Benchmarks

To get maximal performance, it is recommended to pass -O3, --check-bounds=no, and --math-mode=fast as command line flags to julia.

Run perf/runbenchmarks.jl to see benchmark results for the Atlas robot (v5) in the following scenarios:

Compute the joint-space mass matrix.
Compute both the mass matrix and a geometric Jacobian from the left hand to the right foot.
Do inverse dynamics.
Do forward dynamics.

Note that results on Travis builds are not at all representative because of code coverage. Results on a reasonably fast machine at commit fa1b725:

Output of versioninfo():

Julia Version 0.7.0-beta.133
Commit 60174a9 (2018-07-03 20:03 UTC)
Platform Info:
  OS: Linux (x86_64-linux-gnu)
  CPU: Intel(R) Core(TM) i7-6950X CPU @ 3.00GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-6.0.0 (ORCJIT, broadwell)

Mass matrix:

  memory estimate:  0 bytes
  allocs estimate:  0
  --------------
  minimum time:     6.618 μs (0.00% GC)
  median time:      6.883 μs (0.00% GC)
  mean time:        7.158 μs (0.00% GC)
  maximum time:     40.992 μs (0.00% GC)

Mass matrix and Jacobian from left hand to right foot:

  memory estimate:  0 bytes
  allocs estimate:  0
  --------------
  minimum time:     7.442 μs (0.00% GC)
  median time:      7.839 μs (0.00% GC)
  mean time:        7.840 μs (0.00% GC)
  maximum time:     43.941 μs (0.00% GC)

Note the low additional cost of computing a Jacobian when the mass matrix is already computed. This is because RigidBodyDynamics.jl caches intermediate computation results.

Inverse dynamics:

  memory estimate:  0 bytes
  allocs estimate:  0
  --------------
  minimum time:     7.954 μs (0.00% GC)
  median time:      8.246 μs (0.00% GC)
  mean time:        8.456 μs (0.00% GC)
  maximum time:     34.537 μs (0.00% GC)

Forward dynamics:

  memory estimate:  0 bytes
  allocs estimate:  0
  --------------
  minimum time:     29.740 μs (0.00% GC)
  median time:      36.015 μs (0.00% GC)
  mean time:        36.014 μs (0.00% GC)
  maximum time:     186.809 μs (0.00% GC)