Photonic Processing Units (PPUs) for Cloud Computing: Architectural Challenges and Framework Design

Philip, Roshin (2025) Photonic Processing Units (PPUs) for Cloud Computing: Architectural Challenges and Framework Design. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (919kB) | Preview

Abstract

Photonic Processing Units (PPUs) have emerged as a promising solution to overcome the latency and energy limitations of traditional electronic accelerators in AI workloads. However, their integration into cloud-native environments remains a significant architectural challenge. This research extends the CloudSim 3.0 simulation toolkit with a custom DelayBroker to enable end-to-end modeling—including data-in, computation, and data-out phases—under both zero-delay and realistic network-delay (10 ms one-way, 1Gb/s) scenarios. A matrix multiplication kernel was used as the benchmark, executed across varied matrix sizes (2000×2000, 3000×3000, 5000×5000) to analyze sensitivity to workload scaling. Simulations were performed on three architectures: CPU (5M MIPS), GPU (15M MIPS), and PPU (138M MIPS, derived from Chen et al.’s 9.2× optical speedup). The results confirm significant performance gains for PPUs, showing up to 21.9× speedup over CPU and 7.4× over GPU in compute-only runs. These findings demonstrate the feasibility of photonic acceleration in cloud contexts and provide a foundation for future hybrid scheduling frameworks.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Gupta, Punit UNSPECIFIED
Uncontrolled Keywords:	Photonic Processing Unit; CloudSim; DelayBroker; AI acceleration; MIPS scaling; matrix workload; network delay
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence T Technology > T Technology (General) > Information Technology > Cloud computing
Divisions:	School of Computing > Master of Science in Cloud Computing
Depositing User:	Ciara O'Brien
Date Deposited:	30 Mar 2026 13:36
Last Modified:	30 Mar 2026 13:36
URI:	https://norma.ncirl.ie/id/eprint/9254

Actions (login required)

View Item