Raúl Murillo: “Reducir datos y acelerar cálculos con menos energía”

Raúl Murillo: “Reducir datos y acelerar cálculos con menos energía”

by Raúl Murillo

Formal & Physical Sciences Computing & Cybersecurity UYComputer scienceUYFComputer architecture and logic design
🎙 Raúl Murillo 👥 20K 📅 October 24, 2025 ⏱ 15 min 👁 111 📄 expert opinion 🧭 2026-08-06
Available in: English (current) Français

Keywords

computer arithmeticdecimal floating-pointhardware acceleratorsdeep neural networksenergy efficiency

Summary

In this interview, Raúl Murillo, winner of the 2025 SCIE-Fundación BBVA Young Researchers Award, discusses his research on alternative arithmetic formats for computers. He explains that his work focuses on decimal arithmetic, which is more complex than binary due to the representation of fractional numbers. He has developed new formats and hardware circuits to perform operations more efficiently, reducing data size and energy consumption. He highlights applications in scientific computing and AI, particularly deep neural networks, where many operations can be optimized. He addresses concerns about AI risks, noting ongoing efforts in academia, industry, and government to mitigate them. He also discusses challenges in adopting new standards and the difficulty of hardware design. He shares his motivation for research, emphasizing the ‘Eureka’ moments and the importance of perseverance and collaboration. He comments on the state of computer science in Spain, noting high talent but challenges in academic careers compared to industry. Finally, he underscores the importance of computer science in the 21st century, comparing it to the wheel or steam engine.

171 words

Critical Evaluation

The interview provides a clear and insightful overview of Raúl Murillo’s research in computer arithmetic, a specialized but crucial area for improving computational efficiency. The value of the information lies in its explanation of how alternative arithmetic formats can reduce data size and accelerate calculations, directly impacting energy consumption in AI and scientific computing. The argumentation is solid, as Murillo logically connects his technical work to broader societal benefits, such as more efficient AI systems. The scientific rigor is high, given that he is an award-winning researcher and his explanations are technically accurate. However, the video lacks external sources or references to specific publications, which limits the ability to verify claims independently. The title accurately reflects the content, focusing on reducing data and accelerating calculations with less energy. The interview also touches on important issues like AI risks and the state of research in Spain, providing a well-rounded perspective. The main weakness is the absence of detailed technical depth, as the format is a short interview rather than a technical talk. Overall, the video is a valuable introduction to the field, suitable for those interested in the intersection of hardware and AI efficiency.

193 words

Title / Content Match

The title accurately reflects the core theme of reducing data size and accelerating computations with lower energy consumption.

Quality & Reliability

8/10

The video features an award-winning researcher explaining his own work in computer arithmetic, with clear technical explanations and no obvious errors. However, it is an interview without external sources or peer-reviewed references, limiting verifiability.

Chapters

Cited Sources

  • Fundación BBVA — Official website of the foundation that awards the prize and hosts the interview.
  • Fundación BBVA LinkedIn — LinkedIn page of the foundation, providing additional context about their initiatives.

Concurring Sources

  • Fundación BBVA — The foundation's website likely contains information about the award and the researcher's profile, aligning with the video's content.

Contribution & Novelties

The video presents Raúl Murillo’s original contributions to alternative arithmetic formats, which are novel in their integration into hardware and neural networks. This work is at the forefront of improving computational efficiency.

Pour aller plus loin :

  • IEEE 754 — Standard for floating-point arithmetic, relevant to understanding the context of alternative formats.
  • Deep learning — Neural networks are a key application area for the efficiency gains discussed.
  • Computer arithmetic — Overview of the field, providing background on the challenges and approaches.

81 words

Radar Profile

The radar profile shows high scores in information quality and reliability, with moderate scores in quantity and technical level, indicating a focused and credible interview with room for more depth.

Reliability 8/10