1805045 - Codepen demo (https://codepen.io/Mertl/full/GRJOgjL) is slower than Chrome

Status: NEW

(Core :: JavaScript Engine, task, P2)

Description

[Mayank Bansal]

Go to https://codepen.io/Mertl/full/GRJOgjL

https://share.firefox.dev/3VO71EN

Chrome is visually much more smooth.

Comment 1

[Jan de Mooij [:jandem]]

This one doesn't load for me?

Based on the profile they're using scripted getters on the global I think. I can try to write a patch to optimize that similar to native getters, but it would be nice to verify.

Comment 2

[Mayank Bansal]

(In reply to Jan de Mooij [:jandem] from comment #1)

This one doesn't load for me?

To load the page, do a full/force reload (Ctrl+shift+r). See bug 1800575

Comment 3

[Jan de Mooij [:jandem]]

(In reply to Mayank Bansal from comment #2)

To load the page, do a full/force reload (Ctrl+shift+r). See bug 1800575

Thanks! That fixed it.

Comment 4

[Jan de Mooij [:jandem]]

Mayank, did bug 1805199 improve this bug and bug 1805612 for you?

Comment 5

[Mayank Bansal]

Profile with latest Nightly: https://share.firefox.dev/3Yu6p9c

Comment 6

[Mayank Bansal]

Chrome is visually smoother

Comment 7

[Jan de Mooij [:jandem]]

Does setting javascript.options.use_fdlibm_for_sin_cos_tan = true make any difference for you? We seem to spend a fair amount of time in cos.

This looks pretty smooth for me on Mac and based on the profile we fixed the lowest-hanging fruit. It will be easier to analyze this more when we have JIT code information in the profiler, hopefully soon.

Comment 8

[Mayank Bansal]

(In reply to Jan de Mooij [:jandem] from comment #7)

Does setting javascript.options.use_fdlibm_for_sin_cos_tan = true make any difference for you? We seem to spend a fair amount of time in cos.

Profile: https://share.firefox.dev/3FBTIAu

Comment 9

[Mayank Bansal]

https://share.firefox.dev/4cA1E3O