Skip to content

Latest commit

 

History

History
64 lines (48 loc) · 2.96 KB

README.md

File metadata and controls

64 lines (48 loc) · 2.96 KB

Aeon Compute Engine

This is a compute engine for the Aeon tool (http://biodivine.fi.muni.cz/aeon/).

If you downloaded a pre-built binary, you can just run it to start the engine. The binary will output current address and port on which the compute engine is running. Later, when performing actions, there will some additional logging output. The engine will automatically run on localhost:8000 which is the default setting for the online client. Hence once the engine is running, you should be able to connect immediately.

If for some reason, you can't run the engine on localhost:8000, you can configure the address and port using environmental variables AEON_ADDR and AEON_PORT. Then you have to input this address and port also into the online client when connecting.

To configure the amount of parallel workers that are used by the binary, you can use the environmental variable RAYON_NUM_THREADS (by default, the number of workers will be 2x the available CPUs, but not all workers are necessarily used during all computations).

Note that some computations may require substantial amount of RAM.

More information about the operation of Aeon can be found in the official manual at http://biodivine.fi.muni.cz/aeon/manual.pdf

Building from source

To build the engine from source, you need a nightly Rust toolchain. Once you install Rust, you can use rustup default nightly to configure the default toolchain. Then you can simply run cargo build --release in project root directory. The binary will be located in target/release/biodivine-aeon-server.

Standalone experiments

To run analysis of a single model in command line, you can use the following command:

cargo run --release --bin experiment < path/to/model.aeon

As the model path, you can use one of the .aeon models provided in the benchmarks directory (note that some might require non-trivial time and memory).

Note that running the experiments from command line also requires Rust nightly toolchain.

If you have cargo make installed, you can also run the same command more easily with:

cargo make experiment < path/to/model.aeon

Here are expected run times for some of the benchmark models (measured on a 32-core workstation with 64GB of memory):

Model File Time (1CPU) Time (32CPU)
Asymmetric Cell Division (A)_parametric.aeon 0:05.62 0:03.39
Budding yeast (Orlando)_parametric.aeon 0:35.22 0:02.93
TCR signalisation_parametric.aeon 0:26.61 0:04.42
Drosophila cell cycle_parametric.aeon 27:48.1 1:42.28
Fission Yeast Cell Cycle_parametric.aeon 25:20.9 4:00.29
Mammalian Cell Cycle_parametric.aeon 38:39.6 8:02.14
Budding yeast (Irons)_parametric.aeon timeout 52:28.1