A digital SRAM-based compute-in-memory macro for weight-stationary dynamic matrix multiplication in Transformer attention score computation Article Swipe
Yu JianYi
,
Wang Yuxuan
,
Fu Xiang
,
Qiao Fei
,
Wang, Ying
,
Yuan Rui
,
Liu Liyuan
,
Shi Cong
·
YOU?
·
· 2025
· Open Access
·
YOU?
·
· 2025
· Open Access
·
Compute-in-memory (CIM) techniques are widely employed in energy-efficient artificial intelligent (AI) processors. They alleviate power and latency bottlenecks caused by extensive data movements between compute and storage units. This work proposes a digital CIM macro to compute Transformer attention. To mitigate dynamic matrix multiplication that is unsuitable for the common weight-stationary CIM paradigm, we reformulate the attention score computation process based on a combined QK-weight matrix, so that inputs can be directly fed to CIM cells to obtain the score results. Moreover, the involved binomial matrix multiplication operation is decomposed into 4 groups of bit-serial shifting and additions, without costly physical multipliers in the CIM. We maximize the energy efficiency of the CIM circuit through zero-value bit-skipping, data-driven word line activation, read-write separate 6T cells and bit-alternating 14T/28T adders. The proposed CIM macro was implemented using a 65-nm process. It occupied only 0.35 mm2 area, and delivered a 42.27 GOPS peak performance with 1.24 mW power consumption at a 1.0 V power supply and a 100 MHz clock frequency, resulting in 34.1 TOPS/W energy efficiency and 120.77 GOPS/mm2 area efficiency. When compared to the CPU and GPU, our CIM macro is 25x and 13x more energy efficient on practical tasks, respectively. Compared with other Transformer-CIMs, our design exhibits at least 7x energy efficiency and at least 2x area efficiency improvements when scaled to the same technology node, showcasing its potential for edge-side intelligent applications.
Related Topics
Concepts
Macro
Computation
Transformer
Computer science
Efficient energy use
Electronic engineering
Matrix multiplication
Energy consumption
Real-time computing
Electrical efficiency
Computer hardware
Multiplication (music)
Power (physics)
Latency (audio)
Algorithm
Dynamic demand
Energy (signal processing)
Embedded system
Automotive engineering
Matrix (chemical analysis)
Engineering
Simulation
Transformation matrix
Control engineering
Metadata
- Type
- article
- Landing Page
- http://arxiv.org/abs/2511.12152
- https://arxiv.org/pdf/2511.12152
- OA Status
- green
- OpenAlex ID
- https://openalex.org/W7106093713
All OpenAlex metadata
Raw OpenAlex JSON
- OpenAlex ID
-
https://openalex.org/W7106093713Canonical identifier for this work in OpenAlex
- Title
-
A digital SRAM-based compute-in-memory macro for weight-stationary dynamic matrix multiplication in Transformer attention score computationWork title
- Type
-
articleOpenAlex work type
- Publication year
-
2025Year of publication
- Publication date
-
2025-11-15Full publication date if available
- Authors
-
Yu JianYi, Wang Yuxuan, Fu Xiang, Qiao Fei, Wang, Ying, Yuan Rui, Liu Liyuan, Shi CongList of authors in order
- Landing page
-
https://arxiv.org/abs/2511.12152Publisher landing page
- PDF URL
-
https://arxiv.org/pdf/2511.12152Direct link to full text PDF
- Open access
-
YesWhether a free full text is available
- OA status
-
greenOpen access status per OpenAlex
- OA URL
-
https://arxiv.org/pdf/2511.12152Direct OA link when available
- Concepts
-
Macro, Computation, Transformer, Computer science, Efficient energy use, Electronic engineering, Matrix multiplication, Energy consumption, Real-time computing, Electrical efficiency, Computer hardware, Multiplication (music), Power (physics), Latency (audio), Algorithm, Dynamic demand, Energy (signal processing), Embedded system, Automotive engineering, Matrix (chemical analysis), Engineering, Simulation, Transformation matrix, Control engineeringTop concepts (fields/topics) attached by OpenAlex
- Cited by
-
0Total citation count in OpenAlex
Full payload
| id | https://openalex.org/W7106093713 |
|---|---|
| doi | |
| ids.openalex | https://openalex.org/W7106093713 |
| fwci | 0.0 |
| type | article |
| title | A digital SRAM-based compute-in-memory macro for weight-stationary dynamic matrix multiplication in Transformer attention score computation |
| biblio.issue | |
| biblio.volume | |
| biblio.last_page | |
| biblio.first_page | |
| topics[0].id | https://openalex.org/T10054 |
| topics[0].field.id | https://openalex.org/fields/17 |
| topics[0].field.display_name | Computer Science |
| topics[0].score | 0.6980013847351074 |
| topics[0].domain.id | https://openalex.org/domains/3 |
| topics[0].domain.display_name | Physical Sciences |
| topics[0].subfield.id | https://openalex.org/subfields/1708 |
| topics[0].subfield.display_name | Hardware and Architecture |
| topics[0].display_name | Parallel Computing and Optimization Techniques |
| topics[1].id | https://openalex.org/T12808 |
| topics[1].field.id | https://openalex.org/fields/22 |
| topics[1].field.display_name | Engineering |
| topics[1].score | 0.08088371902704239 |
| topics[1].domain.id | https://openalex.org/domains/3 |
| topics[1].domain.display_name | Physical Sciences |
| topics[1].subfield.id | https://openalex.org/subfields/2208 |
| topics[1].subfield.display_name | Electrical and Electronic Engineering |
| topics[1].display_name | Ferroelectric and Negative Capacitance Devices |
| topics[2].id | https://openalex.org/T10502 |
| topics[2].field.id | https://openalex.org/fields/22 |
| topics[2].field.display_name | Engineering |
| topics[2].score | 0.07450518757104874 |
| topics[2].domain.id | https://openalex.org/domains/3 |
| topics[2].domain.display_name | Physical Sciences |
| topics[2].subfield.id | https://openalex.org/subfields/2208 |
| topics[2].subfield.display_name | Electrical and Electronic Engineering |
| topics[2].display_name | Advanced Memory and Neural Computing |
| is_xpac | False |
| apc_list | |
| apc_paid | |
| concepts[0].id | https://openalex.org/C166955791 |
| concepts[0].level | 2 |
| concepts[0].score | 0.6759232878684998 |
| concepts[0].wikidata | https://www.wikidata.org/wiki/Q629579 |
| concepts[0].display_name | Macro |
| concepts[1].id | https://openalex.org/C45374587 |
| concepts[1].level | 2 |
| concepts[1].score | 0.6350135207176208 |
| concepts[1].wikidata | https://www.wikidata.org/wiki/Q12525525 |
| concepts[1].display_name | Computation |
| concepts[2].id | https://openalex.org/C66322947 |
| concepts[2].level | 3 |
| concepts[2].score | 0.6319902539253235 |
| concepts[2].wikidata | https://www.wikidata.org/wiki/Q11658 |
| concepts[2].display_name | Transformer |
| concepts[3].id | https://openalex.org/C41008148 |
| concepts[3].level | 0 |
| concepts[3].score | 0.5874205827713013 |
| concepts[3].wikidata | https://www.wikidata.org/wiki/Q21198 |
| concepts[3].display_name | Computer science |
| concepts[4].id | https://openalex.org/C2742236 |
| concepts[4].level | 2 |
| concepts[4].score | 0.5607659220695496 |
| concepts[4].wikidata | https://www.wikidata.org/wiki/Q924713 |
| concepts[4].display_name | Efficient energy use |
| concepts[5].id | https://openalex.org/C24326235 |
| concepts[5].level | 1 |
| concepts[5].score | 0.4614282250404358 |
| concepts[5].wikidata | https://www.wikidata.org/wiki/Q126095 |
| concepts[5].display_name | Electronic engineering |
| concepts[6].id | https://openalex.org/C17349429 |
| concepts[6].level | 3 |
| concepts[6].score | 0.4595150053501129 |
| concepts[6].wikidata | https://www.wikidata.org/wiki/Q1049914 |
| concepts[6].display_name | Matrix multiplication |
| concepts[7].id | https://openalex.org/C2780165032 |
| concepts[7].level | 2 |
| concepts[7].score | 0.4504646062850952 |
| concepts[7].wikidata | https://www.wikidata.org/wiki/Q16869822 |
| concepts[7].display_name | Energy consumption |
| concepts[8].id | https://openalex.org/C79403827 |
| concepts[8].level | 1 |
| concepts[8].score | 0.4266704022884369 |
| concepts[8].wikidata | https://www.wikidata.org/wiki/Q3988 |
| concepts[8].display_name | Real-time computing |
| concepts[9].id | https://openalex.org/C118993495 |
| concepts[9].level | 3 |
| concepts[9].score | 0.4236331880092621 |
| concepts[9].wikidata | https://www.wikidata.org/wiki/Q5042828 |
| concepts[9].display_name | Electrical efficiency |
| concepts[10].id | https://openalex.org/C9390403 |
| concepts[10].level | 1 |
| concepts[10].score | 0.37361639738082886 |
| concepts[10].wikidata | https://www.wikidata.org/wiki/Q3966 |
| concepts[10].display_name | Computer hardware |
| concepts[11].id | https://openalex.org/C2780595030 |
| concepts[11].level | 2 |
| concepts[11].score | 0.34897342324256897 |
| concepts[11].wikidata | https://www.wikidata.org/wiki/Q3860309 |
| concepts[11].display_name | Multiplication (music) |
| concepts[12].id | https://openalex.org/C163258240 |
| concepts[12].level | 2 |
| concepts[12].score | 0.3377116024494171 |
| concepts[12].wikidata | https://www.wikidata.org/wiki/Q25342 |
| concepts[12].display_name | Power (physics) |
| concepts[13].id | https://openalex.org/C82876162 |
| concepts[13].level | 2 |
| concepts[13].score | 0.3328794240951538 |
| concepts[13].wikidata | https://www.wikidata.org/wiki/Q17096504 |
| concepts[13].display_name | Latency (audio) |
| concepts[14].id | https://openalex.org/C11413529 |
| concepts[14].level | 1 |
| concepts[14].score | 0.31427013874053955 |
| concepts[14].wikidata | https://www.wikidata.org/wiki/Q8366 |
| concepts[14].display_name | Algorithm |
| concepts[15].id | https://openalex.org/C45872418 |
| concepts[15].level | 3 |
| concepts[15].score | 0.3135182857513428 |
| concepts[15].wikidata | https://www.wikidata.org/wiki/Q5318966 |
| concepts[15].display_name | Dynamic demand |
| concepts[16].id | https://openalex.org/C186370098 |
| concepts[16].level | 2 |
| concepts[16].score | 0.2937272787094116 |
| concepts[16].wikidata | https://www.wikidata.org/wiki/Q442787 |
| concepts[16].display_name | Energy (signal processing) |
| concepts[17].id | https://openalex.org/C149635348 |
| concepts[17].level | 1 |
| concepts[17].score | 0.290127158164978 |
| concepts[17].wikidata | https://www.wikidata.org/wiki/Q193040 |
| concepts[17].display_name | Embedded system |
| concepts[18].id | https://openalex.org/C171146098 |
| concepts[18].level | 1 |
| concepts[18].score | 0.2780430316925049 |
| concepts[18].wikidata | https://www.wikidata.org/wiki/Q124192 |
| concepts[18].display_name | Automotive engineering |
| concepts[19].id | https://openalex.org/C106487976 |
| concepts[19].level | 2 |
| concepts[19].score | 0.2767188847064972 |
| concepts[19].wikidata | https://www.wikidata.org/wiki/Q685816 |
| concepts[19].display_name | Matrix (chemical analysis) |
| concepts[20].id | https://openalex.org/C127413603 |
| concepts[20].level | 0 |
| concepts[20].score | 0.26863333582878113 |
| concepts[20].wikidata | https://www.wikidata.org/wiki/Q11023 |
| concepts[20].display_name | Engineering |
| concepts[21].id | https://openalex.org/C44154836 |
| concepts[21].level | 1 |
| concepts[21].score | 0.26688918471336365 |
| concepts[21].wikidata | https://www.wikidata.org/wiki/Q45045 |
| concepts[21].display_name | Simulation |
| concepts[22].id | https://openalex.org/C165443888 |
| concepts[22].level | 3 |
| concepts[22].score | 0.2566603124141693 |
| concepts[22].wikidata | https://www.wikidata.org/wiki/Q1482183 |
| concepts[22].display_name | Transformation matrix |
| concepts[23].id | https://openalex.org/C133731056 |
| concepts[23].level | 1 |
| concepts[23].score | 0.25207793712615967 |
| concepts[23].wikidata | https://www.wikidata.org/wiki/Q4917288 |
| concepts[23].display_name | Control engineering |
| keywords[0].id | https://openalex.org/keywords/macro |
| keywords[0].score | 0.6759232878684998 |
| keywords[0].display_name | Macro |
| keywords[1].id | https://openalex.org/keywords/computation |
| keywords[1].score | 0.6350135207176208 |
| keywords[1].display_name | Computation |
| keywords[2].id | https://openalex.org/keywords/transformer |
| keywords[2].score | 0.6319902539253235 |
| keywords[2].display_name | Transformer |
| keywords[3].id | https://openalex.org/keywords/efficient-energy-use |
| keywords[3].score | 0.5607659220695496 |
| keywords[3].display_name | Efficient energy use |
| keywords[4].id | https://openalex.org/keywords/matrix-multiplication |
| keywords[4].score | 0.4595150053501129 |
| keywords[4].display_name | Matrix multiplication |
| keywords[5].id | https://openalex.org/keywords/energy-consumption |
| keywords[5].score | 0.4504646062850952 |
| keywords[5].display_name | Energy consumption |
| keywords[6].id | https://openalex.org/keywords/electrical-efficiency |
| keywords[6].score | 0.4236331880092621 |
| keywords[6].display_name | Electrical efficiency |
| language | |
| locations[0].id | pmh:oai:arXiv.org:2511.12152 |
| locations[0].is_oa | True |
| locations[0].source.id | https://openalex.org/S4306400194 |
| locations[0].source.issn | |
| locations[0].source.type | repository |
| locations[0].source.is_oa | True |
| locations[0].source.issn_l | |
| locations[0].source.is_core | False |
| locations[0].source.is_in_doaj | False |
| locations[0].source.display_name | arXiv (Cornell University) |
| locations[0].source.host_organization | https://openalex.org/I205783295 |
| locations[0].source.host_organization_name | Cornell University |
| locations[0].source.host_organization_lineage | https://openalex.org/I205783295 |
| locations[0].license | |
| locations[0].pdf_url | https://arxiv.org/pdf/2511.12152 |
| locations[0].version | submittedVersion |
| locations[0].raw_type | text |
| locations[0].license_id | |
| locations[0].is_accepted | False |
| locations[0].is_published | False |
| locations[0].raw_source_name | |
| locations[0].landing_page_url | http://arxiv.org/abs/2511.12152 |
| indexed_in | arxiv |
| authorships[0].author.id | https://openalex.org/A2671184843 |
| authorships[0].author.orcid | |
| authorships[0].author.display_name | Yu JianYi |
| authorships[0].author_position | first |
| authorships[0].raw_author_name | Yu, Jianyi |
| authorships[0].is_corresponding | True |
| authorships[1].author.id | https://openalex.org/A1965576348 |
| authorships[1].author.orcid | |
| authorships[1].author.display_name | Wang Yuxuan |
| authorships[1].author_position | middle |
| authorships[1].raw_author_name | Wang, Yuxuan |
| authorships[1].is_corresponding | False |
| authorships[2].author.id | https://openalex.org/A1977120960 |
| authorships[2].author.orcid | |
| authorships[2].author.display_name | Fu Xiang |
| authorships[2].author_position | middle |
| authorships[2].raw_author_name | Fu, Xiang |
| authorships[2].is_corresponding | False |
| authorships[3].author.id | https://openalex.org/A2023561171 |
| authorships[3].author.orcid | |
| authorships[3].author.display_name | Qiao Fei |
| authorships[3].author_position | middle |
| authorships[3].raw_author_name | Qiao, Fei |
| authorships[3].is_corresponding | False |
| authorships[4].author.id | https://openalex.org/A1843046921 |
| authorships[4].author.orcid | https://orcid.org/0000-0001-6025-2670 |
| authorships[4].author.display_name | Wang, Ying |
| authorships[4].author_position | middle |
| authorships[4].raw_author_name | Wang, Ying |
| authorships[4].is_corresponding | False |
| authorships[5].author.id | https://openalex.org/A2139533833 |
| authorships[5].author.orcid | https://orcid.org/0000-0001-5165-8898 |
| authorships[5].author.display_name | Yuan Rui |
| authorships[5].author_position | middle |
| authorships[5].raw_author_name | Yuan, Rui |
| authorships[5].is_corresponding | False |
| authorships[6].author.id | https://openalex.org/A2112714395 |
| authorships[6].author.orcid | https://orcid.org/0000-0001-8010-9126 |
| authorships[6].author.display_name | Liu Liyuan |
| authorships[6].author_position | middle |
| authorships[6].raw_author_name | Liu, Liyuan |
| authorships[6].is_corresponding | False |
| authorships[7].author.id | https://openalex.org/A2152690874 |
| authorships[7].author.orcid | |
| authorships[7].author.display_name | Shi Cong |
| authorships[7].author_position | last |
| authorships[7].raw_author_name | Shi, Cong |
| authorships[7].is_corresponding | False |
| has_content.pdf | False |
| has_content.grobid_xml | False |
| is_paratext | False |
| open_access.is_oa | True |
| open_access.oa_url | https://arxiv.org/pdf/2511.12152 |
| open_access.oa_status | green |
| open_access.any_repository_has_fulltext | False |
| created_date | 2025-11-19T00:00:00 |
| display_name | A digital SRAM-based compute-in-memory macro for weight-stationary dynamic matrix multiplication in Transformer attention score computation |
| has_fulltext | False |
| is_retracted | False |
| updated_date | 2025-11-19T23:39:43.309859 |
| primary_topic.id | https://openalex.org/T10054 |
| primary_topic.field.id | https://openalex.org/fields/17 |
| primary_topic.field.display_name | Computer Science |
| primary_topic.score | 0.6980013847351074 |
| primary_topic.domain.id | https://openalex.org/domains/3 |
| primary_topic.domain.display_name | Physical Sciences |
| primary_topic.subfield.id | https://openalex.org/subfields/1708 |
| primary_topic.subfield.display_name | Hardware and Architecture |
| primary_topic.display_name | Parallel Computing and Optimization Techniques |
| cited_by_count | 0 |
| locations_count | 1 |
| best_oa_location.id | pmh:oai:arXiv.org:2511.12152 |
| best_oa_location.is_oa | True |
| best_oa_location.source.id | https://openalex.org/S4306400194 |
| best_oa_location.source.issn | |
| best_oa_location.source.type | repository |
| best_oa_location.source.is_oa | True |
| best_oa_location.source.issn_l | |
| best_oa_location.source.is_core | False |
| best_oa_location.source.is_in_doaj | False |
| best_oa_location.source.display_name | arXiv (Cornell University) |
| best_oa_location.source.host_organization | https://openalex.org/I205783295 |
| best_oa_location.source.host_organization_name | Cornell University |
| best_oa_location.source.host_organization_lineage | https://openalex.org/I205783295 |
| best_oa_location.license | |
| best_oa_location.pdf_url | https://arxiv.org/pdf/2511.12152 |
| best_oa_location.version | submittedVersion |
| best_oa_location.raw_type | text |
| best_oa_location.license_id | |
| best_oa_location.is_accepted | False |
| best_oa_location.is_published | False |
| best_oa_location.raw_source_name | |
| best_oa_location.landing_page_url | http://arxiv.org/abs/2511.12152 |
| primary_location.id | pmh:oai:arXiv.org:2511.12152 |
| primary_location.is_oa | True |
| primary_location.source.id | https://openalex.org/S4306400194 |
| primary_location.source.issn | |
| primary_location.source.type | repository |
| primary_location.source.is_oa | True |
| primary_location.source.issn_l | |
| primary_location.source.is_core | False |
| primary_location.source.is_in_doaj | False |
| primary_location.source.display_name | arXiv (Cornell University) |
| primary_location.source.host_organization | https://openalex.org/I205783295 |
| primary_location.source.host_organization_name | Cornell University |
| primary_location.source.host_organization_lineage | https://openalex.org/I205783295 |
| primary_location.license | |
| primary_location.pdf_url | https://arxiv.org/pdf/2511.12152 |
| primary_location.version | submittedVersion |
| primary_location.raw_type | text |
| primary_location.license_id | |
| primary_location.is_accepted | False |
| primary_location.is_published | False |
| primary_location.raw_source_name | |
| primary_location.landing_page_url | http://arxiv.org/abs/2511.12152 |
| publication_date | 2025-11-15 |
| publication_year | 2025 |
| referenced_works_count | 0 |
| abstract_inverted_index.4 | 91 |
| abstract_inverted_index.V | 160 |
| abstract_inverted_index.a | 31, 62, 136, 147, 158, 164 |
| abstract_inverted_index.2x | 216 |
| abstract_inverted_index.6T | 123 |
| abstract_inverted_index.7x | 210 |
| abstract_inverted_index.It | 139 |
| abstract_inverted_index.To | 39 |
| abstract_inverted_index.We | 105 |
| abstract_inverted_index.at | 157, 208, 214 |
| abstract_inverted_index.be | 70 |
| abstract_inverted_index.by | 19 |
| abstract_inverted_index.in | 6, 102, 170 |
| abstract_inverted_index.is | 45, 88, 190 |
| abstract_inverted_index.mW | 154 |
| abstract_inverted_index.of | 93, 110 |
| abstract_inverted_index.on | 61, 197 |
| abstract_inverted_index.so | 66 |
| abstract_inverted_index.to | 35, 73, 76, 182, 222 |
| abstract_inverted_index.we | 53 |
| abstract_inverted_index.1.0 | 159 |
| abstract_inverted_index.100 | 165 |
| abstract_inverted_index.13x | 193 |
| abstract_inverted_index.25x | 191 |
| abstract_inverted_index.CIM | 33, 51, 74, 112, 131, 188 |
| abstract_inverted_index.CPU | 184 |
| abstract_inverted_index.MHz | 166 |
| abstract_inverted_index.The | 129 |
| abstract_inverted_index.and | 15, 25, 96, 125, 145, 163, 175, 185, 192, 213 |
| abstract_inverted_index.are | 3 |
| abstract_inverted_index.can | 69 |
| abstract_inverted_index.fed | 72 |
| abstract_inverted_index.for | 47, 230 |
| abstract_inverted_index.its | 228 |
| abstract_inverted_index.mm2 | 143 |
| abstract_inverted_index.our | 187, 205 |
| abstract_inverted_index.the | 48, 55, 78, 82, 103, 107, 111, 183, 223 |
| abstract_inverted_index.was | 133 |
| abstract_inverted_index.(AI) | 10 |
| abstract_inverted_index.0.35 | 142 |
| abstract_inverted_index.1.24 | 153 |
| abstract_inverted_index.34.1 | 171 |
| abstract_inverted_index.CIM. | 104 |
| abstract_inverted_index.GOPS | 149 |
| abstract_inverted_index.GPU, | 186 |
| abstract_inverted_index.They | 12 |
| abstract_inverted_index.This | 28 |
| abstract_inverted_index.When | 180 |
| abstract_inverted_index.area | 178, 217 |
| abstract_inverted_index.data | 21 |
| abstract_inverted_index.into | 90 |
| abstract_inverted_index.line | 119 |
| abstract_inverted_index.more | 194 |
| abstract_inverted_index.only | 141 |
| abstract_inverted_index.peak | 150 |
| abstract_inverted_index.same | 224 |
| abstract_inverted_index.that | 44, 67 |
| abstract_inverted_index.when | 220 |
| abstract_inverted_index.with | 152, 202 |
| abstract_inverted_index.word | 118 |
| abstract_inverted_index.work | 29 |
| abstract_inverted_index.(CIM) | 1 |
| abstract_inverted_index.42.27 | 148 |
| abstract_inverted_index.65-nm | 137 |
| abstract_inverted_index.area, | 144 |
| abstract_inverted_index.based | 60 |
| abstract_inverted_index.cells | 75, 124 |
| abstract_inverted_index.clock | 167 |
| abstract_inverted_index.least | 209, 215 |
| abstract_inverted_index.macro | 34, 132, 189 |
| abstract_inverted_index.node, | 226 |
| abstract_inverted_index.other | 203 |
| abstract_inverted_index.power | 14, 155, 161 |
| abstract_inverted_index.score | 57, 79 |
| abstract_inverted_index.using | 135 |
| abstract_inverted_index.120.77 | 176 |
| abstract_inverted_index.TOPS/W | 172 |
| abstract_inverted_index.caused | 18 |
| abstract_inverted_index.common | 49 |
| abstract_inverted_index.costly | 99 |
| abstract_inverted_index.design | 206 |
| abstract_inverted_index.energy | 108, 173, 195, 211 |
| abstract_inverted_index.groups | 92 |
| abstract_inverted_index.inputs | 68 |
| abstract_inverted_index.matrix | 42, 85 |
| abstract_inverted_index.obtain | 77 |
| abstract_inverted_index.scaled | 221 |
| abstract_inverted_index.supply | 162 |
| abstract_inverted_index.tasks, | 199 |
| abstract_inverted_index.units. | 27 |
| abstract_inverted_index.widely | 4 |
| abstract_inverted_index.14T/28T | 127 |
| abstract_inverted_index.adders. | 128 |
| abstract_inverted_index.between | 23 |
| abstract_inverted_index.circuit | 113 |
| abstract_inverted_index.compute | 24, 36 |
| abstract_inverted_index.digital | 32 |
| abstract_inverted_index.dynamic | 41 |
| abstract_inverted_index.latency | 16 |
| abstract_inverted_index.matrix, | 65 |
| abstract_inverted_index.process | 59 |
| abstract_inverted_index.storage | 26 |
| abstract_inverted_index.through | 114 |
| abstract_inverted_index.without | 98 |
| abstract_inverted_index.Compared | 201 |
| abstract_inverted_index.GOPS/mm2 | 177 |
| abstract_inverted_index.binomial | 84 |
| abstract_inverted_index.combined | 63 |
| abstract_inverted_index.compared | 181 |
| abstract_inverted_index.directly | 71 |
| abstract_inverted_index.employed | 5 |
| abstract_inverted_index.exhibits | 207 |
| abstract_inverted_index.involved | 83 |
| abstract_inverted_index.maximize | 106 |
| abstract_inverted_index.mitigate | 40 |
| abstract_inverted_index.occupied | 140 |
| abstract_inverted_index.physical | 100 |
| abstract_inverted_index.process. | 138 |
| abstract_inverted_index.proposed | 130 |
| abstract_inverted_index.proposes | 30 |
| abstract_inverted_index.results. | 80 |
| abstract_inverted_index.separate | 122 |
| abstract_inverted_index.shifting | 95 |
| abstract_inverted_index.Moreover, | 81 |
| abstract_inverted_index.QK-weight | 64 |
| abstract_inverted_index.alleviate | 13 |
| abstract_inverted_index.attention | 56 |
| abstract_inverted_index.delivered | 146 |
| abstract_inverted_index.edge-side | 231 |
| abstract_inverted_index.efficient | 196 |
| abstract_inverted_index.extensive | 20 |
| abstract_inverted_index.movements | 22 |
| abstract_inverted_index.operation | 87 |
| abstract_inverted_index.paradigm, | 52 |
| abstract_inverted_index.potential | 229 |
| abstract_inverted_index.practical | 198 |
| abstract_inverted_index.resulting | 169 |
| abstract_inverted_index.additions, | 97 |
| abstract_inverted_index.artificial | 8 |
| abstract_inverted_index.attention. | 38 |
| abstract_inverted_index.bit-serial | 94 |
| abstract_inverted_index.decomposed | 89 |
| abstract_inverted_index.efficiency | 109, 174, 212, 218 |
| abstract_inverted_index.frequency, | 168 |
| abstract_inverted_index.read-write | 121 |
| abstract_inverted_index.showcasing | 227 |
| abstract_inverted_index.techniques | 2 |
| abstract_inverted_index.technology | 225 |
| abstract_inverted_index.unsuitable | 46 |
| abstract_inverted_index.zero-value | 115 |
| abstract_inverted_index.Transformer | 37 |
| abstract_inverted_index.activation, | 120 |
| abstract_inverted_index.bottlenecks | 17 |
| abstract_inverted_index.computation | 58 |
| abstract_inverted_index.consumption | 156 |
| abstract_inverted_index.data-driven | 117 |
| abstract_inverted_index.efficiency. | 179 |
| abstract_inverted_index.implemented | 134 |
| abstract_inverted_index.intelligent | 9, 232 |
| abstract_inverted_index.multipliers | 101 |
| abstract_inverted_index.performance | 151 |
| abstract_inverted_index.processors. | 11 |
| abstract_inverted_index.reformulate | 54 |
| abstract_inverted_index.improvements | 219 |
| abstract_inverted_index.applications. | 233 |
| abstract_inverted_index.bit-skipping, | 116 |
| abstract_inverted_index.respectively. | 200 |
| abstract_inverted_index.multiplication | 43, 86 |
| abstract_inverted_index.bit-alternating | 126 |
| abstract_inverted_index.energy-efficient | 7 |
| abstract_inverted_index.Compute-in-memory | 0 |
| abstract_inverted_index.Transformer-CIMs, | 204 |
| abstract_inverted_index.weight-stationary | 50 |
| cited_by_percentile_year | |
| countries_distinct_count | 0 |
| institutions_distinct_count | 8 |
| citation_normalized_percentile.value | 0.8790426 |
| citation_normalized_percentile.is_in_top_1_percent | False |
| citation_normalized_percentile.is_in_top_10_percent | True |