bug
d296a23 Merge pull request #129 from komatits/devel
a63cd7b added utils/Roland_Sylvain_gravity/output_file_of_the_code.txt and utils/Roland_Sylvain_gravity/saved_observation_grid_real_x_y_z_used_by_the_code.txt.bz2 and updated utils/Roland_Sylvain_gravity/constants.h.in
f3b5ba3 Merge pull request #130 from komatits/devel
a4eba40 added utils/Roland_Sylvain_gravity/HOWTO_use_the_code_for_Roland_Sylvain.txt and renamed the two submission scripts for 600 cores
a77c84d Merge pull request #131 from komatits/devel
47ecff9 start working on better OCL cleanups
1ae63dd update DATA submodule
724fa2d Fix array dimensions of movie volumes.
0cb72f2 Merge pull request #133 from QuLogic/fix-ifort
ba2c480 Fix make clean from subdirectories.
862ac9a Remove backup target.
4fc56a1 Correct license in several files.
7db6445 Clean up C code slightly.
4c0e111 Remove comment about s362ani + attenuation.
911ac0e Remove some commented and unused code.
a9fd39a Use real(..,kind=CUSTOM_REAL) for type conversion.
672c799 Use MPI_STATUS_IGNORE in MPI calls.
180110a Remove explicitly imported but unused module vars.
7607904 Fix real-to-integer conversion warnings.
3a2cf6a Make SAC type conversion explicit.
0d652ae Remove unused constants.
38469fd Fix spelling of some constants.
0d35ab5 Combine duplicate get_global subroutines.
2419de2 Combine duplicates of sort_array_coordinates.
55b14dc Replace Numerical Routines sorting code.
0705089 Remove get_value_string subroutine.
e504052 Normalize case of some Fortran keywords.
88864c7 Remove err_occurred function and commented code.
9fa2185 Remove unnecessary string trimming.
4c04d81 Remove Par_file variable prefixes.
8711be6 Small spelling fixes in comments.
f10c4ab Remove scripts that do nothing.
eefe3ac Change a misleading filename.
bc156ed Fix script executable bits.
968b09a Replace bubble sort with heap sort.
8222716 Remove specific subroutines from inlining on XLF.
5bec0e8 Fix warning about doubly-defined macro.
ab8a7ce Correct definition of Fortran preprocessor macros.
6099212 Fix definition of FORCE_VECTORIZATION macro.
ce0e08f Merge pull request #135 from QuLogic/fix-xlf
e0db75e added utils/Roland_Sylvain_gravity/display_g_vector_norm_SPECFEM3D_GLOBE_with_GMT_absolute.sh utils/Roland_Sylvain_gravity/display_g_vector_components_SPECFEM3D_GLOBE_with_GMT_absolute.sh utils/Roland_Sylvain_gravity/display_G_tensor_components_SPECFEM3D_GLOBE_with_GMT_absolute.sh
ad18112 Merge pull request #136 from komatits/devel
f12a92e Reduce DT for AK135F_NO_MUD.
fcaf7ee Merge branch 'cleanup' of github.com:QuLogic/specfem3d_globe into devel
35fef95 Fix model_attenuation sorting.
e5ae7c2 fix div by 0 bug: OpenCL gpu backend can't provide information about memory usage
a35d9a3 improve OCL cleanups
677d61e Merge branch 'devel' into opencl
6e100c2 last DATA commit
dceb26a cosmetics changes
0ddd37a revert to efficient version, without OCL kernel qualifiers
5525ec3 Merge pull request #137 from QuLogic/ak135-timestep
22e808a added a link to the developers' wiki
fe780c1 Merge pull request #140 from komatits/devel
c34e165 removed white spaces from the examples using a script, to reduce their size before we move them to GitHub; also suppressed the seismograms for s362ani + attenuation that were affected by the s362ani + Q bug, since the reference seismograms were wrong.
c02fafd Merge pull request #141 from komatits/devel
0b8177e removed a few old files in the DATA directory
2f567d8 Merge pull request #142 from komatits/devel
52d1523 merged compute_optimized_dumping_undo_att.f90 as a subroutine into the main code instead as a separate program
88c7688 Merge pull request #144 from komatits/devel
70b7397 fixed a declaration problem in src/meshfem3D/compute_volumes_and_areas.F90 when configuring with --disable-vectorization Also created empty directory DATABASES_MPI/, since it is required by the default example in DATA/Par_file
e3cd687 Merge pull request #145 from komatits/devel
6d7da3d Merge pull request #146 from Nanosim-LIG/opencl
c4e38d5 added comment about total memory size per core to undo attenuation on the TGCC Curie machi ne in Paris
23b1b22 Merge pull request #147 from komatits/devel
88ad6ff Remove submodule
1bc4138 Merge branch 'master' into devel
c672433 Fix ADIOS routine that was not compiling because of module changes
d7e975d Fix parallel compilation as a dependency was added
c285e06 par_file has options that were created by the opencl dev.
fa9148d moved ATTENUATION_1D_WITH_3D_STORAGE from the Par_file to setup/constants.h.in
10a92ce fixed a gfortran warning
c75d818 Merge pull request #155 from komatits/devel
d396c8a Re-update m4 submodule.
3cfb525 added utils/BOAST_framework_to_develop_the_CUDA_and_OpenCL_routines_of_SPECFEM/tutorials_BOAST_for_OpenCL_and_CUDA_programming_of_SPECFEM3D_kernels_Brice_Videau_Kevin_Pouget_Jean_Francois_Mehaut_jun2014.pdf
f002db5 added utils/BOAST_web_site_official_URL.txt
57fe9f3 Merge pull request #158 from komatits/devel
4be3008 moved a file
e70ef6f updated the URL of BOAST
d6f39eb put Matthieu Lefebvre's support for ADIOS_FOR_UNDO_ATTENUATION back (they had been lost when CIG got rid of the Git submodules)
bc58688 Merge pull request #161 from komatits/devel
b88f853 added a stop statement if the first time step entered by the user is negative
346cc77 added utils/trick_that_can_be_used_to_easily_create_weak_scaling_curves.txt
fd5af76 added the script download_the_whole_topography_database_if_you_want_other_topographic_models.bash
22de8b7 acknowledge Geza Seriani and Enrico Priolo in the users manual
dba0bc5 Add binary version of topography file.
df73113 Modify topography loader to use binary.
657aad2 Tell git to treat topography files as binary.
741ee67 Merge pull request #167 from QuLogic/binary-topography
6c2a930 Optimize ascii2bin script a bit.
0cc3848 Merge pull request #168 from QuLogic/faster-binary-conversion
ef85836 renamed DATA/topo_bathy/ascii2bin.py to DATA/topo_bathy/convert_etopo_files_from_specfem_ASCII_to_binary_topo_format_ascii2bin.py
bac1d1c improved a print statement in download_the_whole_topography_database_if_you_want_other_topographic_models.bash
01408eb fixed a print statement in download_the_whole_topography_database_if_you_want_other_topographic_models.bash
ffeb7b6 now that the s362ani bug has been fixed, re-added EXAMPLES/small_benchmark_run_to_test_more_complex_Earth with new (clean) reference seismograms
b79c294 added directory OUTPUT_FILES
b17a1fe Fix warning about BOZ constant outside of DATA.
4e3f2c6 added option EXACT_UNDOING_TO_DISK, which saves the forward run to disk in order to compute exact sensitivity kernels without having to undo the forward run numerically; it will read it back from disk instead. This is for validation purposes only, since the disk files will typically require tens of terabytes (yes, tera). However this will become affordable uin a few years.
32685b6 Merge remote-tracking branch 'origin/devel'
More information about the CIG-COMMITS
mailing list