Skip to main content

HIGH PERFORMANCE, SAFE, SECURE, OPEN-SOURCE LEVER-AGED RISC-V DOMAIN-SPECIFIC ECOSYSTEMS

PCI2023-143358

•
Funding agency name Agencia Estatal de Investigación
Funding agency acronym AEI
Program Programa Estatal para Afrontar las Prioridades de Nuestro Entorno
Subprogram Subprograma Estatal de Internacionalización
Call Proyectos Programación Conjunta Internacional
Call year 2023
State Plan 2021-2023 Plan Estatal de Investigación Científica y Técnica y de Innovación
Beneficiary centre UNIVERSITAT POLITÈCNICA DE VALÈNCIA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 5
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Tesis de maestría (MasterThesis)

Acceleration of Quantum LDPC Decoding with a RISC-V-Based Vector Processor, Aceleración de la decodificación de LDPC Cuánticos con un procesador vectorial basado en RISC-V, Aceleració de la decodificació de LPDC Cuántics amb un procesador vectorial RISC-V

  • Fernández Andonegui, Iker
[CA] La computació quàntica promet resoldre problemes més enllà de l’abast dels sistemes clàssics, però la seua escalabilitat depén d’una correcció d’errors quàntics (Quantum Error Correction, QEC) eficient. Entre els esquemes disponibles, els codis Quantum Low-Density Parity-Check (QLDPC) destaquen pel seu potencial per a detectar errors en processadors quàntics a gran escala. Al mateix temps, RISC-V ha sorgit com una arquitectura oberta i flexible, i la seua extensió vectorial ofereix un marc adequat per a explorar el paral·lelisme en la descodificació de QLDPC. Aprofitar aquestes capacitats pot ajudar a reduir les altes latències de descodificació que actualment limiten la QEC pràctica. En aquest treball, estudiem com els algorismes QLDPC es poden adaptar a l’extensió vectorial de RISC-V (RVV), i explorem la possibilitat d’introduir instruccions especialitzades per a accelerar encara més les operacions crítiques de descodificació., [EN] Quantum computing promises to solve problems beyond the reach of classical systems, but its scalability depends on efficient quantum error correction (QEC). Among the available schemes, Quantum Low-Density Parity-Check (QLDPC) codes stand out for their potential to detect errors in large-scale quantum processors. At the same time, RISC-V has emerged as an open and flexible architecture, and its vector extension provides a suitable framework to explore parallelism in QLDPC decoding. Leveraging these capabilities can help reduce the high decoding latencies that currently limit practical QEC. In this work, we study how QLDPC algorithms can be adapted to the RISC-V Vector Extension (RVV) and explore the possibility of introducing specialized instructions to further accelerate critical decoding operations., [ES] La computación cuántica promete resolver problemas más allá del alcance de los sistemas clásicos, pero su escalabilidad depende de una corrección de errores cuánticos (Quantum Error Correction, QEC) eficiente. Entre los esquemas disponibles, los códigos Quantum Low-Density Parity-Check (QLDPC) destacan por su potencial para detectar errores en procesadores cuánticos a gran escala. Al mismo tiempo, RISC-V ha surgido como una arquitectura abierta y flexible, y su extensión vectorial ofrece un marco adecuado para explorar el paralelismo en la descodificación de QLDPC. Aprovechar estas capacidades puede ayudar a reducir las altas latencias de descodificación que actualmente limitan la QEC práctica. En este trabajo, estudiamos cómo los algoritmos QLDPC pueden adaptarse a la extensión vectorial de RISC-V (RVV), y exploramos la posibilidad de introducir instrucciones especializadas para acelerar aún más las operaciones críticas de descodificación.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Expanding SafeSU capabilities by leveraging security frameworks for contention monitoring in complex SoCs

  • Andreu-Cerezo, Pablo|||0000-0003-2070-9271
  • López Rodríguez, Pedro Juan|||0000-0003-4544-955X
  • Hernández Luz, Carles|||0000-0003-1192-6223
  • Alcaide, Sergi
  • Abella, Jaume
[EN] The increased performance requirements of applications running on safety-critical systems have led to the use of complex platforms with several CPUs, GPUs, and AI accelerators. However, higher platform and system complexity challenge performance verification and validation since timing interference across tasks occurs in unobvious ways, hence defeating attempts to optimize application consolidation informedly during design phases and validating that mutual interference across tasks is within bounds during test phases. In that respect, the SafeSU has been proposed to extend inter-task interference monitoring capabilities in simple systems. However, modern mixed-criticality systems are complex, with multilayered interconnects, shared caches, and hardware accelerators. To that end, this paper proposes a non-intrusive add-on approach for monitoring interference across tasks in multilayer heterogeneous systems implemented by leveraging existing security frameworks and the SafeSU infrastructure. The feasibility of the proposed approach has been validated in an RTL RISC-V-based multicore SoC with support for AI hardware acceleration. Our results show that our approach can safely track contention and properly break down contention cycles across the different sources of interference, hence guiding optimization and validation processes., This work has received funding from the Chips Joint Undertaking project ISOLDE no. 101112274 and the Spanish Agencia Estatal de Investigacion PCI2023-143358 grant. Pablo Andreu has received funding from Generalitat Valenciana, Spain via "Subvenciones para la contratacion de personal investigador predoctoral" with reference CIACIF/2021/412. Carles Hernandez is partially supported by Spanish Ministry of Science, Innovation and Universities under "Ramon y Cajal", fellowship No. RYC2020-030685-I. Funding for open access charge: CRUE-Universitat Politecnica de Valencia.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Proyecto fin de carrera (BachelorThesis)

Performance Improvements of an Open-source L2 Cache, Millores de rendiment a una memoria cache L2 de codi obert, Mejoras de rendimiento a una memoria caché L2 de código abierto

  • Roche Peris, David
[CA] L’alt interés industrial en els processadors RISC-V de codi obert han generat la necessitat de nous components més complexos que puguen optimitzar al màxim les prestacions dels processadors. Les caches compartides, generalment l’L2 en processadors embeguts d’altes prestacions, són un dels components més crítics en els Sistemes en un Chip (SoCs). No obstant això, encara que existeixen molts processadors RISC-V multinucli, les implementacions de caches L2 actuals són generalment molt més simples que les presents en processadors comercials. Aquest projecte presenta les millores d’una cache L2 de codi obert, implementant optimitzacions ben estudiades i presents en productes comercials per millorar les prestacions de la cache. Concretament, es posa enfasi en la implementació de dues millores: particionat basat en vies, i transferències no bloquejants. La cache ha sigut descrita en RTL i prototipada i testejada en una FPGA. Els resultats mostren que aquesta nova cache millora signicativament les prestacions i la qualitat de servei (QoS) dels sistemes multinucli., [ES] El objetivo de este proyecto es generar una caché L2 de código abierto de alto rendimiento adecuada para entornos multi-núcleo en aplicaciones de seguridad funcional. Este proyecto utilizará la caché L2 de código abierto de Frontgrade Gaisler como punto de partida (de su biblioteca de código abierto GRLIB). En primer lugar, la caché será caracterizada, analizando las limitaciones del sistema. Luego, esta caché será ampliada mediante la adición de nuevas características que mejoren el rendimiento del diseño utilizando como principios la modularidad y la simplicidad. Finalmente, el sistema será evaluado de forma experimental cuantificando las mejoras obtenidas., [EN] The aim of this project is to produce a high-performance open- source l2 cache suitable for multi-core, safety-critical environments. This project will use Frontgrade Gaisler s open-source L2-cache as a starting point (from their open-source library GRLIB). Firstly, the cache will be characterized, analyzing the limitations of the system. Then, the cache will be extended by adding new features that improve the performance of the design using as principles modu- larity and simplicity. Finally, the system will be tested, and the performance gain will be quantified. The project will focus on three main features: i) the implementation of a par- titioning scheme that reduces the unpredictability of multi-core systems; ii) the addition of a frontend module that makes the design more simple and modular, making it easily portable to other systems using different bus technologies; and iii) the parallelization of the design, making the cache non-blocking, allowing the bus to continue making requests in the presence of misses. 1
UPCommons. Portal del coneixement obert de la UPC
Publication Artículo científico (JournalArticle)

Expanding SafeSU capabilities by leveraging security frameworks for contention monitoring in complex SoCs

  • Andreu, Pablo
  • Alcaide Portet, Sergi|||0000-0001-8561-7795
  • Lopez, Pedro
  • Abella Ferrer, Jaume|||0000-0001-7951-4028
  • Hernandez, Carles
The increased performance requirements of applications running on safety-critical systems have led to the use of complex platforms with several CPUs, GPUs, and AI accelerators. However, higher platform and system complexity challenge performance verification and validation since timing interference across tasks occurs in unobvious ways, hence defeating attempts to optimize application consolidation informedly during design phases and validating that mutual interference across tasks is within bounds during test phases. In that respect, the SafeSU has been proposed to extend inter-task interference monitoring capabilities in simple systems. However, modern mixed-criticality systems are complex, with multilayered interconnects, shared caches, and hardware accelerators. To that end, this paper proposes a non-intrusive add-on approach for monitoring interference across tasks in multilayer heterogeneous systems implemented by leveraging existing security frameworks and the SafeSU infrastructure. The feasibility of the proposed approach has been validated in an RTL RISC-V-based multicore SoC with support for AI hardware acceleration. Our results show that our approach can safely track contention and properly break down contention cycles across the different sources of interference, hence guiding optimization and validation processes., This work has received funding from the Chips Joint Undertaking project ISOLDE no. 101112274 and the Spanish Agencia Estatal de Investigación PCI2023-143358 grant. Pablo Andreu has received funding from Generalitat Valenciana, Spain via “Subvenciones para la contratación de personal investigador predoctoral” with reference CIACIF/2021/412. Carles Hernández is partially supported by Spanish Ministry of Science, Innovation and Universities under “Ramón Cajal”, fellowship No. RYC2020-030685-I. Funding for open access charge: CRUE-Universitat Politecnica de Valencia ., Peer Reviewed
Recercat. Dipósit de la Recerca de Catalunya
Publication Artículo científico (JournalArticle)

Expanding SafeSU capabilities by leveraging security frameworks for contention monitoring in complex SoCs

  • Andreu, Pablo
  • Alcaide Portet, Sergi
  • Lopez, Pedro
  • Abella Ferrer, Jaume
  • Hernandez, Carles
The increased performance requirements of applications running on safety-critical systems have led to the use of complex platforms with several CPUs, GPUs, and AI accelerators. However, higher platform and system complexity challenge performance verification and validation since timing interference across tasks occurs in unobvious ways, hence defeating attempts to optimize application consolidation informedly during design phases and validating that mutual interference across tasks is within bounds during test phases. In that respect, the SafeSU has been proposed to extend inter-task interference monitoring capabilities in simple systems. However, modern mixed-criticality systems are complex, with multilayered interconnects, shared caches, and hardware accelerators. To that end, this paper proposes a non-intrusive add-on approach for monitoring interference across tasks in multilayer heterogeneous systems implemented by leveraging existing security frameworks and the SafeSU infrastructure. The feasibility of the proposed approach has been validated in an RTL RISC-V-based multicore SoC with support for AI hardware acceleration. Our results show that our approach can safely track contention and properly break down contention cycles across the different sources of interference, hence guiding optimization and validation processes., This work has received funding from the Chips Joint Undertaking project ISOLDE no. 101112274 and the Spanish Agencia Estatal de Investigación PCI2023-143358 grant. Pablo Andreu has received funding from Generalitat Valenciana, Spain via “Subvenciones para la contratación de personal investigador predoctoral” with reference CIACIF/2021/412. Carles Hernández is partially supported by Spanish Ministry of Science, Innovation and Universities under “Ramón Cajal”, fellowship No. RYC2020-030685-I. Funding for open access charge: CRUE-Universitat Politecnica de Valencia ., Peer Reviewed, Postprint (published version)