SPONGEMM

Getting Started

SPONGE 2.0 documentation: Getting Started.

Organized by GPT-6-Astra. SPONGE 2.0 is under active, iterative development; features, parameters, and documentation may change. Consult the GitHub repository’s docs and skills for the latest version information.

Source snapshot for this page.

Getting Started

SPONGE (Simulation Package Toward Next GEneration of molecular modelling) is a GPU-accelerated molecular dynamics simulation engine supporting CUDA, HIP (AMD GPU / Hygon DCU), and various CPU SIMD backends.

Install pixi

SPONGE uses pixi to manage dependencies and build workflows.

Linux / macOS

curl -fsSL https://pixi.sh/install.sh | bash

China mirror:

curl -fsSL https://conda.spongemm.cn/pixi/install.sh | bash

Windows

powershell -ExecutionPolicy ByPass -c "irm https://pixi.sh/install.ps1 | iex"

China mirror:

powershell -ExecutionPolicy ByPass -c "irm https://conda.spongemm.cn/pixi/install.ps1 | iex"

Install from binary distributions

If you want to use a prebuilt SPONGE binary, install one of the published packages instead of building from source.

Choose a package

PackageHardwarePlatforms
sponge-cuda13NVIDIA GPU (driver >= 570)Linux x86_64, Windows x64
sponge-cuda12NVIDIA GPU (driver >= 525)Linux x86_64, Windows x64
sponge-cpuCPU onlyLinux x86_64 / aarch64, Windows x64, macOS ARM64
sponge-cpu-mpiCPU + MPILinux x86_64 / aarch64

The commands below use sponge-cpu as an example. Replace it with another package from the table when appropriate.

Install in a Pixi project

Create a Pixi project, add the SPONGE channel, and install the package into the project environment:

pixi init sponge-project
cd sponge-project
pixi project channel add https://conda.spongemm.cn
pixi add sponge-cpu
pixi run SPONGE -v

If you are already in a Pixi project that uses conda-forge, only the last three commands are needed.

Install globally with Pixi

Install SPONGE into a Pixi-managed global environment when you want to run it outside a project:

pixi global install \
  --channel https://conda.spongemm.cn \
  --channel conda-forge \
  sponge-cpu
SPONGE -v

Optional: Build SPONGE from source

Most users should install a binary distribution above. To compile SPONGE from source or set up a development environment, follow the Build Guide.

Run a simulation

Prepare a TOML input file mdin.spg.toml:

md_name = "NVT water"
mode = "nvt"
step_limit = 50000
dt = 0.002
cutoff = 8.0
default_in_file_prefix = "WAT"
constrain_mode = "SHAKE"
target_temperature = 300.0
write_information_interval = 1000

[thermostat]
mode = "middle_langevin"
tau = 0.1
seed = 2026

Use the command that matches how SPONGE was installed.

For a global installation:

SPONGE -mdin mdin.spg.toml

For an installation in a Pixi project:

pixi run SPONGE -mdin mdin.spg.toml

For an optional source build in a development environment:

pixi run -e dev-cuda13 SPONGE -mdin mdin.spg.toml

You can also enter that development environment first:

pixi shell -e dev-cuda13
SPONGE -mdin mdin.spg.toml

Run benchmarks

The benchmark tasks are available from a source checkout with a development environment installed:

pixi run -e dev-cuda13 perf-amber       # AMBER force field performance
pixi run -e dev-cuda13 perf-nonortho    # non-orthogonal box
pixi run -e dev-cuda13 vali-thermostat  # thermostat validation
pixi run -e dev-cuda13 vali-barostat    # barostat validation

Next steps