Rework WGLMakie lines #3062

SimonDanisch · 2023-07-13T11:11:57Z

More efficient implementation of lines purely in JS
implement updating
fix WGLMakie plots "interleave" traces when updated chaotically #3050
line patterns
3d lines
colormaps + clean up implementation

This should also be a template for how we want to generate the plot objects in JS in the future, which should lead to a large TTFP improvement.
If I find the time to do the same treatment for all plot types in WGLMakie, it may just get the fastest TTFP of all backends, since it will basically just create dictionaries from plots.

This should also bring us closer to #2977, with a more comprehensive plotting API directly in JS.

MakieBot · 2023-07-13T11:38:52Z

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(display(fig))

	using	create	display	create	display
GLMakie	12.27s (11.74, 13.89) 0.80+-	1.10s (1.06, 1.26) 0.07+-	1.24s (1.16, 1.43) 0.09+-	10.59ms (10.37, 11.29) 0.32+-	150.17ms (145.25, 171.64) 9.52+-
master	12.16s (11.75, 13.81) 0.73+-	1.11s (1.04, 1.29) 0.08+-	1.26s (1.18, 1.41) 0.10+-	10.84ms (10.36, 12.23) 0.70+-	151.98ms (144.89, 172.07) 10.11+-
evaluation	+0.92%, 0.11s invariant (0.15d, 0.79p, 0.77std)	-1.53%, -0.02s invariant (-0.21d, 0.70p, 0.08std)	-1.57%, -0.02s invariant (-0.20d, 0.72p, 0.10std)	-2.37%, -0.25ms invariant (-0.46d, 0.41p, 0.51std)	-1.20%, -1.81ms invariant (-0.18d, 0.74p, 9.82std)
CairoMakie	13.13s (12.98, 13.41) 0.14+-	1.37s (1.30, 1.45) 0.06+-	791.03ms (757.88, 864.86) 35.59+-	14.70ms (14.37, 15.68) 0.45+-	8.22ms (8.02, 8.36) 0.11+-
master	12.91s (12.65, 13.45) 0.26+-	1.33s (1.26, 1.41) 0.05+-	788.27ms (768.97, 817.15) 14.92+-	14.30ms (14.02, 14.41) 0.14+-	8.33ms (8.15, 8.53) 0.11+-
evaluation	+1.70%, 0.22s invariant (1.07d, 0.07p, 0.20std)	+2.74%, 0.04s invariant (0.66d, 0.24p, 0.06std)	+0.35%, 2.77ms invariant (0.10d, 0.85p, 25.26std)	+2.75%, 0.4ms invariant (1.21d, 0.06p, 0.30std)	-1.32%, -0.11ms invariant (-0.97d, 0.10p, 0.11std)
WGLMakie	15.21s (15.03, 15.34) 0.12+-	1.36s (1.31, 1.42) 0.04+-	14.37s (14.18, 14.50) 0.13+-	16.39ms (15.40, 19.64) 1.51+-	1.19s (1.15, 1.22) 0.02+-
master	15.18s (14.98, 15.38) 0.13+-	1.34s (1.29, 1.37) 0.03+-	14.37s (14.15, 14.60) 0.19+-	15.87ms (15.43, 16.56) 0.41+-	1.15s (1.09, 1.20) 0.04+-
evaluation	+0.17%, 0.03s invariant (0.22d, 0.69p, 0.12std)	+1.83%, 0.02s invariant (0.69d, 0.22p, 0.04std)	-0.04%, -0.01s invariant (-0.04d, 0.95p, 0.16std)	+3.18%, 0.52ms invariant (0.47d, 0.41p, 0.96std)	+3.54%, 0.04s slower X (1.24d, 0.04p, 0.03std)

staticfloat · 2023-08-03T15:18:37Z

@SimonDanisch I just tried out the current state of the branch (b9e2bad) and it seems to have completely solved #3050! I can no longer trigger that kind of issue.

github-actions · 2023-08-16T13:45:23Z