PalGPU: A Cost-Efficient Shared CPU and GPU Computing Platform for AI and Scientific Workloads in Palestine

Description

PalGPU is a proposed shared computing platform that aims to improve access to affordable CPU and GPU resources in Palestine. It will connect GPU-enabled computers contributed by individuals and organizations and provide access to designated CPU resources at the university’s Centre of Excellence. Through a centralized web application, users will submit supported AI and scientific computing workloads. The platform will provide resource registration, job queues, resource-aware scheduling, monitoring, execution history, cost estimation, budget alerts, and GPU provider earnings tracking. A companion mobile application will support job monitoring and notifications. The project will support parallel CPU workloads and investigate multi-node CPU execution and multi-GPU training where infrastructure permits. Evaluation will assess performance, resource utilization, computing costs, and technical and economic feasibility in Palestine, including potential demand and comparison with selected international cloud services.