DKRZ projects: Issueshttps://swprojects.dkrz.de/redmine/https://swprojects.dkrz.de/redmine/redmine/favicon.ico?17095821032012-11-26T15:24:20ZDKRZ projects
Redmine YAXT - Feature #311 (Closed): stripify xt_xmap_dist_dir_newhttps://swprojects.dkrz.de/redmine/issues/3112012-11-26T15:24:20ZJoerg Behrensbehrens@dkrz.de
<p>Following the idea of using bounding boxes for xmap constructions we could consider to implement a stripe-set form of all involved ingredients.<br />This avoids vectorization to death of big compact index lists (better scalability). This is suitable for a one-dim distributed directory which might not scale perfectly but probably is more cost effective for medium sized problems.</p> YAXT - Bug #302 (Closed): Naminghttps://swprojects.dkrz.de/redmine/issues/3022012-04-23T12:36:25ZThomas Jahnsjahns@dkrz.de
<p>The names of xt_idxlist and xt_idxvec are the inverse of conventional nomenclature:</p>
<ul>
<li>a vector is an ordered list of elements, but also with random access</li>
<li>a list is an ordered collection of elements, potentially with duplications</li>
</ul>
<p>so I'll rename both.</p> ScalES-PPM - Task #300 (Closed): Use generalized rectilinear generation routine in sparse mask in...https://swprojects.dkrz.de/redmine/issues/3002012-02-13T12:16:28ZThomas Jahnsjahns@dkrz.deScalES-PPM - Bug #299 (Closed): Fix array passing in C/Fortran interface files from P*TYPE* to *T...https://swprojects.dkrz.de/redmine/issues/2992012-02-10T12:03:21ZThomas Jahnsjahns@dkrz.de
<p>Some routines pass arrays incorrectly from Fortran to C. This does currently yield correct results for all compilers tested, i.e. is not an acute problem, but should be fixed for clarity of semantics.</p> ScalES-PPM - Bug #284 (Closed): Fix MPI_Comm_f2c testhttps://swprojects.dkrz.de/redmine/issues/2842011-09-19T12:53:03ZThomas Jahnsjahns@dkrz.de
<p>The test does not account for MPI_Comm_f2c implemented as macro, i.e. mpi.h is not included.</p> Quick Enqueue - Feature #275 (Closed): Add default wall time limit argument to configure script.https://swprojects.dkrz.de/redmine/issues/2752011-06-03T16:10:35ZThomas Jahnsjahns@dkrz.deQuick Enqueue - Feature #274 (Closed): Add default SMT-setting argument to configure script.https://swprojects.dkrz.de/redmine/issues/2742011-06-03T15:58:14ZThomas Jahnsjahns@dkrz.deQuick Enqueue - Feature #273 (Closed): Add default memory-limit argument to configure script.https://swprojects.dkrz.de/redmine/issues/2732011-06-03T15:43:51ZThomas Jahnsjahns@dkrz.deQuick Enqueue - Feature #272 (Closed): Add default class argument to configure script.https://swprojects.dkrz.de/redmine/issues/2722011-06-03T15:42:32ZThomas Jahnsjahns@dkrz.deQuick Enqueue - Bug #271 (Closed): Catch SIGINT and cancel jobhttps://swprojects.dkrz.de/redmine/issues/2712011-06-03T14:12:42ZThomas Jahnsjahns@dkrz.de
<p>The job is currently not cancelled automatically, but this would be desirable on signal exit.</p> Quick Enqueue - Feature #270 (Closed): Improve site-configuration.https://swprojects.dkrz.de/redmine/issues/2702011-06-01T16:35:59ZThomas Jahnsjahns@dkrz.de
<p>Some aspects of per-site-configuration should be improved by providing optional switches to configure:</p>
<ul>
<li>The method for establishing the temporary directory, which must be available on submit and execution host, should be configurable.</li>
<li>The following should be configurable:
<ul>
<li>default class</li>
<li>default per-task memory limit</li>
<li>default smt-mode</li>
<li>default wall clock time limit</li>
</ul></li>
</ul> Quick Enqueue - Feature #268 (Closed): Let user choose the job-namehttps://swprojects.dkrz.de/redmine/issues/2682011-06-01T16:25:19ZThomas Jahnsjahns@dkrz.de
<p>Currently the job name is derived from the first argument. It should be adjustable via optional arguments to distinguish separate tasks running with the same command.</p> ScalES-PPM - Bug #261 (Closed): Check wether make dist-check workshttps://swprojects.dkrz.de/redmine/issues/2612011-03-24T16:25:28ZThomas Jahnsjahns@dkrz.deScalES-PPM - Task #259 (Closed): Decide on first public version numberhttps://swprojects.dkrz.de/redmine/issues/2592011-03-24T16:24:08ZThomas Jahnsjahns@dkrz.de
<p>A <a href="https://www.dkrz.de/redmine/doodles/1" class="external">doodle</a> has been setup.</p> ScalES-PPM - Feature #212 (Closed): Add cache database to dependency trackerhttps://swprojects.dkrz.de/redmine/issues/2122010-11-19T10:18:40ZThomas Jahnsjahns@dkrz.de
<p>On platforms where the Fortran compiler is slow (I'm looking at you xlf) to generate preprocessor output, it would be highly desirable to cache the dependency results, and not reparse the source code of unchanged files. Such an approach would have to account for several preconditions:</p>
<ul>
<li>The environment and flags between runs would ideally be unchanged, perhaps excluding known problem cases.</li>
<li>The source file in question must not have changed, as well as any file it includes.</li>
</ul>
<p>Please add further conditions to account for.</p>