Commit message (Collapse) | Author | Age | |
---|---|---|---|
* | Rename new max_time_levels parameter as max_timelevels to be consistent with ↵ | Ian Hinder | 2013-10-15 |
| | | | | other uses of "timelevels" in parameter names in Carpet | ||
* | Carpet: remove duplicate myproc variable in SetupGH | Roland Haas | 2013-09-27 |
| | |||
* | Carpet: fix no_split_directions implementation | Roland Haas | 2013-09-27 |
| | | | | had forgotten a level of indirection | ||
* | Carpet: add new group tag no_split_directions | Roland Haas | 2013-09-27 |
| | |||
* | Carpet: correct wording of error message about number of time levels | Roland Haas | 2013-09-25 |
| | |||
* | Carpet: Initialise number of timelevels at the right time | Erik Schnetter | 2013-09-25 |
| | | | | Also clean up code structure for setting number of time levels. | ||
* | Carpet: Comment out unused routine | Erik Schnetter | 2013-09-25 |
| | |||
* | Carpet: Modernize error reporting | Erik Schnetter | 2013-09-25 |
| | |||
* | Carpet: Introduce parameter max_time_levels | Erik Schnetter | 2013-09-25 |
| | | | | This parameter defines the maximum number of time levels that a variable can have when being used for time prolongation. This allows using more time levels than prolongation_order_time+1. | ||
* | Timers: Move all timer-related code into a new thorn Timers | Erik Schnetter | 2013-08-07 |
| | |||
* | Carpet: Output list of compute nodes only in verbose mode | Erik Schnetter | 2013-05-26 |
| | |||
* | Carpet: Remove unused, broken code | Erik Schnetter | 2013-05-26 |
| | |||
* | Carpet: Do not restrict variables marked as "don't restrict" | Erik Schnetter | 2013-04-04 |
| | | | | | Do not restrict variables marked as op_none or op_sync. Mark grid arrays as op_sync, not op_copy, which would still prolongate and restrict them. | ||
* | Revert part of "Requirements: Beautify code" | Ian Hinder | 2013-04-04 |
| | | | | | This reverts part of commit 40f113decf0d0de18aeabb8cce90bdc835f56fdd which looks like it was committed accidentally and causes tests to fail. | ||
* | Carpet, Requirements: Pass timelevel_offset | Erik Schnetter | 2013-04-03 |
| | | | | | | Introduce global variable timelevel_offset in Carpet. Pass timelevel_offset from Carpet to Requirements. Use timelevel_offset when determining which timelevels are accessed. | ||
* | Requirements: Beautify code | Erik Schnetter | 2013-04-03 |
| | |||
* | Replace abs() by fabs() | Erik Schnetter | 2013-04-02 |
| | | | | Replace calls to abs() with floating arguments by calls to fabs(), to ensure the results are not accidentally truncated to integer. | ||
* | Requirements: New thorn | Erik Schnetter | 2013-04-02 |
| | | | | Move requirement handling code out of Carpet and into a new thorn. | ||
* | Carpet: Beautify screen output | Erik Schnetter | 2013-02-08 |
| | |||
* | Carpet: Avoid unused and uninitialised variables | Erik Schnetter | 2012-11-25 |
| | |||
* | CarpetLib: add volume averaged cc eno operator | Roland Haas | 2012-11-20 |
| | |||
* | Introduce parameter "additional_overlap_zones" to control the amount of ↵ | Christian Reisswig | 2012-09-26 |
| | | | | | | | | additional (user-requested) overlap zones. This is similar to additional_buffer_zones. Also fixed a bug in reporting the number of overlap zones in stdout. ("buffers" instead of "overlaps") was used. | ||
* | Carpet: Check schedule reads/writes statements at run time | Erik Schnetter | 2012-09-11 |
| | |||
* | Carpet: Move CCTK_INFO calls out of OpenMP parallel region | Erik Schnetter | 2012-09-11 |
| | |||
* | Carpet: Introduce more timers to measure communication time | Erik Schnetter | 2012-09-11 |
| | |||
* | Introduce cctk_ash, retire cctk_lssh | Erik Schnetter | 2012-09-11 |
| | | | | | | | | | Introduce cctk_ash, describing the process-local array shape that has been allocated. This may be larger than cctk_lsh, the process-local shape that should be used. Retire cctk_lssh and related infrastructure to handle staggered grid functions. | ||
* | Carpet: Output informative message if HAVE_SCHED_GETAFFINITY is missing | Erik Schnetter | 2012-09-11 |
| | |||
* | Carpet: output grid structures to file in Initialize instead of in SetupGH | Roland Haas | 2012-09-11 |
| | | | | | This is so that IOUtils has had time to initialize itself and we can query IO_TruncateOutputFiles with confidence. | ||
* | Carpet: Update setting CPU affinity | Erik Schnetter | 2012-09-11 |
| | |||
* | Carpet: check ghost size against restrion operator order | Roland Haas | 2012-09-11 |
| | | | | if use_higher_order_restriction = yes | ||
* | Carpet…: Rename buffer2 to overlap zones; correct implementation in dh | Erik Schnetter | 2012-09-11 |
| | | | | | | Rename buffer2 zones to overlap zones. Complete implementing overlap zones in the dh class. | ||
* | Carpet: Correct setting core affinity | Erik Schnetter | 2012-09-11 |
| | | | | Correct upper loop bound | ||
* | Carpet: Correct condition for using buffer2 zones | Erik Schnetter | 2012-09-11 |
| | |||
* | Carpet: Correct error in thread id handling | Erik Schnetter | 2012-09-11 |
| | |||
* | Carpet: Remove debug output | Erik Schnetter | 2012-09-11 |
| | |||
* | Carpet: Set CPU affinity correctly for multiple MPI processes per host | Erik Schnetter | 2012-09-11 |
| | |||
* | Carpet: get thread number from CarpetLib's dist object | Roland Haas | 2012-09-11 |
| | |||
* | Backed out changeset 74ddd545dbb1 | Roland Haas | 2012-09-11 |
| | | | | | I give up. I cannot get this right today. The previous patch would have allowed a thread to switch sockets which is undesirable. | ||
* | Carpet: let thread run on any core [n0,n0+num_threads) instead of binding to | Roland Haas | 2012-09-11 |
| | | | | | | | single core This is the behaviour of eg. Kraken's aprun --cc numa_node. Avoids using omp_get_thread_num(). | ||
* | Backed out changeset 27a049a40a81 | Roland Haas | 2012-09-11 |
| | | | | | sigh, there _is_ a difference between omp_get_num_threads() and omp_get_thread_num() after all. | ||
* | Carpet: replace omp_get_num_threads() by mynthreads to compile without OpenMP | Roland Haas | 2012-09-11 |
| | |||
* | Carpet: Check CPU affinity per thread, not per process | Erik Schnetter | 2012-09-11 |
| | |||
* | Carpet: Don't warn when there are more cores then threads | Erik Schnetter | 2012-09-11 |
| | |||
* | Carpet: Don't use CPU_COUNT | Erik Schnetter | 2012-09-11 |
| | |||
* | Carpet: Add parameter set_cpuaffinity | Erik Schnetter | 2012-09-11 |
| | |||
* | Carpet: Correct representation of host and model names | Erik Schnetter | 2012-09-11 |
| | |||
* | Carpet: Initialise Carpet::in_analysis_bin | Erik Schnetter | 2012-09-11 |
| | |||
* | Carpet: Check number of processes and threads on startup | Erik Schnetter | 2012-09-11 |
| | | | | Compare number of MPI processes and OpenMP threads to environment variables CACTUS_NUM_PROCS and CACTUS_NUM_THREADS. | ||
* | Merge | Erik Schnetter | 2012-09-11 |
|\ | |||
| * | Re-organise some refluxing internals | Erik Schnetter | 2012-09-11 |
| | | | | | | | | | | | | | | | | | | | | | | Re-organise some of the internal details of refluxing. Refluxing requires restricting fluxes from fine to coarse grids. Previously, Carpet would internally store bounding boxes that were offset by 1/2 grid point, and adjust (correct) these boxes in various places in a rather ad-hoc manner. This is now cleaned up. Remove the (unused) "accumulate" operator, and the (unused) "accumulate" prolongation type. |