RT Conference Proceedings
T1 Configurable Ultra-High-Throughput QRD FPGA Accelerators for small matrices.
A1 Hormigo Jiménez, Pablo
A1 Hormigo-Aguilar, Javier
K1 Algoritmos computacionales
K1 Factorización de operadores
K1 Matrices lógicas programables por el usuario
AB QR decomposition is an essential operation in matrixalgebra that is applicable in many fields, such as signalprocessing, automatic control, communications, and physics simulations.The QRD computation is the system’s bottleneck inmany of these applications. This paper presents a configurableultra-high-throughput accelerator for FPGAs designed usingHigh-Level Synthesis language. The accelerator is arranged asa 2D-systolic array of Givens rotators based on the CORDICalgorithm. Its dimension can be configured at compilation timeto fit the matrix size. Similarly, the degree of parallelism in therotators can be configured, such as the computation throughputgoes from one n × n matrix every 2n clock cycles up to onematrix every clock cycle.
PB IEEE
YR 2025
FD 2025-12-11
LK https://hdl.handle.net/10630/41156
UL https://hdl.handle.net/10630/41156
LA eng
NO https://conferences.ieeeauthorcenter.ieee.org/author-ethics/guidelines-and-policies/post-publication-policies/#accepted (24 meses de embargo)
DS RIUMA. Repositorio Institucional de la Universidad de Málaga
RD 20 ene 2026