About the jerkiness on the movement...

Canary is now up to v41. It seems like they have some other fix in place, as it was it terrible shape post-revert, but now seems to be functioning pretty well. The only kink is that if you bring a game up in the browser while it's windowed, the vsync fails...until you resize the window. After resizing it, canary will sync very well, whether windowed or maximized.

Haven't tried 39 stable yet...I'm updating to it now.

Honestly, at this point I'm not that worried about chrome, though I'd hate to have to use a version of node-webkit based on v38. The real problem is Firefox. It's just abysmal, even on simple games. It jerks, halts, drops frames, and screen tears (sometimes horizontally, which is a new one on me). Project Silk appears to address this...but what's going on with it? I may be wrong, but it looks like Silk has been getting less and less attention over the last couple months.

Firefox never quite matched chrome's performance, but I don't know how it got this bad, and I don't know why this isn't higher on their todo list...you can see all these things on webpages as much as with gaming. I'm going to dl the developer and nightly versions tonight and see if there are any improvements in the pipeline for this (though looking at the feature lists, my hopes aren't very high).

FYI, Chrome Stable v39 is a bit better than 38, but it still jerks and halts more than IE or the current Canary(smooth as butter once it locks in).

I see major improvements both in latest Chrome (Version 39.0.2171.65 m) and Canary (Version 41.0.2225.0 canary (64-bit)).

The problem still rise when running a test with letterbox scale. Making the browser full screen (F11) fixes that.

The same goes with node web-kit. For some reason without letterbox scale in my system I don't experience anything like the terrible things I read about in the forum the past 2 weeks. Letterbox scale however, does affect the performance, introducing heavy jerking when running at full screen.

When I run my data gathering test on Canary 41.0.2225..0 the spread of dt values is quite depressingly similar to chrome 38. However, Chrome 39.0.2171.65 m is very good, with only a small rate of frame drops....

Colludium , by the way, your test still produces the same integer spike values (I'm cursed, no other logical explanation exist), but this time the scrolling background move without any jerkiness!

Weird..... Consistently inconsistent!

Colludium

What do you see with Canary when you start in windowed mode? What about when you resize the window; does this clear up any janking? I just tested a couple games to judge the subjective performance, but I was getting great results fullscreen and maximized...the only issue was when I started the browser windowed. In those cases, a simple resize would fix it.

eli0s

Sorry you were never able to figure that out, but I'm glad your subjective results are better.

TiAm, I'll post images from the test tomorrow but I saw Canary was worse performing than Chrome 39... Window residing made no difference to the results in either case.

Lunatrap thanks for the nw installer! wow, nice to see my game be smooth again!!

TiAm , oh well, it's nice to be different I guess :O

But yes, since the jerking is much-much-much less than before, I won't complain

I've get the issue as well in the isometric rpg I'm working on. So far I've only addressed the problems minimally with the idea I'll get back to it once I have the game play programmed with their respective assets. Optimization is usually something that comes last and only to the extent it's needed. In the back of my mind this may mean reducing the scrolling by introducing smaller levels, but just more of them. I may also remove the Scroll-To behavior and implement some scrolling logic that only executes only if the character reaches the edge of the screen.

I agree it's annoying, but I've been programming long enough to expect these kinds of issues with development tools, and part of the art of programming is finding creative ways to work around them.

So is there any consensus on which is performing better, chrome 39 or canary 41? I have only been using canary but i will try chrome again today.

Canary seems better to me, except that if you start it windowed it jerks terribly til you resize the window...which is a really strange glitch I must say. I'm sure they'll stomp it out at some point. 39 is better than 38, almost as good as canary, but it still jerks a bit for me, which canary doesn't once it locks in.

I just tested with Chrome, Version 41.0.2237.0 canary (64-bit), and the issue appears to be gone, at least for me. I'm not sure how long it will take for Node Webkit to catch up, but for now I'll just keep using the canary build for testing.

Hey, this thread is back!

plauk

Hmm...I'm still seeing the same bug as before; if I start canary windowed, vsync fails, terrible jank. Resize/Maximize/Fullscreen corrects vsync, starting Maximized/Fullscreen locks vsync correctly from the start. So, 90% there...but the same as it was a week ago.