%0 Conference Paper %1 814703ca784d4ca78de96151d0e76152 %A Incardona, Pietro %A Gupta, Aryaman %A Yaskovets, Serhii %A Sbalzarini, Ivo F. %B Euro-Par 2022 %C Germany %D 2023 %E Singer, Jeremy %E Elkhatib, Yehia %E Blanco Heras, Dora %E Diehl, Patrick %E Brown, Nick %E Ilic, Aleksandar %I Springer Science and Business Media B.V. %K topic_lifescience C++ FIS_scads GPU, algorithms, generic layout, memory multi-core, performance portability tuples, %P 109--120 %R 10.1007/978-3-031-31209-0_8 %T A C++ Library for Memory Layout and Performance Portability of Scientific Applications %X We present a C++14 library for performance portability of scientific computing codes across CPU and GPU architectures. Our library combines generic data structures like vectors, multi-dimensional arrays, maps, graphs, and sparse grids with basic, reusable algorithms like convolutions, sorting, prefix sum, reductions, and scan. The memory layout of the data structures is adapted at compile-time using tuples with optional memory mirroring between CPU and GPU. We combine this transparent memory mapping with generic algorithms under two alternative programming interfaces: a CUDA-like kernel interface for multi-core CPUs, Nvidia GPUs, and AMD GPUs, as well as a lambda interface. We validate and benchmark the presented library using micro-benchmarks, showing that the abstractions introduce negligible performance overhead, and we compare performance against the current state of the art. %@ 9783031312083

@inproceedings{814703ca784d4ca78de96151d0e76152, abstract = {We present a C++14 library for performance portability of scientific computing codes across CPU and GPU architectures. Our library combines generic data structures like vectors, multi-dimensional arrays, maps, graphs, and sparse grids with basic, reusable algorithms like convolutions, sorting, prefix sum, reductions, and scan. The memory layout of the data structures is adapted at compile-time using tuples with optional memory mirroring between CPU and GPU. We combine this transparent memory mapping with generic algorithms under two alternative programming interfaces: a CUDA-like kernel interface for multi-core CPUs, Nvidia GPUs, and AMD GPUs, as well as a lambda interface. We validate and benchmark the presented library using micro-benchmarks, showing that the abstractions introduce negligible performance overhead, and we compare performance against the current state of the art.}, added-at = {2024-11-28T16:27:18.000+0100}, address = {Germany}, author = {Incardona, Pietro and Gupta, Aryaman and Yaskovets, Serhii and Sbalzarini, {Ivo F.}}, biburl = {https://puma.scadsai.uni-leipzig.de/bibtex/26ad8a430709195b480fd0d14a1a28291/scadsfct}, booktitle = {Euro-Par 2022}, doi = {10.1007/978-3-031-31209-0_8}, editor = {Singer, Jeremy and Elkhatib, Yehia and {Blanco Heras}, Dora and Diehl, Patrick and Brown, Nick and Ilic, Aleksandar}, interhash = {41a9bf1430b11b1516624991b701260d}, intrahash = {6ad8a430709195b480fd0d14a1a28291}, isbn = {9783031312083}, keywords = {topic_lifescience C++ FIS_scads GPU, algorithms, generic layout, memory multi-core, performance portability tuples,}, language = {English}, note = {Publisher Copyright: {\textcopyright} 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 28th International European Conference on Parallel and Distributed Computing , Euro-Par 2022 ; Conference date: 22-08-2022 Through 26-08-2022}, pages = {109--120}, publisher = {Springer Science and Business Media B.V.}, series = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}, timestamp = {2024-11-28T17:41:27.000+0100}, title = {A C++ Library for Memory Layout and Performance Portability of Scientific Applications}, year = 2023 }