Narcís Rodas

Narcís Rodas

Auto-generated profile, based on publicly available information

Narcís Rodas

Barcelona Supercomputing Center, Barcelona, Spain

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
21 November 2024 in IEEE Transactions on Computers
Verified
Journal Article
Closed access
JFJordi Fornt
Jordi Fornt
EREnrico Reggiani
Enrico Reggiani
PFPau Fontova-Musté
Pau Fontova-Musté
NRNarcís Rodas
Narcís Rodas
APAlessandro Pappalardo
Alessandro Pappalardo
Jaume AbellaJaume Abella
Jaume Abella
Efficiently computing Deep Neural Networks (DNNs) has become a primary challenge in today's computers, especially on devices targeting mobile or edge applications. Recent progress on Post-Training Quantization (PTQ) and Quantization-Aware Training (QAT) has shown that the key to high energy efficiency lies in executing deep learning models with low- (8- to 5-bit) or ultra-low-precision (4- to 2-bit). Unfortunately, current Central Processing Unit (CPU) architectures and Instruction Set Architectures (ISAs) present severe limitations on the range of data sizes supported to compute DNN kernels. In this work, we present Mix-GEMM, a hardware-software co-designed architecture that enables RISC-V processors to efficiently compute arbitrary mixed-precision DNN kernels, supporting all data size combinations from 8- to 2-bit. By applying binary segmentation, our architecture can scale its throughput by decreasing the data size of the operands, resulting in a flexible approach capable of leveraging state-of-the-art QAT and PTQ to achieve high energy efficiency at a very low cost. Evaluating our Mix-GEMM architecture in a dual-issue in-order RISC-V processor shows that we are able to boost its performance and energy efficiency by up to $44\times$ and $11\times$ with respect to the baseline processor, with an area overhead of only 2%. This allows our extended processor to execute state-of-the-art DNNs with significantly higher performance and energy efficiency than the standard FP32 precision, while retaining almost the same model accuracy.

17 October 2025
Verified
Conference Paper
Open access
MNMohammadreza Esmali Nojehdeh
Mohammadreza Esmali Nojehdeh
HMHossein Mokhtarnia
Hossein Mokhtarnia
JPJulian Pavon
Julian Pavon
NRNarcís Rodas
Narcís Rodas
RBRoger Figueras Bagué
Roger Figueras Bagué
EAEduard Ayguadé
Eduard Ayguadé