Difference between revisions of "CamCASP/Releases"
Jump to navigation
Jump to search
import>Am592 |
import>Am592 |
||
| (3 intermediate revisions by the same user not shown) | |||
| Line 2: | Line 2: | ||
==Branches== |
==Branches== |
||
| + | ===5.6 Planned=== |
||
| + | Not yet branched. Still in Trunk/ |
||
| + | |||
| + | * Lillistollen and Wheatley's Stockholder partitioning |
||
| + | * Fifth row atomic basis sets. Decisions need to be made about these as there aren't so many options |
||
| + | * More and better scripts to aid potential development. |
||
| + | --[[User:am592|alston]] 12:48, 16 February 2011 (GMT) |
||
| + | |||
===5.5-dev=== |
===5.5-dev=== |
||
Branched from trunk on the 25-Mar-10. |
Branched from trunk on the 25-Mar-10. |
||
| Line 11: | Line 19: | ||
* New propagator module: Implements the ALDA(X) kernel in the manner described in Bukowski et al. CPL 414, 111(2005). |
* New propagator module: Implements the ALDA(X) kernel in the manner described in Bukowski et al. CPL 414, 111(2005). |
||
| + | ** There are more options in this module. |
||
| + | ** The PBE/AC ALDAX route is the fastest and consumes the least memory. |
||
| + | ** PBE0/AC ALDAX+CHF is fast, but memory intensive. |
||
| + | ** The old propagator module can still be accessed. These deliver the highest accuracy, but are slow and require integrals from DALTON. |
||
* 10-fold speed-up in calculating 2 and 3-centre Coulomb integrals. |
* 10-fold speed-up in calculating 2 and 3-centre Coulomb integrals. |
||
* Revision 21041 compiles with ifort. Binary is almost twice as fast as with pgf90. |
* Revision 21041 compiles with ifort. Binary is almost twice as fast as with pgf90. |
||
| + | |||
| + | * Massive improvements to the speed of the distribution algorithm. Improvements by two orders of magnitude. |
||
| + | |||
| + | * Better scripts for easier use of the code. |
||
| + | |||
| + | * Energy scan for first order energies a factor of 10 faster. |
||
| + | |||
| + | ==Creating a TarBall== |
||
| + | Here are instructions from Anthony about creating a tar-ball: |
||
| + | # You will need a directory ${HOME}/www/restrict/camcasp/src/ (you can delete it afterwards.) The source tarball will be put there. (The www directory is my copy of the website.) |
||
| + | # You also need to edit camcasp/distrib/source_files to include any new files. Include the camcasp-5.6/ in the name. (You don't need to change that, even if your camcasp path is something else. In fact you mustn't change it.) (I don't use e.g. camcasp-5.6/src/*90 because that would include unwanted stuff.) |
||
| + | # Then chdir to camcasp/distrib and run |
||
| + | <pre> |
||
| + | make source VERSION=5.6 PATCHLEVEL=00 |
||
| + | </pre> |
||
| + | |||
| + | This creates the SOURCE tarball. We also need a tarball of the BASE (basis sets, examples, etc.). To do this use: |
||
| + | <pre> |
||
| + | make base-only VERSION=5.6 PATCHLEVEL=00 |
||
| + | </pre> |
||
| + | I've modified the make file (trivially) to allow making the BASE only without trying to upload things to the web server. To do that, use |
||
| + | <pre> |
||
| + | make base VERSION=5.6 PATCHLEVEL=00 |
||
| + | </pre> |
||
| + | --[[User:am592|alston]] 16:41, 23 February 2011 (GMT) |
||
Latest revision as of 16:41, 23 February 2011
CamCASP => Releases and Branches
Branches
5.6 Planned
Not yet branched. Still in Trunk/
- Lillistollen and Wheatley's Stockholder partitioning
- Fifth row atomic basis sets. Decisions need to be made about these as there aren't so many options
- More and better scripts to aid potential development.
--alston 12:48, 16 February 2011 (GMT)
5.5-dev
Branched from trunk on the 25-Mar-10.
- New program: res2disp added. Source code in src/res2disp/
Calculates dispersion energies of crystals using isotropic dispersion models.
- Kernel integrals constructed in CamCASP.
- New propagator module: Implements the ALDA(X) kernel in the manner described in Bukowski et al. CPL 414, 111(2005).
- There are more options in this module.
- The PBE/AC ALDAX route is the fastest and consumes the least memory.
- PBE0/AC ALDAX+CHF is fast, but memory intensive.
- The old propagator module can still be accessed. These deliver the highest accuracy, but are slow and require integrals from DALTON.
- 10-fold speed-up in calculating 2 and 3-centre Coulomb integrals.
- Revision 21041 compiles with ifort. Binary is almost twice as fast as with pgf90.
- Massive improvements to the speed of the distribution algorithm. Improvements by two orders of magnitude.
- Better scripts for easier use of the code.
- Energy scan for first order energies a factor of 10 faster.
Creating a TarBall
Here are instructions from Anthony about creating a tar-ball:
- You will need a directory ${HOME}/www/restrict/camcasp/src/ (you can delete it afterwards.) The source tarball will be put there. (The www directory is my copy of the website.)
- You also need to edit camcasp/distrib/source_files to include any new files. Include the camcasp-5.6/ in the name. (You don't need to change that, even if your camcasp path is something else. In fact you mustn't change it.) (I don't use e.g. camcasp-5.6/src/*90 because that would include unwanted stuff.)
- Then chdir to camcasp/distrib and run
make source VERSION=5.6 PATCHLEVEL=00
This creates the SOURCE tarball. We also need a tarball of the BASE (basis sets, examples, etc.). To do this use:
make base-only VERSION=5.6 PATCHLEVEL=00
I've modified the make file (trivially) to allow making the BASE only without trying to upload things to the web server. To do that, use
make base VERSION=5.6 PATCHLEVEL=00
--alston 16:41, 23 February 2011 (GMT)