Benchmarking : Smil vs scikit-image

Morphological Image Libraries

Computer platforms configuration

Test platforms

We choose two computers of different types :

  • nestor : a desktop computer with one CPU, four cores and 8 threads;

  • taurus : a typical server with two CPUs, 12 cores each and 48 threads.

This choice allows us to check how both libraries make use of a high number of virtual CPUs (OpenMP), even if many morphological algorithms can't be easily parallelized.

taurus (server)

  • Computer Dell PowerEdge R540

  • CPU : 2 X Intel(R) Xeon(R) Gold 5118 CPU @ 2.30GHz

  • 12 cores - 2 threads - Total 48 VCPUs

  • Memory : 384 GiB

  • Operating system - Linux Ubuntu 18.04

  • Smil Python version : 1.0.0-dev

  • conda 4.10.3

  • scikit-image version : 0.18.1

  • Numpy 1.20.3

  • Python 3.8.11

nestor (desktop computer)

  • Desktop computer Dell OptiPlex 7010

  • CPU : 1 x Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz

  • 4 cores - 2 threads - Total 8 VCPUs

  • Memory : 16 GiB

  • Operating system - Linux Ubuntu 20.04

  • Smil Python version : 1.0.0-dev

  • scikit-image version : 0.18.2

  • Numpy 1.21.0

  • Python 3.8.10