Hi
Please consider joining the weekly Einstein Toolkit phone call at 10 am US central time on Mondays. As usual, you can find instructions how to join on the following web site:
http://einsteintoolkit.org/community/support/
In short: the number is (+1) 225-578-4942 or (+1) 866-573-0359 and the conference id is 118682#.
Some tickets within the last two weeks that are still open: - 778: Compiling PittNullCode is slow - 1648: simfactory default for <1 node jobs - 1648: MPI thorn should auto configure - 1650: piraha parses "-0." incorrectly (reviewed ok - but not closed) - 1618: Compile with adaptive MPI (new patches)
Other projects include - Make Illinois code work with standard ET - git transition
As always, please add to this list if you like.
Frank Loeffler
Minutes from the Einstein Toolkit Meeting on August 11, 2014.
Present:
Frank, Peter, Barry, Josef, Ian, Steve, Josh, Jonah.
Tickets:
778: PittNullCode is slow to compile. Frank is not sure what to do about it. According to Josef there is a constraint function that is slow. Josef usually does not compile this file. Ask Bela and or Roland about suggestions about what to do. Check if newer Intel compilers compiles it faster. Is it an option to remove it from ET.
?: Issue with using less than a full node in Simfactory. Ian suggest to postpone support until Simfactory3. Steve suggest a simple fix, but Ian thinks it will be to difficult. Simfactory3 needs some attention (Erik, Barry, Ian). One person is using it in production at the AEI for managing simulations. The code is rewritten and much more maintainable.
1648: Extract information from mpiCC to figure out how to compile Cactus with MPI. Steve wants to know if it works on system he hasn't tried. Try it on the usual ET release test machines. Steve will propose another patch to take into account comments from Ian.
1650: Patch is reviewed ok. Steve will commit and close.
1618: Steve looked at the patch and found that not all comments about previous has been addressed. Steve is talking to Jian and Geng Bin. Keep the ticket open.
1518: Patch for expression parsing to fix inconsistency in Cactus expression parsing. Standard test suites pass. Needs somebody to review the patch.
Git transition:
Almost final version of repositories are available. A few minor issues to beresolved.
1: Some arrangements have thorns that are not part of ET. 2: PUGHIO thorns have been moved into PUGH. 3: Some username mappings are missing. Frank will resolve. 4: Some thorns have changes associated with tags. This is not possible in git.
Barry invites people to try out the repositories. Barry has a new ET thornlist. Frank has given Barry commit right to manifest so he shoiuld commit. Barry will send a summary with instructions to the mailing list.
Cheers,
Peter
On Aug 11, 2014, at 11:47 , Peter Diener diener@cct.lsu.edu wrote:
Minutes from the Einstein Toolkit Meeting on August 11, 2014.
Present:
Frank, Peter, Barry, Josef, Ian, Steve, Josh, Jonah.
Tickets:
778: PittNullCode is slow to compile. Frank is not sure what to do about it. According to Josef there is a constraint function that is slow. Josef usually does not compile this file. Ask Bela and or Roland about suggestions about what to do. Check if newer Intel compilers compiles it faster. Is it an option to remove it from ET.
If I recall correctly, then this function is written in a whole-array style. Rewriting this with explicit index operations and surrounding it by a loop should improve compile time significantly.
-erik
On 12 Aug 2014, at 16:06, Erik Schnetter schnetter@cct.lsu.edu wrote:
On Aug 11, 2014, at 11:47 , Peter Diener diener@cct.lsu.edu wrote:
Minutes from the Einstein Toolkit Meeting on August 11, 2014.
Present:
Frank, Peter, Barry, Josef, Ian, Steve, Josh, Jonah.
Tickets:
778: PittNullCode is slow to compile. Frank is not sure what to do about it. According to Josef there is a constraint function that is slow. Josef usually does not compile this file. Ask Bela and or Roland about suggestions about what to do. Check if newer Intel compilers compiles it faster. Is it an option to remove it from ET.
If I recall correctly, then this function is written in a whole-array style. Rewriting this with explicit index operations and surrounding it by a loop should improve compile time significantly.
Roland already did this, and saw a speed improvement (see the ticket). I believe this improvement only helped with older compilers though. With 14.0.0, the file still takes 20 minutes to compile.
Present: Ian, Roland, Eloisa Bentivegna, Frank, Zach, Barry, Peter, Steve, Erik
Git transition: * Ian is testing the git repositories with Jenkins * need someone to test with GetComponents * commit messages will likely not work, will open up mailing list for bitbucket once first commit comes in * transition is ready to go, Frank has turned off svn commits and Barry will re-generate the git repositories * with the exception of simfactory we will also convert the "other" repositories https://docs.einsteintoolkit.org/et-docs/Repository_transition * Frank, Barry, Ian, Erik, Roland can add committers (and new admins) to bitbucket repos, add new committers as needed. Committers need bitbucket accounts.
MDH code: * vector potential code in ET: need science driver. Poll ET list for interest, approach main users of MHD code directly.
Yours, Roland
On Mon, Sep 22, 2014 at 12:42 PM, Roland Haas < roland.haas@physics.gatech.edu> wrote:
Git transition:
- Ian is testing the git repositories with Jenkins
- need someone to test with GetComponents
- commit messages will likely not work, will open up mailing list for
bitbucket once first commit comes in
- transition is ready to go, Frank has turned off svn commits and Barry
will re-generate the git repositories
- with the exception of simfactory we will also convert the "other"
repositories https://docs.einsteintoolkit.org/et-docs/Repository_transition
- Frank, Barry, Ian, Erik, Roland can add committers (and new admins) to
bitbucket repos, add new committers as needed. Committers need bitbucket accounts.
I have finished re-generating the new git repositories. They are available at https://bitbucket.org/cactuscode/ and https://bitbucket.org/einsteintoolkit/. The git_testing.th thornlist can be used to checkout the new repositories and can be downloaded from https://bitbucket.org/einsteintoolkit/manifest/raw/master/git_testing.th.
I have verified that I can checkout and build the Einstein Toolkit using GetComponents with this new thornlist. Ian has got Jenkins up and running using it (https://build.barrywardell.net/job/EinsteinToolkitGit/) and all tests currently pass. Commit messages also appear to be working and are forwarded to the commits@cactuscode.org and commits@einsteintoolkit.org mailing lists, respectively.
Please go ahead and try it out, and report any outstanding issues.
Barry
On Mon, Sep 22, 2014 at 11:21:43PM -0400, Barry Wardell wrote:
Please go ahead and try it out, and report any outstanding issues.
What is the status of Outflow? The still current svn thornlist lists its github repo as the one to be used, while the new git thornlist points (obviously) to the new git repo on bitbucket. Was the source/history from github used for the new repository?
Frank
Yes, the Outflow repository from GitHub was merged into the EinsteinAnalysis repository on BitBucket. I think Roland is planning to put a note in the GitHub repository pointing to the new location.
On Tue, Sep 23, 2014 at 2:42 PM, Frank Loeffler knarf@cct.lsu.edu wrote:
On Mon, Sep 22, 2014 at 11:21:43PM -0400, Barry Wardell wrote:
Please go ahead and try it out, and report any outstanding issues.
What is the status of Outflow? The still current svn thornlist lists its github repo as the one to be used, while the new git thornlist points (obviously) to the new git repo on bitbucket. Was the source/history from github used for the new repository?
Frank
users@lists.einsteintoolkit.org
-
Barry Wardell -
Erik Schnetter -
Frank Loeffler -
Ian Hinder -
Peter Diener -
Roland Haas