Matches in SemOpenAlex for { <https://semopenalex.org/work/W4385488484> ?p ?o ?g. }
Showing items 1 to 89 of
89
with 100 items per page.
- W4385488484 abstract "Transformers' compute- intensive operations pose enormous challenges for their deployment in resource- constrained EdgeAI / tiny ML devices. As an established neural network compression technique, quantization reduces the hardware computational and memory resources. In particular, fixed-point quantization is desirable to ease the computations using lightweight blocks, like adders and multipliers, of the underlying hardware. However, deploying fully-quantized Transformers on existing general-purpose hardware, generic AI accelerators, or specialized architectures for Transformers with floating-point units might be infeasible and/or inefficient. Towards this, we propose SwiftTron, an efficient specialized hardware accelerator designed for Quantized Transformers. SwiftTron supports the execution of different types of Transformers' operations (like Attention, Softmax, GELU, and Layer Normalization) and accounts for diverse scaling factors to perform correct computations. We synthesize the complete SwiftTron architecture in a 65 nm CMOS technology with the ASIC design flow. Our Accelerator executes the RoBERTa-base model in 1.83 ns, while consuming 33.64 mW power, and occupying an area of 273 mm <sup xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>2</sup> • To ease the reproducibility, the RTL of our SwiftTron architecture is released at https://github.com/albertomarchisio/SwiftTron." @default.
- W4385488484 created "2023-08-03" @default.
- W4385488484 creator A5005190949 @default.
- W4385488484 creator A5026016005 @default.
- W4385488484 creator A5042642081 @default.
- W4385488484 creator A5074150684 @default.
- W4385488484 creator A5083674839 @default.
- W4385488484 creator A5090095861 @default.
- W4385488484 date "2023-06-18" @default.
- W4385488484 modified "2023-09-26" @default.
- W4385488484 title "SwiftTron: An Efficient Hardware Accelerator for Quantized Transformers" @default.
- W4385488484 cites W2143205602 @default.
- W4385488484 cites W2798919674 @default.
- W4385488484 cites W2974437627 @default.
- W4385488484 cites W2979826702 @default.
- W4385488484 cites W3017024317 @default.
- W4385488484 cites W3035083896 @default.
- W4385488484 cites W3035251378 @default.
- W4385488484 cites W3038988173 @default.
- W4385488484 cites W3047848469 @default.
- W4385488484 cites W3098873988 @default.
- W4385488484 cites W3108426037 @default.
- W4385488484 cites W3138516171 @default.
- W4385488484 cites W3184454880 @default.
- W4385488484 cites W3196923642 @default.
- W4385488484 cites W4200376271 @default.
- W4385488484 cites W4282981748 @default.
- W4385488484 cites W4308128267 @default.
- W4385488484 doi "https://doi.org/10.1109/ijcnn54540.2023.10191521" @default.
- W4385488484 hasPublicationYear "2023" @default.
- W4385488484 type Work @default.
- W4385488484 citedByCount "0" @default.
- W4385488484 crossrefType "proceedings-article" @default.
- W4385488484 hasAuthorship W4385488484A5005190949 @default.
- W4385488484 hasAuthorship W4385488484A5026016005 @default.
- W4385488484 hasAuthorship W4385488484A5042642081 @default.
- W4385488484 hasAuthorship W4385488484A5074150684 @default.
- W4385488484 hasAuthorship W4385488484A5083674839 @default.
- W4385488484 hasAuthorship W4385488484A5090095861 @default.
- W4385488484 hasBestOaLocation W43854884842 @default.
- W4385488484 hasConcept C11413529 @default.
- W4385488484 hasConcept C118524514 @default.
- W4385488484 hasConcept C119599485 @default.
- W4385488484 hasConcept C127413603 @default.
- W4385488484 hasConcept C149635348 @default.
- W4385488484 hasConcept C164620267 @default.
- W4385488484 hasConcept C165801399 @default.
- W4385488484 hasConcept C28855332 @default.
- W4385488484 hasConcept C41008148 @default.
- W4385488484 hasConcept C45374587 @default.
- W4385488484 hasConcept C46362747 @default.
- W4385488484 hasConcept C66322947 @default.
- W4385488484 hasConcept C76155785 @default.
- W4385488484 hasConcept C77390884 @default.
- W4385488484 hasConcept C82876162 @default.
- W4385488484 hasConcept C9390403 @default.
- W4385488484 hasConceptScore W4385488484C11413529 @default.
- W4385488484 hasConceptScore W4385488484C118524514 @default.
- W4385488484 hasConceptScore W4385488484C119599485 @default.
- W4385488484 hasConceptScore W4385488484C127413603 @default.
- W4385488484 hasConceptScore W4385488484C149635348 @default.
- W4385488484 hasConceptScore W4385488484C164620267 @default.
- W4385488484 hasConceptScore W4385488484C165801399 @default.
- W4385488484 hasConceptScore W4385488484C28855332 @default.
- W4385488484 hasConceptScore W4385488484C41008148 @default.
- W4385488484 hasConceptScore W4385488484C45374587 @default.
- W4385488484 hasConceptScore W4385488484C46362747 @default.
- W4385488484 hasConceptScore W4385488484C66322947 @default.
- W4385488484 hasConceptScore W4385488484C76155785 @default.
- W4385488484 hasConceptScore W4385488484C77390884 @default.
- W4385488484 hasConceptScore W4385488484C82876162 @default.
- W4385488484 hasConceptScore W4385488484C9390403 @default.
- W4385488484 hasLocation W43854884841 @default.
- W4385488484 hasLocation W43854884842 @default.
- W4385488484 hasOpenAccess W4385488484 @default.
- W4385488484 hasPrimaryLocation W43854884841 @default.
- W4385488484 hasRelatedWork W2105185821 @default.
- W4385488484 hasRelatedWork W2115844004 @default.
- W4385488484 hasRelatedWork W2118220389 @default.
- W4385488484 hasRelatedWork W2162220252 @default.
- W4385488484 hasRelatedWork W2526025708 @default.
- W4385488484 hasRelatedWork W2890159928 @default.
- W4385488484 hasRelatedWork W2908267042 @default.
- W4385488484 hasRelatedWork W3010492628 @default.
- W4385488484 hasRelatedWork W978089954 @default.
- W4385488484 hasRelatedWork W2506672464 @default.
- W4385488484 isParatext "false" @default.
- W4385488484 isRetracted "false" @default.
- W4385488484 workType "article" @default.