Matches in SemOpenAlex for { <https://semopenalex.org/work/W4383219896> ?p ?o ?g. }
Showing items 1 to 66 of
66
with 100 items per page.
- W4383219896 abstract "Specialized processors designed to accelerate tensor operations are evolving faster than conventional processors. This trend of architectural innovations greatly benefits artificial intelligence (AI) workloads. However, it is unknown how well AI-optimized accelerators can be retargeted to scientific applications. To answer this question we explore (1) whether a typical scientific modeling kernel can be mapped efficiently to tensor operations and (2) whether this approach is portable across diverse processors and AI accelerators. In this paper we implement two versions of tracer advection in an ocean-modeling application using PyTorch and evaluate these on one CPU, two GPUs, and Google's TPU. Our findings are that scientific modeling can observe both a performance boost and improved portability by mapping key computational kernels to tensor operations." @default.
- W4383219896 created "2023-07-06" @default.
- W4383219896 creator A5024805727 @default.
- W4383219896 creator A5039337781 @default.
- W4383219896 creator A5060855642 @default.
- W4383219896 date "2023-02-25" @default.
- W4383219896 modified "2023-09-27" @default.
- W4383219896 title "Harnessing Extreme Heterogeneity for Ocean Modeling with Tensors" @default.
- W4383219896 cites W2020699102 @default.
- W4383219896 cites W2902076893 @default.
- W4383219896 cites W2921025948 @default.
- W4383219896 cites W2977970870 @default.
- W4383219896 cites W3100649155 @default.
- W4383219896 doi "https://doi.org/10.1145/3587278.3595645" @default.
- W4383219896 hasPublicationYear "2023" @default.
- W4383219896 type Work @default.
- W4383219896 citedByCount "0" @default.
- W4383219896 crossrefType "proceedings-article" @default.
- W4383219896 hasAuthorship W4383219896A5024805727 @default.
- W4383219896 hasAuthorship W4383219896A5039337781 @default.
- W4383219896 hasAuthorship W4383219896A5060855642 @default.
- W4383219896 hasBestOaLocation W43832198961 @default.
- W4383219896 hasConcept C111919701 @default.
- W4383219896 hasConcept C113775141 @default.
- W4383219896 hasConcept C114614502 @default.
- W4383219896 hasConcept C154945302 @default.
- W4383219896 hasConcept C155281189 @default.
- W4383219896 hasConcept C173608175 @default.
- W4383219896 hasConcept C199360897 @default.
- W4383219896 hasConcept C202444582 @default.
- W4383219896 hasConcept C26517878 @default.
- W4383219896 hasConcept C33923547 @default.
- W4383219896 hasConcept C41008148 @default.
- W4383219896 hasConcept C459310 @default.
- W4383219896 hasConcept C63000827 @default.
- W4383219896 hasConcept C74193536 @default.
- W4383219896 hasConceptScore W4383219896C111919701 @default.
- W4383219896 hasConceptScore W4383219896C113775141 @default.
- W4383219896 hasConceptScore W4383219896C114614502 @default.
- W4383219896 hasConceptScore W4383219896C154945302 @default.
- W4383219896 hasConceptScore W4383219896C155281189 @default.
- W4383219896 hasConceptScore W4383219896C173608175 @default.
- W4383219896 hasConceptScore W4383219896C199360897 @default.
- W4383219896 hasConceptScore W4383219896C202444582 @default.
- W4383219896 hasConceptScore W4383219896C26517878 @default.
- W4383219896 hasConceptScore W4383219896C33923547 @default.
- W4383219896 hasConceptScore W4383219896C41008148 @default.
- W4383219896 hasConceptScore W4383219896C459310 @default.
- W4383219896 hasConceptScore W4383219896C63000827 @default.
- W4383219896 hasConceptScore W4383219896C74193536 @default.
- W4383219896 hasLocation W43832198961 @default.
- W4383219896 hasOpenAccess W4383219896 @default.
- W4383219896 hasPrimaryLocation W43832198961 @default.
- W4383219896 hasRelatedWork W1574423241 @default.
- W4383219896 hasRelatedWork W1832263773 @default.
- W4383219896 hasRelatedWork W2025462778 @default.
- W4383219896 hasRelatedWork W2117014006 @default.
- W4383219896 hasRelatedWork W2624440775 @default.
- W4383219896 hasRelatedWork W3198509963 @default.
- W4383219896 hasRelatedWork W4243260870 @default.
- W4383219896 hasRelatedWork W4250047567 @default.
- W4383219896 hasRelatedWork W4316035610 @default.
- W4383219896 hasRelatedWork W792983 @default.
- W4383219896 isParatext "false" @default.
- W4383219896 isRetracted "false" @default.
- W4383219896 workType "article" @default.