Matches in SemOpenAlex for { <https://semopenalex.org/work/W3133241465> ?p ?o ?g. }
Showing items 1 to 91 of
91
with 100 items per page.
- W3133241465 endingPage "859" @default.
- W3133241465 startingPage "845" @default.
- W3133241465 abstract "Deep neural networks are widely used in personalized recommendation systems. Unlike regular DNN inference workloads, recommendation inference is memory-bound due to the many random memory accesses needed to lookup the embedding tables. The inference is also heavily constrained in terms of latency because producing a recommendation for a user must be done in about tens of milliseconds. In this paper, we propose MicroRec, a high-performance inference engine for recommendation systems. MicroRec accelerates recommendation inference by (1) redesigning the data structures involved in the embeddings to reduce the number of lookups needed and (2) taking advantage of the availability of High-Bandwidth Memory (HBM) in FPGA accelerators to tackle the latency by enabling parallel lookups. We have implemented the resulting design on an FPGA board including the embedding lookup step as well as the complete inference process. Compared to the optimized CPU baseline (16 vCPU, AVX2-enabled), MicroRec achieves 13.8~14.7x speedup on embedding lookup alone and 2.5$~5.4x speedup for the entire recommendation inference in terms of throughput. As for latency, CPU-based engines needs milliseconds for inferring a recommendation while MicroRec only takes microseconds, a significant advantage in real-time recommendation systems." @default.
- W3133241465 created "2021-03-01" @default.
- W3133241465 creator A5007322337 @default.
- W3133241465 creator A5015371403 @default.
- W3133241465 creator A5018202004 @default.
- W3133241465 creator A5032503782 @default.
- W3133241465 creator A5035131147 @default.
- W3133241465 creator A5041232755 @default.
- W3133241465 creator A5055093027 @default.
- W3133241465 creator A5057864403 @default.
- W3133241465 creator A5066481174 @default.
- W3133241465 creator A5069651430 @default.
- W3133241465 creator A5071396514 @default.
- W3133241465 creator A5071868053 @default.
- W3133241465 date "2021-03-15" @default.
- W3133241465 modified "2023-09-26" @default.
- W3133241465 title "MicroRec: Efficient Recommendation Inference by Hardware and Data Structure Solutions" @default.
- W3133241465 doi "https://doi.org/10.3929/ethz-b-000470540" @default.
- W3133241465 hasPublicationYear "2021" @default.
- W3133241465 type Work @default.
- W3133241465 sameAs 3133241465 @default.
- W3133241465 citedByCount "3" @default.
- W3133241465 countsByYear W31332414652020 @default.
- W3133241465 countsByYear W31332414652021 @default.
- W3133241465 crossrefType "journal-article" @default.
- W3133241465 hasAuthorship W3133241465A5007322337 @default.
- W3133241465 hasAuthorship W3133241465A5015371403 @default.
- W3133241465 hasAuthorship W3133241465A5018202004 @default.
- W3133241465 hasAuthorship W3133241465A5032503782 @default.
- W3133241465 hasAuthorship W3133241465A5035131147 @default.
- W3133241465 hasAuthorship W3133241465A5041232755 @default.
- W3133241465 hasAuthorship W3133241465A5055093027 @default.
- W3133241465 hasAuthorship W3133241465A5057864403 @default.
- W3133241465 hasAuthorship W3133241465A5066481174 @default.
- W3133241465 hasAuthorship W3133241465A5069651430 @default.
- W3133241465 hasAuthorship W3133241465A5071396514 @default.
- W3133241465 hasAuthorship W3133241465A5071868053 @default.
- W3133241465 hasConcept C119857082 @default.
- W3133241465 hasConcept C149635348 @default.
- W3133241465 hasConcept C154945302 @default.
- W3133241465 hasConcept C173608175 @default.
- W3133241465 hasConcept C2776214188 @default.
- W3133241465 hasConcept C41008148 @default.
- W3133241465 hasConcept C41608201 @default.
- W3133241465 hasConcept C42935608 @default.
- W3133241465 hasConcept C557471498 @default.
- W3133241465 hasConcept C68339613 @default.
- W3133241465 hasConcept C76155785 @default.
- W3133241465 hasConcept C82876162 @default.
- W3133241465 hasConceptScore W3133241465C119857082 @default.
- W3133241465 hasConceptScore W3133241465C149635348 @default.
- W3133241465 hasConceptScore W3133241465C154945302 @default.
- W3133241465 hasConceptScore W3133241465C173608175 @default.
- W3133241465 hasConceptScore W3133241465C2776214188 @default.
- W3133241465 hasConceptScore W3133241465C41008148 @default.
- W3133241465 hasConceptScore W3133241465C41608201 @default.
- W3133241465 hasConceptScore W3133241465C42935608 @default.
- W3133241465 hasConceptScore W3133241465C557471498 @default.
- W3133241465 hasConceptScore W3133241465C68339613 @default.
- W3133241465 hasConceptScore W3133241465C76155785 @default.
- W3133241465 hasConceptScore W3133241465C82876162 @default.
- W3133241465 hasLocation W31332414651 @default.
- W3133241465 hasOpenAccess W3133241465 @default.
- W3133241465 hasPrimaryLocation W31332414651 @default.
- W3133241465 hasRelatedWork W140566078 @default.
- W3133241465 hasRelatedWork W1991892717 @default.
- W3133241465 hasRelatedWork W2461383018 @default.
- W3133241465 hasRelatedWork W2510609312 @default.
- W3133241465 hasRelatedWork W2828086783 @default.
- W3133241465 hasRelatedWork W2887300277 @default.
- W3133241465 hasRelatedWork W2968989294 @default.
- W3133241465 hasRelatedWork W2982446375 @default.
- W3133241465 hasRelatedWork W2982507465 @default.
- W3133241465 hasRelatedWork W3039202310 @default.
- W3133241465 hasRelatedWork W3042267743 @default.
- W3133241465 hasRelatedWork W3082835045 @default.
- W3133241465 hasRelatedWork W3111747337 @default.
- W3133241465 hasRelatedWork W3131663603 @default.
- W3133241465 hasRelatedWork W3140099337 @default.
- W3133241465 hasRelatedWork W3147053158 @default.
- W3133241465 hasRelatedWork W3159558974 @default.
- W3133241465 hasRelatedWork W3160778734 @default.
- W3133241465 hasRelatedWork W3188694252 @default.
- W3133241465 hasRelatedWork W3201621211 @default.
- W3133241465 hasVolume "3" @default.
- W3133241465 isParatext "false" @default.
- W3133241465 isRetracted "false" @default.
- W3133241465 magId "3133241465" @default.
- W3133241465 workType "article" @default.