Research
I spent a decade in computer architecture research before moving into industry. My work focused on GPUs, memory systems, and task parallelism.
Research Interests
Industry Research
Ericsson
At Ericsson, I led the development of 5G digital twins using NVIDIA Omniverse. We built simulation tools for radio wave propagation in realistic 3D environments — enabling real-time prediction of how antennas behave in different conditions.
This work was:
- Featured in NVIDIA's GTC Keynote (twice)
- Presented directly to Jensen Huang
- Covered by NVIDIA Blog, Ericsson Blog, and industry press
I worked closely with NVIDIA's Omniverse team as an early adopter, contributing to the platform's development for industrial simulation use cases.
Academic Research
Uppsala University
I completed my PhD with the Uppsala Architecture Research Team (UART), one of Europe's leading computer architecture groups. My research explored how task-based parallel programsinteract with memory systems, and how to optimize scheduling for better performance and energy efficiency.
I also developed tools for analyzing graphical workloads on tile-based GPUs — work that was recognized with multiple best paper awards.
Advisors: David Black-Schaffer and Erik Hagersten
FaMAF, Universidad Nacional de Córdoba
As a member of the GPGPU Computing Group, I conducted interdisciplinary research on GPGPU computing tailored to physical simulations using CUDA.
My key achievement was developing the world's first GPU-powered Parallel Adaptive Mesh Refinement (AMR) framework for large-scale physical simulations. This work was presented at several international conferences.
Advisors: Prof. Oscar Reula (PhD) and Prof. Nicolás Wolovick (PhD)
Academic Leadership & Teaching
Throughout my academic career, I contributed to education at both undergraduate and graduate levels, designing courses and mentoring students in computer architecture and systems programming.
- Main Instructor / Course Designer — Uppsala University (2015–2018)
- Advanced Computer Architecture (PhD/MSc)
- Operating Systems II (MSc)
- Teaching Assistant — Uppsala University & Universidad Nacional de Córdoba (2010–2018)
- Computer Architecture
- Operating Systems
- Parallel Programming
- Introduction to Computing
Research Software & Tools
I developed several tools for analyzing memory behavior and performance in parallel and graphical workloads.
GLTraceSim
A trace-driven simulator for analyzing graphical workloads on tile-based GPUs. Enables detailed memory and performance analysis of graphics rendering pipelines.
Tech: C++, Python, OpenGL
TaskInsight
A framework for understanding how task scheduling decisions affect memory behavior and application performance in task-based parallel programs.
Tech: C++, Intel PIN, OmpSs
StatTask
A statistical tool for reuse distance analysis in task-based applications, enabling cache performance prediction without cycle-accurate simulation.
Tech: C++, Python, Statistical Modeling
Theses
PhD Thesis (2018)
Understanding Task Parallelism: Providing Insight about Scheduling, Memory, and Performance for CPUs and Graphics
Uppsala University
Licentiate Thesis (2017)
Modeling the Interactions Between Tasks and the Memory System
Uppsala University
MSc Thesis (2013)
A GPU-powered Parallel Adaptive Mesh Refinement Framework for Large-Scale Physical Simulations
Universidad Nacional de Córdoba — Advisors: Prof. Oscar Reula, Prof. Nicolás Wolovick
Selected Publications
Awards
| Paper | Venue | Year | Award |
|---|---|---|---|
| Tail-PASS: Resource-based Cache Management for Tiled Graphics Rendering Hardware | IEEE ISPA | 2018 | Best Paper Award |
| Behind the Scenes: Memory Analysis of Graphical Workloads on Tile-based GPUs | IEEE ISPASS | 2018 | Best Paper Award |
| Understanding the Interplay Between Task Scheduling, Memory and Performance | ACM SPLASH-SRC | 2017 | Best Paper Award |
| How Does Memory Affect the Performance of Tasks? | ISC High Performance | 2017 | Best Poster Award |
| Rapid Performance Prediction for Out-of-Order Cores | IEEE IPDPS | 2016 | Best Poster Award |
| StatTask: Reuse Distance Analysis for Task-Based Applications | RAPIDO @ HiPEAC | 2015 | Best Paper Award |
Journal Articles
- Sampled Simulation of Task-Based Programs
T Grass, TE Carlson, A Rico, G Ceballos, et al.
IEEE Transactions on Computers, 2018 - Analyzing Performance Variation of Task Schedulers with TaskInsight
G Ceballos, T Grass, A Hugo, D Black-Schaffer
Parallel Computing, 2018 - Exploring Scheduling Effects on Task Performance with TaskInsight
G Ceballos, A Hugo, E Hagersten, D Black-Schaffer
Supercomputing Frontiers and Innovations, 2017
Conference Papers
- Tail-PASS: Resource-based Cache Management for Tiled Graphics Rendering Hardware — IEEE ISPA, 2018
- Behind the Scenes: Memory Analysis of Graphical Workloads on Tile-based GPUs — IEEE ISPASS, 2018
- TaskInsight: Understanding Task Schedules Effects on Memory and Performance — PMAM @ HPCA, 2017
- Analyzing Graphics Workloads on Tile-Based GPUs — IEEE IISWC, 2017
- Formalizing Data Locality in Task Parallel Applications — ICA3PP, 2016
- StatTask: Reuse Distance Analysis for Task-Based Applications — RAPIDO @ HiPEAC, 2015
Academic Profiles
Service to the Research Community
- Journal Reviewer — IEEE Transactions on Parallel and Distributed Systems (TPDS), Cluster Computing
- Program Committee — RAPIDO Workshop @ HiPEAC
- Industry Judging — NVIDIA Omniverse Developer Contest
Honors
- Heidelberg Laureate Forum (2018) — Selected among 200 participants worldwide
- Best GPA in Computer Science (2013) — Universidad Nacional de Córdoba
- Google Student Ambassador (2012) — Universidad Nacional de Córdoba