Matches in SemOpenAlex for { <https://semopenalex.org/work/W4205546691> ?p ?o ?g. }
- W4205546691 endingPage "168184" @default.
- W4205546691 startingPage "168162" @default.
- W4205546691 abstract "Convolution is the most time-consuming operation in modern deep artificial neural networks, so its performance is crucial for fast inference. One of the standard approaches to fast convolution computation is to use GeMM-based convolution algorithms relying on efficient general matrix multiplication (GeMM) from optimized BLAS libraries. However, commonly used GeMM-based algorithms may cause significant memory overhead or avoid it only at the cost of worse performance. In this paper, we propose a novel convolution algorithm, p-im2col, based on a well-known im2col algorithm that avoids memory overhead by splitting a single multiplication of a large matrix into several multiplications of smaller matrices. We theoretically and experimentally compare our algorithm with two other GeMM-based algorithms: im2col, which is widely used as a baseline, and the memory-efficient kn2row-aa. We measure the inference time of these algorithms on central processing units of x86, x86_64, ARM, and MIPS architectures for a large set of convolutional parameters. The proposed algorithm demonstrates a speedup over im2col and kn2row-aa in a number of cases and a significant reduction in additional memory requirements compared to im2col. Based on our experiments, we present a new convolution algorithm selection scheme that considers memory restrictions, CPU architecture, and convolutional parameters and provides a noticeable advantage over each particular algorithm." @default.
- W4205546691 created "2022-01-25" @default.
- W4205546691 creator A5019711919 @default.
- W4205546691 creator A5047135094 @default.
- W4205546691 creator A5066631478 @default.
- W4205546691 creator A5077963262 @default.
- W4205546691 date "2021-01-01" @default.
- W4205546691 modified "2023-10-16" @default.
- W4205546691 title "p-im2col: Simple Yet Efficient Convolution Algorithm With Flexibly Controlled Memory Overhead" @default.
- W4205546691 cites W2073061372 @default.
- W4205546691 cites W2107356030 @default.
- W4205546691 cites W2139774022 @default.
- W4205546691 cites W2194775991 @default.
- W4205546691 cites W2519653196 @default.
- W4205546691 cites W2531409750 @default.
- W4205546691 cites W2600746833 @default.
- W4205546691 cites W2734572653 @default.
- W4205546691 cites W2738749209 @default.
- W4205546691 cites W2792311073 @default.
- W4205546691 cites W2793947836 @default.
- W4205546691 cites W2891540872 @default.
- W4205546691 cites W2902446117 @default.
- W4205546691 cites W2922260520 @default.
- W4205546691 cites W2951894856 @default.
- W4205546691 cites W2963037989 @default.
- W4205546691 cites W2963163009 @default.
- W4205546691 cites W2963809228 @default.
- W4205546691 cites W2972235195 @default.
- W4205546691 cites W2988314300 @default.
- W4205546691 cites W2990467641 @default.
- W4205546691 cites W3018105153 @default.
- W4205546691 cites W3022079201 @default.
- W4205546691 cites W3034344052 @default.
- W4205546691 cites W3041293089 @default.
- W4205546691 cites W3093756531 @default.
- W4205546691 cites W3119260092 @default.
- W4205546691 cites W3119889735 @default.
- W4205546691 cites W3157788795 @default.
- W4205546691 cites W3161273196 @default.
- W4205546691 cites W3164506637 @default.
- W4205546691 cites W3168636294 @default.
- W4205546691 cites W3183407731 @default.
- W4205546691 cites W3185437734 @default.
- W4205546691 cites W4287633886 @default.
- W4205546691 cites W639708223 @default.
- W4205546691 doi "https://doi.org/10.1109/access.2021.3135690" @default.
- W4205546691 hasPublicationYear "2021" @default.
- W4205546691 type Work @default.
- W4205546691 citedByCount "3" @default.
- W4205546691 countsByYear W42055466912022 @default.
- W4205546691 countsByYear W42055466912023 @default.
- W4205546691 crossrefType "journal-article" @default.
- W4205546691 hasAuthorship W4205546691A5019711919 @default.
- W4205546691 hasAuthorship W4205546691A5047135094 @default.
- W4205546691 hasAuthorship W4205546691A5066631478 @default.
- W4205546691 hasAuthorship W4205546691A5077963262 @default.
- W4205546691 hasBestOaLocation W42055466911 @default.
- W4205546691 hasConcept C111335779 @default.
- W4205546691 hasConcept C111919701 @default.
- W4205546691 hasConcept C11413529 @default.
- W4205546691 hasConcept C114614502 @default.
- W4205546691 hasConcept C121332964 @default.
- W4205546691 hasConcept C154945302 @default.
- W4205546691 hasConcept C170723468 @default.
- W4205546691 hasConcept C17349429 @default.
- W4205546691 hasConcept C173608175 @default.
- W4205546691 hasConcept C199360897 @default.
- W4205546691 hasConcept C2524010 @default.
- W4205546691 hasConcept C2777904410 @default.
- W4205546691 hasConcept C2779960059 @default.
- W4205546691 hasConcept C2780595030 @default.
- W4205546691 hasConcept C33923547 @default.
- W4205546691 hasConcept C41008148 @default.
- W4205546691 hasConcept C45347329 @default.
- W4205546691 hasConcept C50644808 @default.
- W4205546691 hasConcept C62520636 @default.
- W4205546691 hasConcept C68339613 @default.
- W4205546691 hasConcept C81363708 @default.
- W4205546691 hasConcept C84114770 @default.
- W4205546691 hasConceptScore W4205546691C111335779 @default.
- W4205546691 hasConceptScore W4205546691C111919701 @default.
- W4205546691 hasConceptScore W4205546691C11413529 @default.
- W4205546691 hasConceptScore W4205546691C114614502 @default.
- W4205546691 hasConceptScore W4205546691C121332964 @default.
- W4205546691 hasConceptScore W4205546691C154945302 @default.
- W4205546691 hasConceptScore W4205546691C170723468 @default.
- W4205546691 hasConceptScore W4205546691C17349429 @default.
- W4205546691 hasConceptScore W4205546691C173608175 @default.
- W4205546691 hasConceptScore W4205546691C199360897 @default.
- W4205546691 hasConceptScore W4205546691C2524010 @default.
- W4205546691 hasConceptScore W4205546691C2777904410 @default.
- W4205546691 hasConceptScore W4205546691C2779960059 @default.
- W4205546691 hasConceptScore W4205546691C2780595030 @default.
- W4205546691 hasConceptScore W4205546691C33923547 @default.
- W4205546691 hasConceptScore W4205546691C41008148 @default.
- W4205546691 hasConceptScore W4205546691C45347329 @default.
- W4205546691 hasConceptScore W4205546691C50644808 @default.
- W4205546691 hasConceptScore W4205546691C62520636 @default.