Yuri Torres
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Yuri Torres has a total of 22 co-authors. Here are the top co-authors by number of shared publications:
- Diego R. Llanos
Dpto. de Informática, Universidad de Valladolid, Valladolid, Spain
15 shared publications - Arturo Gonzalez-Escribano
Universidad de Valladolid
14 shared publications - Manuel de Castro
Universidad de Valladolid
8 shared publications - Rocío Carratalá-Sáez
University of Valencia
6 shared publications - Francisco‐José Andújar‐Muñoz
Universidad de Valladolid
5 shared publications - Roberto R. Osorio
CITIC, Computer Architecture Group, Universidade da Coruña, A Coruña, Spain
4 shared publications
Latest Publications
- Open Access
- 6 October 2026
- Published by Elsevier in SSRN Electronic Journal
EPSILOD-Het: Accelerating stencil applications in highly heterogeneous systems with a parallel skeleton
Stencil computations are a representative class of applications in HPC. In this work, we extend EPSILOD, a parallel skeleton for Iterative Stencil Loops in distributed heterogeneous systems to support Intel FPGAs. We call the extended version EPSILOD-Het. EPSILOD-Het can compute Iterative Stencil Loops concurrently on CPUs, GPUs and FPGAs, allowing the resources of highly heterogeneous systems to be combined within the same execution. We discuss the integration of FPGAs into portability layers and parallel skeletons, highlighting challenges, possible solutions, and limitations. We conduct experimental studies to assess EPSILOD-Het's performance in different aspects. Our results suggest that: (a) EPSILOD-Het achieves comparable FPGA performance for basic 2D and 3D stencils to implementations using both FPGA reference versions and programs generated with another portability layer; (b) that the FPGA implementation in EPSILOD-Het offers competitive throughput and the highest energy efficiency among the devices tested, especially for compute-intensive stencils.; and (c) that EPSILOD-Het efficiently combines GPU and FPGA resources in the tested scenarios using a static load-balancing strategy, while some CPU-inclusive executions reveal limitations. For example, by concurrently using a GPU and an FPGA, EPSILOD-Het can compute up to 115\% more domain elements than the GPU alone in the same time, or the same domain elements in up to 52.15\% less time.
- 6 October 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessEPSILOD-Het: Accelerating stencil applications in highly heterogeneous systems with a parallel skeleton
Stencil computations are a representative class of applications in HPC. In this work, we extend EPSILOD, a parallel skeleton for Iterative Stencil Loops in distributed heterogeneous systems to support Intel FPGAs. We call the extended version EPSILOD-Het. EPSILOD-Het can compute Iterative Stencil Loops concurrently on CPUs, GPUs and FPGAs, allowing the resources of highly heterogeneous systems to be combined within the same execution. We discuss the integration of FPGAs into portability layers and parallel skeletons, highlighting challenges, possible solutions, and limitations. We conduct experimental studies to assess EPSILOD-Het's performance in different aspects. Our results suggest that: (a) EPSILOD-Het achieves comparable FPGA performance for basic 2D and 3D stencils to implementations using both FPGA reference versions and programs generated with another portability layer; (b) that the FPGA implementation in EPSILOD-Het offers competitive throughput and the highest energy efficiency among the devices tested, especially for compute-intensive stencils.; and (c) that EPSILOD-Het efficiently combines GPU and FPGA resources in the tested scenarios using a static load-balancing strategy, while some CPU-inclusive executions reveal limitations. For example, by concurrently using a GPU and an FPGA, EPSILOD-Het can compute up to 115\% more domain elements than the GPU alone in the same time, or the same domain elements in up to 52.15\% less time.
- Open Access
- 2 July 2026
- Published by Springer Nature
- p. 106-118
Exploiting Highly Heterogeneous Systems with Stencil Applications
While CPUs, GPUs, and FPGAs present particular advantages for different classes of High Performance Computing Applications, Iterative Stencil Loop (ISL) is a class of parallel applications with efficient implementations for all of them. Thus, this class is an interesting case of use to test the potential of using simultaneously different classes of heterogeneous devices. EPSILOD is a parallel skeleton framework designed to easily program and deploy ISL applications on heterogeneous platforms. It provides a programming abstraction and a transparent coordination system to work with different types of devices. In this work, we improve EPSILOD and its performance portability layer to support key features for developing and operating optimized kernels on FPGAs. The new EPSILOD version allows the execution of efficient stencil programs simultaneously on CPUs, GPUs, and FPGA accelerators. We discuss how to test the computing power of each different device and use this information in an EPSILOD data-partition policy to obtain a balanced load distribution across them. We present an experimental study using a classical heat-transfer stencil example in a highly heterogeneous system to show the efficiency of EPSILOD programs when using a CPU, a GPU, and an FPGA simultaneously.
- 2 July 2026
- Published by Springer Nature
- p. 106-118
Open AccessExploiting Highly Heterogeneous Systems with Stencil Applications
While CPUs, GPUs, and FPGAs present particular advantages for different classes of High Performance Computing Applications, Iterative Stencil Loop (ISL) is a class of parallel applications with efficient implementations for all of them. Thus, this class is an interesting case of use to test the potential of using simultaneously different classes of heterogeneous devices. EPSILOD is a parallel skeleton framework designed to easily program and deploy ISL applications on heterogeneous platforms. It provides a programming abstraction and a transparent coordination system to work with different types of devices. In this work, we improve EPSILOD and its performance portability layer to support key features for developing and operating optimized kernels on FPGAs. The new EPSILOD version allows the execution of efficient stencil programs simultaneously on CPUs, GPUs, and FPGA accelerators. We discuss how to test the computing power of each different device and use this information in an EPSILOD data-partition policy to obtain a balanced load distribution across them. We present an experimental study using a classical heat-transfer stencil example in a highly heterogeneous system to show the efficiency of EPSILOD programs when using a CPU, a GPU, and an FPGA simultaneously.
- Open Access
- 12 May 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Reconfigurable Technology and Systems
- Vol. 19 (2) , 1-25
Pipelined FPGA Implementation of a Differential Evolution Engine for Optimization of Scientific Models
Custom computing machines implemented on FPGAs have emerged as a powerful solution for tackling computationally intensive tasks, leveraging their capacity for deep pipelining and parallel memory access. Differential Evolution (DE), a robust optimization algorithm, combined with adaptive numerical integration methods, is widely used to optimize parameter values in diverse scientific models. These tasks involve extensive floating-point computations, making FPGAs an ideal platform for efficiently accelerating their execution. In this work, we present a flexible and scalable FPGA architecture optimized for DE. This architecture is tailored to solve complex, resource-intensive optimization problems and is easily customizable for various models and integration methods. To demonstrate its efficacy, we evaluate two case studies: The Hodgkin–Huxley model for neuron action potentials and the Circadian clock model of Arabidopsis thaliana . Our architecture integrates adaptive numerical methods with DE and achieves significant performance and energy efficiency gains over CPU and GPU implementations while maintaining versatility across applications. Our architecture’s modular design enables seamless adaptation across different scientific contexts, enabling further optimization of resource utilization and expansion of application domains. The results underline the potential of FPGAs as a superior platform for large-scale scientific computation, offering unmatched energy efficiency and computational throughput for highly demanding tasks. The code developed to carry out this work is publicly available at https://github.com/mdccUVa/de-fpga .
- 12 May 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Reconfigurable Technology and Systems
- Vol. 19 (2) , 1-25
Open AccessPipelined FPGA Implementation of a Differential Evolution Engine for Optimization of Scientific Models
Custom computing machines implemented on FPGAs have emerged as a powerful solution for tackling computationally intensive tasks, leveraging their capacity for deep pipelining and parallel memory access. Differential Evolution (DE), a robust optimization algorithm, combined with adaptive numerical integration methods, is widely used to optimize parameter values in diverse scientific models. These tasks involve extensive floating-point computations, making FPGAs an ideal platform for efficiently accelerating their execution. In this work, we present a flexible and scalable FPGA architecture optimized for DE. This architecture is tailored to solve complex, resource-intensive optimization problems and is easily customizable for various models and integration methods. To demonstrate its efficacy, we evaluate two case studies: The Hodgkin–Huxley model for neuron action potentials and the Circadian clock model of Arabidopsis thaliana . Our architecture integrates adaptive numerical methods with DE and achieves significant performance and energy efficiency gains over CPU and GPU implementations while maintaining versatility across applications. Our architecture’s modular design enables seamless adaptation across different scientific contexts, enabling further optimization of resource utilization and expansion of application domains. The results underline the potential of FPGAs as a superior platform for large-scale scientific computation, offering unmatched energy efficiency and computational throughput for highly demanding tasks. The code developed to carry out this work is publicly available at https://github.com/mdccUVa/de-fpga .
- Open Access
- 24 October 2025
- Published by Elsevier in Journal of Parallel and Distributed Computing
- Vol. 207
On the development of high-performance, multi-GPU applications on heterogeneous systems leveraging SYCL
- 24 October 2025
- Published by Elsevier in Journal of Parallel and Distributed Computing
- Vol. 207
Open AccessOn the development of high-performance, multi-GPU applications on heterogeneous systems leveraging SYCL
- Open Access
- 11 June 2025
- Published by Springer Nature
- p. 412-424
Challenging Portability Paradigms: FPGA Acceleration Using SYCL and OpenCL
As the interest in FPGA-based accelerators for HPC applications increases, new challenges also arise, especially concerning different programming and portability issues. This paper aims to provide a snapshot of the current state of the FPGA tooling and its problems. To do so, we evaluate the performance portability of two frameworks for developing FPGA solutions for HPC (SYCL and OpenCL) when using them to port a highly-parallel application to FPGAs, using both ND-range and single-task type of kernels. The developer’s general recommendation when using FPGAs is to develop single-task kernels for them, as they are commonly regarded as more suited for such hardware. However, we discovered that, when using high-level approaches such as OpenCL and SYCL to program a highly-parallel application with no FPGA-tailored optimizations, ND-range kernels significantly outperform single-task codes. Specifically, while SYCL struggles to produce efficient FPGA implementations of applications described as single-task codes, its performance excels with ND-range kernels, a result that was unexpectedly favorable.
- 11 June 2025
- Published by Springer Nature
- p. 412-424
Open AccessChallenging Portability Paradigms: FPGA Acceleration Using SYCL and OpenCL
As the interest in FPGA-based accelerators for HPC applications increases, new challenges also arise, especially concerning different programming and portability issues. This paper aims to provide a snapshot of the current state of the FPGA tooling and its problems. To do so, we evaluate the performance portability of two frameworks for developing FPGA solutions for HPC (SYCL and OpenCL) when using them to port a highly-parallel application to FPGAs, using both ND-range and single-task type of kernels. The developer’s general recommendation when using FPGAs is to develop single-task kernels for them, as they are commonly regarded as more suited for such hardware. However, we discovered that, when using high-level approaches such as OpenCL and SYCL to program a highly-parallel application with no FPGA-tailored optimizations, ND-range kernels significantly outperform single-task codes. Specifically, while SYCL struggles to produce efficient FPGA implementations of applications described as single-task codes, its performance excels with ND-range kernels, a result that was unexpectedly favorable.
- 4 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 283
Accelerating Scientific Model Optimization with a Pipelined FPGA-Based Differential Evolution Engine
Custom computing machines on FPGAs excel in solving computationally intensive tasks by leveraging deep pipelines and parallel memory access. This paper introduces a flexible FPGA-based architecture for Differential Evolution (DE), optimized for a variety of scientific models and numerical integration methods. The architecture's modular design allows seamless customization for diverse applications. Two case studies demonstrate the architecture's capabilities: the Hodgkin-Huxley model for neuron action potentials and the Circadian model of Arabidopsis thaliana. These implementations employ double-precision floating-point arithmetic and adaptive numerical integration techniques, addressing the challenges of complex, stiff differential equations. The proposed design outperforms CPUs and GPUs in computational speed and energy efficiency, achieving up to 3.8x faster processing and significant reductions in energy consumption. This work highlights the potential of FPGA platforms for accelerating complex scientific computations while providing insights for future optimizations in resource utilization and broader applicability.
- 4 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 283
Accelerating Scientific Model Optimization with a Pipelined FPGA-Based Differential Evolution Engine
Custom computing machines on FPGAs excel in solving computationally intensive tasks by leveraging deep pipelines and parallel memory access. This paper introduces a flexible FPGA-based architecture for Differential Evolution (DE), optimized for a variety of scientific models and numerical integration methods. The architecture's modular design allows seamless customization for diverse applications. Two case studies demonstrate the architecture's capabilities: the Hodgkin-Huxley model for neuron action potentials and the Circadian model of Arabidopsis thaliana. These implementations employ double-precision floating-point arithmetic and adaptive numerical integration techniques, addressing the challenges of complex, stiff differential equations. The proposed design outperforms CPUs and GPUs in computational speed and energy efficiency, achieving up to 3.8x faster processing and significant reductions in energy consumption. This work highlights the potential of FPGA platforms for accelerating complex scientific computations while providing insights for future optimizations in resource utilization and broader applicability.
- Open Access
- 14 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 50394-50410
Comparing Portability of FPGA High-Level Synthesis Frameworks in the Context of a Highly-Parallel Application
With the growing popularity of FPGA-based accelerators in HPC applications, new challenges have emerged, particularly in terms of programming and portability. This paper provides an overview of the current state of FPGA tools and their limitations. This study evaluates the performance and portability of two frameworks, SYCL and OpenCL, for developing HPC FPGA solutions. The case of porting a highly-parallel application to FPGAs is studied. First, naïve, low-development-effort implementations are presented using both ND-range and single-task types of kernels, and their performance is evaluated. Subsequently, an optimized FPGA-centric approach is presented and assessed using metrics from the compilation framework. Finally, the different approaches presented are implemented using OpenCL and SYCL and their performance is evaluated Results reveal that ND-range kernels offer high portability for highly parallel applications, while single-task codes exhibit significantly lower portability. Additionally, SYCL struggles to generate efficient hardware architectures for this kind of application when described as single-task codes, although its performance when following the ND-range approach is surprisingly high.
- 14 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 50394-50410
Open AccessComparing Portability of FPGA High-Level Synthesis Frameworks in the Context of a Highly-Parallel Application
With the growing popularity of FPGA-based accelerators in HPC applications, new challenges have emerged, particularly in terms of programming and portability. This paper provides an overview of the current state of FPGA tools and their limitations. This study evaluates the performance and portability of two frameworks, SYCL and OpenCL, for developing HPC FPGA solutions. The case of porting a highly-parallel application to FPGAs is studied. First, naïve, low-development-effort implementations are presented using both ND-range and single-task types of kernels, and their performance is evaluated. Subsequently, an optimized FPGA-centric approach is presented and assessed using metrics from the compilation framework. Finally, the different approaches presented are implemented using OpenCL and SYCL and their performance is evaluated Results reveal that ND-range kernels offer high portability for highly parallel applications, while single-task codes exhibit significantly lower portability. Additionally, SYCL struggles to generate efficient hardware architectures for this kind of application when described as single-task codes, although its performance when following the ND-range approach is surprisingly high.
- Open Access
- 21 February 2025
- Published by Qeios Ltd
Review of: "Performant Automatic BLAS Offloading on Unified Memory Architecture with OpenMP First-Touch Style Data Movement"
GENERAL EVALUATION The paper "PERFORMANT AUTOMATIC BLAS OFFLOADING ON UNIFIED MEMORY ARCHITECTURE WITH OPENMP FIRST-TOUCH STYLE DATA MOVEMENT" presents an optimization for automatic BLAS offloading in unified memory architectures using the OpenMP Fi...
- 21 February 2025
- Published by Qeios Ltd
Open AccessReview of: "Performant Automatic BLAS Offloading on Unified Memory Architecture with OpenMP First-Touch Style Data Movement"
GENERAL EVALUATION The paper "PERFORMANT AUTOMATIC BLAS OFFLOADING ON UNIFIED MEMORY ARCHITECTURE WITH OPENMP FIRST-TOUCH STYLE DATA MOVEMENT" presents an optimization for automatic BLAS offloading in unified memory architectures using the OpenMP Fi...
- Open Access
- 27 June 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Computer
- Vol. 57 (7) , 66-76
The Role of Field-Programmable Gate Arrays in the Acceleration of Modern High-Performance Computing Workloads
Reconfigurable hardware circuits, such as field-programmable gate arrays, have gained popularity in the high-performance computing (HPC) community in recent years. Nevertheless, their real contribution to accelerating HPC workloads is unclear in both potential and extent.
- 27 June 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Computer
- Vol. 57 (7) , 66-76
Open AccessThe Role of Field-Programmable Gate Arrays in the Acceleration of Modern High-Performance Computing Workloads
Reconfigurable hardware circuits, such as field-programmable gate arrays, have gained popularity in the high-performance computing (HPC) community in recent years. Nevertheless, their real contribution to accelerating HPC workloads is unclear in both potential and extent.
- Open Access
- 15 April 2024
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 80 (11) , 16630-16653
Performance improvement of the triangular matrix product in commodity clusters
There are many works devoted to improving the matrix product computation, as it is used in a wide variety of scientific applications arising from many different fields. In this work, we propose alternative data distribution policies and communication patterns to reduce the elapsed time when computing triangular matrix products in distributed memory environments. In particular, we focus on commodity clusters, where the number of nodes is limited, proposing alternatives to traditional approaches in order to improve this operation’s performance. Our proposal overcomes the performance results associated with the state-of-the-art libraries, such as ScaLAPACK and SLATE, offering execution times that are up to 30% faster.
- 15 April 2024
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 80 (11) , 16630-16653
Open AccessPerformance improvement of the triangular matrix product in commodity clusters
There are many works devoted to improving the matrix product computation, as it is used in a wide variety of scientific applications arising from many different fields. In this work, we propose alternative data distribution policies and communication patterns to reduce the elapsed time when computing triangular matrix products in distributed memory environments. In particular, we focus on commodity clusters, where the number of nodes is limited, proposing alternatives to traditional approaches in order to improve this operation’s performance. Our proposal overcomes the performance results associated with the state-of-the-art libraries, such as ScaLAPACK and SLATE, offering execution times that are up to 30% faster.