24-05-2026
Supercell v2.1.2 (maintenance release) Mostly a maintenance update: dependencies and packages were refreshed, several minor bugs were fixed, and benchmarks/tests show performance improved by almost 30%. New native builds are now published for Linux ARM64 (Ubuntu) and for macOS Intel and Apple Silicon separately. The x86_64 Linux binary from the site targets AVX-capable CPUs (any x86_64 from 2011 onward) and is built to run on Ubuntu 18.04+, Debian 10+, RHEL 8+ and openSUSE Leap 15.x+. Download here.
03-08-2024
Supercell v2.1.1 (minor improvement) External dependencies updated, minor bugs fixed. Download program here.
03-11-2022
Supercell v2.1.0 A new version of supercell program released with improvement of electrostatic calculation precision and performance. New multi-threading approach allows effectively parallelize program up to 64 threads. Please check benchmark for relevant information.
10-05-2022
New supercell site and packages Welcome to the new supercell site: a lot of useful information has been added. The program itself is now available via snap and for the Linux ARM environment. See download.
02-01-2022
Supercell v2.0.2 (minor bugfix) Symmetry search in very sheared cells is improved. Gemmi and xxHash updated.
20-05-2021
Supercell v2.0 A new version of supercell program with major improvements: performance, portability and new parameters. Please check changelog and benchmark.
23-05-2019
Supercell v1.2 (performance increase) The new version of supercell program is around 4 times faster, than the previous one!
09-02-2019
Microsoft Windows support (experimental) I would like to announce Microsoft Windows platform support. You can download the binary below. An output structures packing is not supported (-a option). The binary is tested on Windows 7 and Windows 10.
24-01-2019
Supercell v1.1 The maximum limit for processing structures has been increased. The new version of supercell program can process up to 1015 total structures. This value is far beyond a reasonable limit due to calculation time. The average program performance is about 10-100 billion structures per day with a standard desktop processor. Feedback is very welcome.