Sluggish crosshair after dynamic updates

[carlos8f]

Great library! Works well, except for sluggish performance after a while. Tried hardware acceleration on and off. Any advice would be appreciated.

https://zenbot.s8f.org/

[andredumas]

That's an interesting one. I observed the issue on that site, but when I attempt to inspect it becomes performant, even if i'm careful not to cause a redraw. Would be helpful if you could provide sample code of the isolated issue and any specific observations.

Nice color scheme btw

[carlos8f]

yeah, it's odd. Will see if I can reproduce the issue with minimal code. Thanks.

[carlos8f]

Page to reproduce the issue: http://bl.ocks.org/carlos8f/raw/ac8677ff6f6432b87b4d7fed25931e7d/

Minimal 2-line fix (disable scale() and differencePath()): http://bl.ocks.org/carlos8f/raw/346f1030d1aaa1d5ba635cc67452214d/

CPU profile: https://gist.github.com/carlos8f/8a0d617739f35f7aa7a9af064c9fc76c#gistcomment-1852166

working on this at https://gist.github.com/carlos8f/8a0d617739f35f7aa7a9af064c9fc76c

[carlos8f]

There's a perf. bottleneck in the scale() function. https://gist.github.com/carlos8f/8a0d617739f35f7aa7a9af064c9fc76c#gistcomment-1852195

[carlos8f]

Sluggish issue is still present after patch, although my CPU profiler is no longer showing scale() as bottleneck

[andredumas]

Nice pick up! Will give the pr a try see if it fixes for me.

Thank you.

[andredumas]

Sluggish issue is still present after patch

Still a problem then?

[carlos8f]

still a problem after those date/string optimizations, just takes slightly longer to accumulate slowness.