Carlos Rojas Morales
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Carlos Rojas Morales has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
- Adrian Cristal
Barcelona Supercomputing Center, Barcelona, Spain
2 shared publications - Francesco Minervini
Barcelona Supercomputing Center, Barcelona, Spain
2 shared publications - Mauro Olivieri
Department of Information Engineering, Electronics and Telecommunications, Sapienza University of Rome, Rome, Italy
2 shared publications - Osman Unsal
Barcelona Supercomputing Center
2 shared publications - Abraham Ruiz
Barcelona Supercomputing Center, Barcelona, Spain
1 shared publication - Alberto Gonzalez
Barcelona Supercomputing Center, Barcelona, Spain
1 shared publication
Latest Publications
- Open Access
- 25 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of the Computer Society
- Vol. 5, 553-565
Enhancing Fault Tolerance in High-Performance Computing: A Real Hardware Case Study on a RISC-V Vector Processing Unit
High-Performance Computing (HPC) systems are designed for large-scale processing and complex dataset analysis leveraging scalability, efficiency, and parallelism, often integrating specialized hardware structures such as Vector Processing Units (VPUs). As these systems have grown in complexity and scale, their vulnerability to errors and failures has become an important and complex issue in the HPC world. Our research addresses this challenge by exploring and implementing advanced fault tolerance techniques inside the Vitruvius+ architecture, a partial out-of-order Vector Processing Unit. To the best of our knowledge, this is the first full RTL-level implementation of instruction replication in an HPC-class vector processor for reliability. Specifically, we investigate the integration and interaction of redundancy mechanisms inside the most sensitive architectural units, obtaining a reduction of 75% in non-silent faults causing system failure, proven by an extensive fault injection simulation campaign, with a hardware overhead of only 7.5% and a negligible variation in clock frequency.
- 25 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of the Computer Society
- Vol. 5, 553-565
Open AccessEnhancing Fault Tolerance in High-Performance Computing: A Real Hardware Case Study on a RISC-V Vector Processing Unit
High-Performance Computing (HPC) systems are designed for large-scale processing and complex dataset analysis leveraging scalability, efficiency, and parallelism, often integrating specialized hardware structures such as Vector Processing Units (VPUs). As these systems have grown in complexity and scale, their vulnerability to errors and failures has become an important and complex issue in the HPC world. Our research addresses this challenge by exploring and implementing advanced fault tolerance techniques inside the Vitruvius+ architecture, a partial out-of-order Vector Processing Unit. To the best of our knowledge, this is the first full RTL-level implementation of instruction replication in an HPC-class vector processor for reliability. Specifically, we investigate the integration and interaction of redundancy mechanisms inside the most sensitive architectural units, obtaining a reduction of 75% in non-silent faults causing system failure, proven by an extensive fault injection simulation campaign, with a hardware overhead of only 7.5% and a negligible variation in clock frequency.
- Top Cited PapersOpen Access
- 1 March 2023
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
- Vol. 20 (2) , 1-25
Vitruvius+: An Area-Efficient RISC-V Decoupled Vector Coprocessor for High Performance Computing Applications
The maturity level of RISC-V and the availability of domain-specific instruction set extensions, like vector processing, make RISC-V a good candidate for supporting the integration of specialized hardware in processor cores for the High Performance Computing (HPC) application domain. In this article, 1 we present Vitruvius+, the vector processing acceleration engine that represents the core of vector instruction execution in the HPC challenge that comes within the EuroHPC initiative. It implements the RISC-V vector extension (RVV) 0.7.1 and can be easily connected to a scalar core using the Open Vector Interface standard. Vitruvius+ natively supports long vectors: 256 double precision floating-point elements in a single vector register. It is composed of a set of identical vector pipelines (lanes), each containing a slice of the Vector Register File and functional units (one integer, one floating point). The vector instruction execution scheme is hybrid in-order/out-of-order and is supported by register renaming and arithmetic/memory instruction decoupling. On a stand-alone synthesis, Vitruvius+ reaches a maximum frequency of 1.4 GHz in typical conditions (TT/0.80V/25°C) using GlobalFoundries 22FDX FD-SOI. The silicon implementation has a total area of 1.3 mm 2 and maximum estimated power of ∼920 mW for one instance of Vitruvius+ equipped with eight vector lanes.
- 1 March 2023
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
- Vol. 20 (2) , 1-25
Top Cited PapersOpen AccessVitruvius+: An Area-Efficient RISC-V Decoupled Vector Coprocessor for High Performance Computing Applications
The maturity level of RISC-V and the availability of domain-specific instruction set extensions, like vector processing, make RISC-V a good candidate for supporting the integration of specialized hardware in processor cores for the High Performance Computing (HPC) application domain. In this article, 1 we present Vitruvius+, the vector processing acceleration engine that represents the core of vector instruction execution in the HPC challenge that comes within the EuroHPC initiative. It implements the RISC-V vector extension (RVV) 0.7.1 and can be easily connected to a scalar core using the Open Vector Interface standard. Vitruvius+ natively supports long vectors: 256 double precision floating-point elements in a single vector register. It is composed of a set of identical vector pipelines (lanes), each containing a slice of the Vector Register File and functional units (one integer, one floating point). The vector instruction execution scheme is hybrid in-order/out-of-order and is supported by register renaming and arithmetic/memory instruction decoupling. On a stand-alone synthesis, Vitruvius+ reaches a maximum frequency of 1.4 GHz in typical conditions (TT/0.80V/25°C) using GlobalFoundries 22FDX FD-SOI. The silicon implementation has a total area of 1.3 mm 2 and maximum estimated power of ∼920 mW for one instance of Vitruvius+ equipped with eight vector lanes.