About the jerkiness on the movement...

Just to add to the above - I downloaded Chrome Canary. It gave a much improved performance and could be considered better than IE....

[attachment=0:2bo52sge][/attachment:2bo52sge]

Colludium

Have you tried Canary yet? I ran your test and it confirmed my subjective opinion: Canary now vsync's about as well as IE.

EDIT: Whoops...you posted right before me...

[attachment=1:6z01cbk3][/attachment:6z01cbk3]

Others:

[attachment=2:6z01cbk3][/attachment:6z01cbk3]

[attachment=0:6z01cbk3][/attachment:6z01cbk3]

And the rest...

TiAm, I did - I posted the results probably just before you did yours. I came to the same conclusion - problem fixed...

Here's my firefox result - slightly worse than yours!

This topic has become way too techy for my understanding!

I'm getting weird results from Colludium 's test in Firefox and both Chrome 38 & Canary. Only IE 11 renders a curve that resembles the tests that you guys posted above...

Chrome 38

Chrome Canary (40)

Firefox 33

IE 11

Am I the only one that gets this? What's going on? I fear now that my PC might be the culprit for the jitter and not the browsers... For example, many members posted that the latest Canary has no jittering problems, I still get almost the same results as with normal Chrome: Every few seconds the sprites that move pause for an instant and then jump a few pixels ahead...

eli0s, for some reason your images haven't posted (for me, at least)... If you're still having trouble with Canary, it might be worth shutting down all non-essential pc processes (anything that isn't your browser) to see if there's something else causing the momentary pauses. I found as well that having only one tab open made a big difference in normal Chrome, as well.

Does this mean anything for CocoonJS (which uses V8)? How about Crosswalk?

Colludium , I am sorry, I just don't know how to use dropbox Please look again in my post above!

eli0s, those Firefox + Chrome results of yours look a bit odd for sure. It's as if the dt that was being reported in those tests was given as whole ms values (14 ms, 15 ms etc) with no scatter of fractional dt values near to those whole millisecond numbers. I would expect more of a scatter of the results like in your IE graph.

Could you post a screen shot of my first dt test (from a couple of pages back in this thread) as that test shows a side-scrolling waterfall display of each dt value? From the graphs above the other test should show pure straight lines of constant value dt readings (which I doubt).... If it does show straight lines then it supports the second test results – and that there's something unusual going on with your browsers' dt values. If it doesn't corroborate then there's something odd with the events in my 2nd demo (!)....

Testing on Chrome canary 40, the jitter and stuttering is almost completely gone from movement for me. Is there a way to use this build of Chrome right now in Crosswalk or do we have to wait for the stable release and Intel add it to Crosswalk?

Colludium ,

I am afraid that the first test also gives similar results for some reason. Here are some screen shots:

Firefox

Chrome

Canary

IE11

Firefox is just a mess with no red dots at all, Chrome and Canary produce almost straight lines with occasional dots all around and only IE renders a graph that resembles the expected result. Of course, as always, in all tests the jitter is there like a common factor...

Woah...!! eli0s, that's beyond my comprehension. Some of your browsers are hitting whole-ms-number dt values every time... Even the Canary values are unexpected and nothing like the other images posted here... I have now reached the limit of my understanding of browsers!!

eli0s

EDIT: Did you run you tests with the browsers windowed? If so, try them Maximized/Full Screen.

Just don't know what to make of that. What is your exact version of Canary (Toolbar -- About Google Chrome)?

Colludium

Could you post a version of your second test that only runs for 20 seconds? Or even lets the user specify run length?

The weirdness continues. I just ran Colludium 's second test, but with Canary windowed this time, and got similar results to eli0s:

[attachment=1:2wrdo31u][/attachment:2wrdo31u]

The reason I tried it windowed? When I try to load the test with Canary maximized, it won't load (huh?). I have to make it windowed. However, after it loads I can maximize it and run the test, which gives me this:

[attachment=0:2wrdo31u][/attachment:2wrdo31u]

I guess Canary's new V-Sync improvments only work with the browser maximized for now...

Colludium , today I get this from firefox:

Chrome & Canary are still the same: Either integer values or 15.99, 17.99 etc... Running your capx with node-webkit also produces the same thing, as expected i guess...

TiAm , it doesn't matter if I run the tests in windowed mode or not. The same thing happen. My Chrome version is 38.0.2125.111 m and it tells me that it's up to date. Canary is 40.0.2208.0 canary (64-bit). I don't know why Chrome doesn't display the 64bit thingy, perhaps it isn't a 64bit version. Is there a normal Chrome 64bit version available...? Anyway, I don't think that has do do anything though.

I don't know what to get out of this, I've even disabled my secondary monitors as was suggested by Tylermon at page 7 and nothing changed on the chrome based browsers. On the other hand, with the secondary monitors disabled, IE11 runs the waterfall benchmark with no scatter what so ever!!!