ManyGUI: A Graphical Tool to Accelerate Many-core Debugging Through Communication, Memory, and Energy Profiling - Pôle Software and Hardware, ARchitectures and Processes Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

ManyGUI: A Graphical Tool to Accelerate Many-core Debugging Through Communication, Memory, and Energy Profiling

Résumé

The debugging and validation of the many-core design is a complex task due to numerous events happening in the system simultaneously. Current state-of-the-art many-cores are strongly based on waveforms and log files to validate their behavior during simulation. Our hypothesis is that, as happened with ASIC development, a Graphical User Interface (GUI) can significantly accelerate the many-core development. To sustain that, we propose an open-source GUI tool called ManyGUI for many-core debugging. ManyGUI is organized in a framework that collects and classifies high-level events during simulation related to computation (executed CPU instructions), memory, and communication (NoC packets). Such events are shown graphically to the developer through a set of intuitive and practical frames. We evaluate ManyGUI in a silicon-proven state-of-the-art open-source manycore called OpenPiton, which uses RISC-V 64-bits CPU, 3 NoCs, and a distributed/shared cache memory organization. Results show that ManyGUI allows the developer to rapidly obtain a comprehensive view of the many-core behavior in terms of communication statistics (packets paths, link utilization), memory statistics (memory access, miss rate), and energy (CPU, memory, and NoC). CCS CONCEPTS • Hardware → Simulation and emulation; • Computer systems organization → Multicore architectures.
Fichier principal
Vignette du fichier
ManyGUI_RAPIDO22.pdf (1.75 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03704278 , version 1 (24-06-2022)

Identifiants

Citer

Marcelo Ruaro, Kevin J. M. Martin. ManyGUI: A Graphical Tool to Accelerate Many-core Debugging Through Communication, Memory, and Energy Profiling. DroneSE and RAPIDO '22: System Engineering for constrained embedded systems, Jun 2022, Budapest, Hungary. pp.39-46, ⟨10.1145/3522784.3522791⟩. ⟨hal-03704278⟩
28 Consultations
46 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More