000014823 001__ 14823
000014823 005__ 20161115100203.0
000014823 04107 $$aeng
000014823 046__ $$k2016-08-21
000014823 100__ $$aKonduri, Aditya
000014823 24500 $$aImplementation of exascale asynchronous solvers for PDEs

000014823 24630 $$n24.$$p24th International Congress of Theoretical and Applied Mechanics - Book of Papers
000014823 260__ $$bInternational Union of Theoretical and Applied Mechanics, 2016
000014823 506__ $$arestricted
000014823 520__ $$2eng$$aTo overcome the challenges likely to be present in exascale systems we recently developed novel asynchrony-tolerant numerical schemes for partial differential equations which relax the need for synchronization between processing elements at a mathematical level. The advantages of such schemes can only be realized if their implementation also aims at alleviating the limiting factors at exascale in terms of hardware, software, and programing frameworks. We present two such implementations with different degrees of asynchrony and their corresponding tradeoffs between numerical accuracy and performance. Analysis and numerical experiments demonstrate their respective potential to increase scalability of solvers in future exascale systems.

000014823 540__ $$aText je chráněný podle autorského zákona č. 121/2000 Sb.
000014823 653__ $$a

000014823 7112_ $$a24th International Congress of Theoretical and Applied Mechanics$$cMontreal (CA)$$d2016-08-21 / 2016-08-26$$gICTAM2016
000014823 720__ $$aKonduri, Aditya
000014823 8560_ $$ffischerc@itam.cas.cz
000014823 8564_ $$s71482$$uhttps://invenio.itam.cas.cz/record/14823/files/TS.FS02-1.05.pdf$$yOriginal version of the author's contribution as presented on CD,  page 3102, code TS.FS02-1.05
.
000014823 962__ $$r13812
000014823 980__ $$aPAPER