Name: Pedro Valero-Lara
Pronouns: he/him/his
Biography:
I am Pedro Valero-Lara, a computer scientist in the Programming Systems Group at Oak Ridge National Laboratory. My interests focused on parallel programming models and math libraries, as they are an essential component in the scientific software ecosystem, and more recently AI. My work addresses performance portability and programming productivity challenges for better scientific software across increasingly diverse and complex heterogeneous HPC architectures. I am a co-principal investigator of the DOE seedling project: Stewardship for Programming Systems and Tools. I have contributed to several DOE-funded software: Kokkos, PLASMA, and Cray LibSci math libraries. My contribution to Kokkos led to the best paper award at the IEEE/ACM Supercomputing 2022 WACCPD workshop. Before ORNL, I worked at Cray Inc., where I was the main developer of the LibSci-Acc library. Before Cray, I worked at the Barcelona Supercomputing Center (BSC), where I founded and led the Linear Algebra and Math Libraries team which contributed to the OpenMP Super Scalar programming model, OmpSs. After that, I joined NVIDIA in which I co-developed the new High-Performance Linpack benchmark that benefits from the use of GPU AI cores. All this effort led to the IEEE Early Career Researcher Award for Excellence in HPC in 2020.
Institution/Lab: Oak Ridge National Laboratory
Website: https://www.ornl.gov/staff-profile/pedro-valero-lara
SRP Collaboration Topic/Title: Challenges/Opportunities for the Extreme Heterogeneity and HPC-AI era
Field or research area: High Performance Computing
Please select all the topical areas that apply to your project:
Computer Science (i.e., architectures, compilers/languages, networks, workflow/edge, experiment automation, containers, neuromorphic computing, programming models, operating systems, sustainable software); High-Performance Computing
Brief Abstract:
This project is aimed at the implementation, evaluation and optimization of novel performance portable and heterogeneous programming solutions for the upcoming extreme heterogeneity era in computing. The project includes the design of novel software solutions on the available software and hardware platforms. Learning objectives for the applicant include: i) develop HPC codes based on performance portable programming models, such as OpenMP/OpenACC, C++, Kokkos, Julia, task-based runtimes, on current HPC and heterogeneous (CPU+GPU) architectures, ii) acquire skills in both, software solutions and HPC codes implementation, iii) gain experience in performance analysis on HPC architectures.
Desired relevant skills, background, or interests:
Computer Science Programming Languages Linux
Other comments:
Do any special requirements apply? In-Person Only; Permanent Resident OK
Other, specify:
Keywords:
High-Performance Computing Parallel Programming AI Heterogeneity
Lightning Talk Title: Challenges/Opportunities for the Extreme Heterogeneity and HPC-AI era