Improve accuracy of set framerate in GLMakie

[ffreyer]

Description

When targeting a framerate GLMakie uses sleep() which is pretty bad at its job:

And thus the frametimes are pretty off. For the default 30fps GLMakie targets we should have 1/30 = 0.033s delay between frames but I get around 0.045:

This pr adds some bookkeeping for the used time per frame. If one frame sleeps for too long the next frame will know about it and sleep for less time. This averages out the frametimes to match the requested framerate.

An alternative option would be have a while (current_time < frametime) yield() which would be better for stability but I assume would be worse for power usage.

Type of change

• [x] Bug fix (non-breaking change which fixes an issue)

Checklist

• [x] Added an entry in CHANGELOG.md (for new features and breaking changes)
• [ ] Added or changed relevant sections in the documentation
• [x] Added unit tests for new algorithms, conversion methods, etc.
• [ ] Added reference image tests for new plotting functions, recipes, visual options, etc.

[MakieBot]

Compile Times benchmark

Note, that these numbers may fluctuate on the CI servers, so take them with a grain of salt. All benchmark results are based on the mean time and negative percent mean faster than the base branch. Note, that GLMakie + WGLMakie run on an emulated GPU, so the runtime benchmark is much slower. Results are from running:

using_time = @ctime using Backend
# Compile time
create_time = @ctime fig = scatter(1:4; color=1:4, colormap=:turbo, markersize=20, visible=true)
display_time = @ctime Makie.colorbuffer(display(fig))
# Runtime
create_time = @benchmark fig = scatter(1:4; color=1:4, colormap=:turbo, markersize=20, visible=true)
display_time = @benchmark Makie.colorbuffer(fig)

	using	create	display	create	display
GLMakie	4.55s (4.40, 4.66) 0.10+-	113.21ms (106.17, 119.99) 5.11+-	566.43ms (541.16, 584.10) 16.45+-	9.17ms (8.79, 9.64) 0.32+-	26.69ms (26.26, 27.18) 0.29+-
master	4.62s (4.54, 4.72) 0.06+-	120.24ms (111.70, 124.89) 4.56+-	583.41ms (552.34, 611.86) 19.48+-	9.57ms (9.01, 10.03) 0.36+-	26.62ms (26.10, 26.90) 0.34+-
evaluation	1.02x invariant, -0.07s (-0.82d, 0.16p, 0.08std)	1.06x faster✅, -7.03ms (-1.45d, 0.02p, 4.83std)	1.03x invariant, -16.98ms (-0.94d, 0.10p, 17.96std)	1.04x faster ✓, -0.41ms (-1.20d, 0.05p, 0.34std)	1.00x invariant, 0.08ms (0.24d, 0.66p, 0.31std)
CairoMakie	3.87s (3.68, 4.00) 0.13+-	113.36ms (109.24, 126.20) 5.83+-	136.96ms (131.80, 149.31) 6.03+-	8.57ms (8.13, 8.86) 0.26+-	980.86μs (970.65, 1005.59) 11.93+-
master	3.77s (3.67, 3.85) 0.08+-	110.70ms (106.02, 115.09) 3.54+-	136.00ms (129.30, 141.88) 5.06+-	8.43ms (8.18, 8.79) 0.19+-	986.62μs (974.12, 996.22) 8.21+-
evaluation	0.97x invariant, 0.1s (0.96d, 0.10p, 0.10std)	0.98x invariant, 2.66ms (0.55d, 0.33p, 4.68std)	0.99x invariant, 0.96ms (0.17d, 0.75p, 5.55std)	0.98x invariant, 0.14ms (0.61d, 0.28p, 0.22std)	1.01x invariant, -5.75μs (-0.56d, 0.32p, 10.07std)
WGLMakie	4.73s (4.68, 4.82) 0.05+-	112.57ms (109.06, 118.52) 3.45+-	9.60s (9.43, 9.76) 0.15+-	10.08ms (9.63, 10.27) 0.22+-	71.74ms (70.81, 72.61) 0.60+-
master	4.73s (4.68, 4.78) 0.03+-	112.47ms (110.55, 114.57) 1.79+-	9.52s (9.41, 9.59) 0.07+-	10.29ms (9.78, 11.42) 0.56+-	71.49ms (70.95, 71.88) 0.39+-
evaluation	1.00x invariant, -0.01s (-0.12d, 0.82p, 0.04std)	1.00x invariant, 0.11ms (0.04d, 0.94p, 2.62std)	0.99x invariant, 0.08s (0.67d, 0.25p, 0.11std)	1.02x invariant, -0.21ms (-0.49d, 0.39p, 0.39std)	1.00x invariant, 0.25ms (0.49d, 0.38p, 0.50std)

[jkrumbiegel]

I saw something on discourse where they were using a different sleep function for higher accuracy, is that an option?

[ffreyer]

I came across https://discourse.julialang.org/t/accuracy-of-sleep/5546 / https://github.com/JuliaLang/julia/issues/12770 but on windows Base.Libc.systemsleep() behaves the same as sleep() for me.