V-Help
← All news
Hardware

Scientists Turn Old Smartphones into Data Center Clusters

Scientists Turn Old Smartphones into Data Center Clusters

Photo: Tom's Hardware

Quick answer

Researchers built a cluster using outdated Pixel smartphones by replacing Android with Linux and removing unnecessary components.

Researchers from the University of California, San Diego (UCSD), in partnership with Google, have proposed an innovative approach to recycling outdated smartphones. As part of the project, old Pixel devices were repurposed into computing clusters for data centers. The primary goal is to reduce electronic waste while creating budget-friendly alternatives to traditional servers.

Scientists discovered that even three-year-old smartphone processors deliver higher single-threaded performance than server-grade CPUs in SPEC benchmarks. To build the cluster, displays, batteries, cameras, and other unnecessary components are removed, and the Android OS is replaced with Linux. This enables the use of orchestrators like Kubernetes and optimized workload management.

Testing revealed that 25–50 smartphones provide computing power equivalent to a dual-socket server CPU. A cluster of 20 devices can support applications for a class of 75 students, while 2,000 smartphones can simultaneously serve hundreds of such groups. The project is particularly relevant for educational institutions with limited IT infrastructure budgets.

While promising, this solution is not suitable for large-scale AI data centers, which require specialized servers with high reliability. However, for universities and small organizations, it offers an effective way to cut costs and carbon footprint. The full system is expected to launch by the end of 2024.

Common questions

Why are scientists using old smartphones for data centers?
Smartphones contain high-performance SoCs that surpass server CPUs in single-threaded power. Recycling reduces e-waste and infrastructure costs for educational and small-scale projects.
What advantages does a smartphone cluster offer?
It is cheaper than traditional servers, operates locally without cloud expenses, and lowers environmental impact. Ideal for universities and budget-constrained organizations.
Can such clusters be used for AI?
No, AI data centers require specialized servers with high reliability and scalability. Smartphones lack the stability needed for large-scale computations.
Share:

Dzen feed: /feed/dzen.xml · RSS: /feed.xml

Why trust this

Prepared by the V-Help editorial team from the primary source with a published date.

Published by: V-Help.ru news desk

Source: Tom's Hardware