summaryrefslogtreecommitdiffstats
path: root/src/CompositeGeometryManager.cpp
Commit message (Collapse)AuthorAgeFilesLines
* Move CUDA include filesWillem Jan Palenstijn2018-06-221-2/+2
|
* iMinds -> imecWillem Jan Palenstijn2018-01-241-1/+1
|
* Update copyright dateWillem Jan Palenstijn2018-01-241-2/+2
|
* Update contact emailWillem Jan Palenstijn2018-01-121-1/+1
|
* Check for CUDA support at run-time in use_cuda()Willem Jan Palenstijn2018-01-091-1/+2
|
* Allow cone_vec for FDKWillem Jan Palenstijn2017-11-221-2/+3
| | | | | | NB: This will let you do mathematically nonsensical things, since FDK does not support non-circular cone geometries. However, this does let you use translated detectors, for example.
* Add experimental accumulate_FP/BP/FDK python functionsWillem Jan Palenstijn2017-11-221-17/+19
|
* Fix output initialization for MODE_ADD CGM jobsWillem Jan Palenstijn2017-10-191-0/+6
|
* Fix warningsWillem Jan Palenstijn2017-02-081-8/+9
|
* Start work on CFloat32Data3DGPU to allow persistent/external GPU memoryWillem Jan Palenstijn2017-02-081-43/+217
|
* Expose the density weighting option of cone_bpWillem Jan Palenstijn2016-12-011-1/+3
| | | | It is now exposed via the new DensityWeighting option of CudaProjector3D.
* Update headers (website+2016)Willem Jan Palenstijn2016-11-281-3/+3
|
* Merge branch 'master' into fdk_custom_filterWillem Jan Palenstijn2016-10-131-140/+233
|\
| * Merge branch 'master' into FDKWillem Jan Palenstijn2016-10-061-137/+184
| |\
| | * Fix inefficient block split logicWillem Jan Palenstijn2016-06-201-5/+19
| | |
| | * Improve volume block reductionWillem Jan Palenstijn2016-06-201-132/+165
| |/ |/| | | | | | | The previous version would make the blocks too large due to inefficient computation of overlap.
| * Use CompositeGeometryManager for FDKWillem Jan Palenstijn2016-04-181-3/+47
|/ | | | | Also fix a number of scaling/weighting issues in FDK, and switch to standard cone_bp with FDKWeighting for the BP step.
* Fix buildWillem Jan Palenstijn2016-03-011-6/+6
|
* Query max texture size instead of hardcoding itWillem Jan Palenstijn2016-02-161-6/+6
|
* Avoid (unlikely) integer overflowWillem Jan Palenstijn2016-02-151-1/+5
|
* Skip some unnecessary splittingWillem Jan Palenstijn2016-02-151-5/+8
|
* Also split volumes in X/Y directions to respect CUDA limitsWillem Jan Palenstijn2016-02-151-21/+240
|
* Refactor FP and BP jobs creation in the composite geometry managerNicola Vigano2016-02-101-14/+25
|
* Add multi-GPU support to CompositeGeometryManagerWillem Jan Palenstijn2016-01-191-114/+320
|
* Fix whitespaceWillem Jan Palenstijn2016-01-191-21/+21
|
* Fix projections parallel to XZ or YZ planesWillem Jan Palenstijn2016-01-141-0/+12
| | | | | The early-cutoff optimization in CompositeGeometryManager was failing to properly handle +/-Inf.
* Add utility functions for creating FP/BP JobListsWillem Jan Palenstijn2015-12-041-2/+111
|
* Add CompositeGeometryManagerWillem Jan Palenstijn2015-12-041-0/+884
This handles FP and BP operations on multiple data objects at once, splitting them to fit in GPU memory where necessary.