Sustainable Software for Science: Happy Earth Day!
As scientists, we’re interested in getting the most out of our data, but as people, we’re also interested in working in the most environmentally responsible ways. Large-scale analysis has the potential to use a lot of energy and to demand high-end hardware, both of which are concerns for the environment. Fortunately, a lot of the features that make CellEngine fast also make it more sustainable. In total, CellEngine only generates about 65% as much CO₂ as a typical home.
Scaling for Sustainability
Sustainable software design entails using less hardware, and using hardware more efficiently. CellEngine operates entirely in the cloud, in low-CO₂ datacenters powered by at least 87% carbon-free energy. The direct reduction of carbon emissions is increased by reducing the overall demand for hardware, because CellEngine users don’t need powerful machines to run algorithms or other heavy-demand uses.
Most of our users don’t have access to computing clusters that would be needed to process thousands of files or millions of events rapidly. Centralization reduces demand for hardware components, reducing the need for rare metals. CellEngine also works to use resources responsibly, scaling certain operations dynamically based on load. This approach frees up computing hardware for other uses when not needed for flow cytometry analysis. This provides an environmental advantage over many research computing clusters, which require resources to build and maintain, but oftentimes sit idle.
Optimizing for Efficiency
Efficiency has been an essential design goal since development began. Due to our continuing optimization efforts over the years, CellEngine uses the same amount of computing hardware today as it did in 2018, despite the number of users increasing significantly. One of the major optimizations CellEngine leverages is single-instruction, multiple-data (SIMD) processing to perform calculations on 16 cells at a time. This processing method uses several times less power per calculation than typical applications, while also reducing the required computing hardware by roughly 16x.
Optimization also frees up another valuable resource: researcher time. Optimizing algorithms and adding tools like automatic gate adjustment increase the amount of time researchers have to review data and complete other tasks.
Conclusion
Sometimes there are conflicts between the ideal solution for research needs and the best approach for the planet. CellEngine’s use of the cloud, heavy optimization, and scaling make it friendly not just for researchers, but also the planet.
If you want to test CellEngine out for yourself, you can sign up for a free trial at https://cellengine.com.