Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Collaborative Research and Development Grants

University of Toronto

$55,000.00

Power-Aware Cache-Based Structure Design

EF 2011-12

Résumé
The leakage power consumed by said caches is substantial and therefore an interesting target for power reduction. Currently, the cache sizes are generally chosen to provide high performance under worst case load conditions. The power management approaches that are currently in uses generally focus on waiting for the caches and surrounding blocks to become fully idle, at which time the caches are flushed and powered down. The central thesis of this research proposal is that the ability to adaptively vary cache size (by power gating pieces of the cache) based on the needs of a given work-load can provide substantial average power and possibly TDP (thermal design power) savings at the expense of limited performance loss. Confirming the above thesis and, if confirmed, recommending an approach is a complicated task and one that needs to be performed on a case by case basis: CPU caches, Graphics core sub-block caches and Server Northbridge TLBs service widely different access patterns and each case needs to be evaluated independently. In addition to dynamic cache sizing (which, for example, can take the form of resizing the number of sets or ways ), additional leakage reduction techniques can be of benefit. These include: (1) dynamic cache decay where cache lines are powered off after a period of inactivity, (2) dynamic cache sub-blocking where not all the data of a cache block is allocated in the cache, (3) zero and frequent pattern compression where frequent values and patterns are encoded in the cache to take up less space, (4) common-prefix compression for tags that reduces the amount of space required to store tags, (5) more effective cache replacement policies that increase the hit rate and the effectiveness of the cache, and (5) alternate SRAM designs that reduce leakage by exploiting data liveness and content. Benefits: -Gaining substantial competitive edge by reducing average power consumption in many products using methods that are not currently in wide use within the industry. -Transfer of technology expertise in computer architecture design methodologies to AMD and to the University. -Forming a, hopefully, permanent collaborative relationship with a recognized expert in low power computer architecture. -Foster the training of highly skilled people building towards a critical mass of highly-skilled computer architects (an area of expertise that Ontario and Canada is lacking). -Strengthen the expertise and value of the AMD Markham site.
Programme
Collaborative Research and Development Grants
Bénéficiaire
Emplacement
Ontario, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data

Tel que publié

Application id
386377-2009
Name nom
Moshovos, Andreas
Department département
Electrical and Computer Engineering, Edward S. Rogers Sr. Department of
Organization id
31
Institution établissement
University of Toronto
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2011
Competition year année de concours
2009
Award amount
55000
Program id
CRDPJ
Program nane en
Collaborative Research and Development Grants
Program name fr
Subventions de recherche et développement coopérative
Group en
Research partnerships programs
Group fr
Programmes de partenariat
Committee code
96
Committee name en
RPP Internal Decision Cttee
Committee name fr
Cté de décision interne - PPR
Area of application code
802
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Information systems and technology
Area of application fr
Systèmes d'information et technologie de l'information
Research subject code
2719
Research subject group en
Information technology
Research subject group fr
Technologie de l'information
Research subject en
Computer architecture and design
Research subject fr
Architecture d'ordinateur et conception
Partie
3
Nb partie
3
Application title
Power-Aware Cache-Based Structure Design
Keyword
NSERC
Application summary
The leakage power consumed by said caches is substantial and therefore an interesting target for power reduction. Currently, the cache sizes are generally chosen to provide high performance under worst case load conditions. The power management approaches that are currently in uses generally focus on waiting for the caches and surrounding blocks to become fully idle, at which time the caches are flushed and powered down. The central thesis of this research proposal is that the ability to adaptively vary cache size (by power gating pieces of the cache) based on the needs of a given work-load can provide substantial average power and possibly TDP (thermal design power) savings at the expense of limited performance loss. Confirming the above thesis and, if confirmed, recommending an approach is a complicated task and one that needs to be performed on a case by case basis: CPU caches, Graphics core sub-block caches and Server Northbridge TLBs service widely different access patterns and each case needs to be evaluated independently. In addition to dynamic cache sizing (which, for example, can take the form of resizing the number of sets or ways ), additional leakage reduction techniques can be of benefit. These include: (1) dynamic cache decay where cache lines are powered off after a period of inactivity, (2) dynamic cache sub-blocking where not all the data of a cache block is allocated in the cache, (3) zero and frequent pattern compression where frequent values and patterns are encoded in the cache to take up less space, (4) common-prefix compression for tags that reduces the amount of space required to store tags, (5) more effective cache replacement policies that increase the hit rate and the effectiveness of the cache, and (5) alternate SRAM designs that reduce leakage by exploiting data liveness and content. Benefits: -Gaining substantial competitive edge by reducing average power consumption in many products using methods that are not currently in wide use within the industry. -Transfer of technology expertise in computer architecture design methodologies to AMD and to the University. -Forming a, hopefully, permanent collaborative relationship with a recognized expert in low power computer architecture. -Foster the training of highly skilled people building towards a critical mass of highly-skilled computer architects (an area of expertise that Ontario and Canada is lacking). -Strengthen the expertise and value of the AMD Markham site.