Experiments for Predictable Execution of GPU Kernels
dc.contributor.author | Kreiliger F. | |
dc.contributor.author | Matějka J. | |
dc.contributor.author | Sojka M. | |
dc.contributor.author | Hanzálek Z. | |
dc.date.accessioned | 2020-02-12T12:24:57Z | |
dc.date.available | 2020-02-12T12:24:57Z | |
dc.date.issued | 2019 | |
dc.identifier | V3S-333028 | |
dc.identifier.citation | KREILIGER, F., et al. Experiments for Predictable Execution of GPU Kernels. In: PROCEEDINGS OF OSPERT 2019 the 15th Annual Workshop on Operating Systems Platforms for Embedded Real-Time Applications. the 15th Annual Workshop on Operating Systems Platforms for Embedded Real-Time Applications, Stuttgart, 2019-07-09. Dresden: TU Dresden, 2019. p. 23-28. Available from: https://ospert19.tudos.org/ospert19-proceedings.pdf | |
dc.identifier.uri | http://hdl.handle.net/10467/86755 | |
dc.description.abstract | Multi-Processor Systems-on-Chip (MPSoC) platforms will definitely power various future autonomous machines. Due to the high complexity of such platforms, it is difficult to achieve timing predictability, reliability and efficient resource utilization at the same time. We believe that time-triggered scheduling in combination with PRedictable Execution Model (PREM) can provide strong safety guarantees, and our longerterm goal is to schedule execution on the whole MPSoC (CPUs and GPU) in time triggered manner. To extend PREM to GPUs, we compare two synchronization mechanisms available on the NVIDIA Tegra X2 platform: one based on pinned memory and another that uses a GPU timer (socalled globaltimer). We found that running the NVIDIA profiler (nvprof) reconfigures the resolution of the globaltimer from 1 µs to 160 ns. By using time-triggered scheduling with such a resolution,itwaspossibletoreduceexecutiontimejitterofatiled 2D convolution kernel from 6.47% to 0.15% while maintaining the same average execution time. | eng |
dc.format.mimetype | application/pdf | |
dc.language.iso | eng | |
dc.publisher | TU Dresden | |
dc.relation.ispartof | PROCEEDINGS OF OSPERT 2019 the 15th Annual Workshop on Operating Systems Platforms for Embedded Real-Time Applications | |
dc.relation.uri | https://ospert19.tudos.org/ospert19-proceedings.pdf | |
dc.rights | Creative Commons Attribution (CC BY) 4.0 | |
dc.rights.uri | http://creativecommons.org/licenses/by/4.0/ | |
dc.subject | predictable execution | eng |
dc.subject | gpu | eng |
dc.subject | nvidia | eng |
dc.subject | tx2 | eng |
dc.subject | prem | eng |
dc.title | Experiments for Predictable Execution of GPU Kernels | eng |
dc.type | stať ve sborníku | cze |
dc.type | conference paper | eng |
dc.relation.projectid | info:eu-repo/grantAgreement/EC/H2020/832011/EU/Thermal-aware Resource Management for Modern Computing Platforms in the Next Generation of Aircraft/THERMAC | |
dc.rights.access | openAccess | |
dc.type.status | Peer-reviewed | |
dc.type.version | publishedVersion | |
dc.relation.conference | the 15th Annual Workshop on Operating Systems Platforms for Embedded Real-Time Applications |
Soubory tohoto záznamu
Tento záznam se objevuje v následujících kolekcích
-
Publikační činnost ČVUT [1323]
Kromě případů, kde je uvedeno jinak, licence tohoto záznamu je Creative Commons Attribution (CC BY) 4.0