The need for reproducibility requires the ability to use containers to move applications from system to system.[6]
Using Singularity containers, developers can work in reproducible environments of their choosing and design, and these complete environments can easily be copied and executed on other platforms.[7]
In 2021 the Singularity open source project split into two projects called Apptainer and SingularityCE.
By the end of 2016, many developers from different research facilities joined forces with the team at Lawrence Berkeley National Laboratory to further the development of Singularity.[10]
Singularity quickly attracted the attention of computing-heavy scientific institutions worldwide:[11]
Stanford University Research Computing Center deployed Singularity on their XStream[12][13] and Sherlock[14] clusters
Various sites of the Open Science Grid Consortium including Fermilab started adopting Singularity;[17] by April 2017, Singularity was deployed on 60% of the Open Science Grid network.[18]
For two years in a row, in 2016 and 2017, Singularity was recognized by HPCwire editors as "One of five new technologies to watch".[19][20]
In 2017 Singularity also won the first place for the category "Best HPC Programming Tool or Technology".[20]
In February 2018 the Sylabs[22] company, founded by the Singularity author, was announced [23] to provide commercial support for Singularity. In October of that year Sylabs released version 3.0.0[24] which was a rewrite in the Go programming language.
Apptainer / Singularity split
In May 2020 Gregory Kurtzer left Sylabs but retained leadership of the Singularity open source project.[25] In May 2021 Sylabs made a fork of the project[26] and called it SingularityCE.[27][28] In November 2021 the Singularity open source project joined the Linux Foundation[29] and was renamed to Apptainer.[30]
Features
Singularity is able to support natively high-performance interconnects, such as InfiniBand[31] and Intel Omni-Path Architecture (OPA).[32]
Similar to the support for InfiniBand and Intel OPA devices, Singularity can support any PCIe-attached device within the compute node, such as graphic accelerators.[33]
Singularity also has native support for Open MPI library by utilizing a hybrid MPI container approach where OpenMPI exists both inside and outside the container.[31]
These features make Singularity increasingly useful in areas such as machine learning, deep learning and most data-intensive workloads where the applications benefit from the high bandwidth and low latency characteristics of these technologies.[34]
Using other enterprise container solutions like Docker in HPC systems would require modifications to the software.[35]
Docker containers can be automatically converted to stand-alone singularity files which can then be submitted to HPC resource managers.[36]
Singularity seamlessly integrates with many resource managers[37] including:
