A look at Dirty Bomb's optimization.


(JamsByBetty) #1

And it doesn’t look good. Lots of inconsistency in drawing frames, resulting in lots of stuttering. This isn’t my data - an acquaintance of mine put this together. But I’ve got a reasonably powerful computer as well and I can confirm it’s not just on his end.

This is something that affects the experiences of all users (spare me the “if it affects everyone it’s fair!” crap) and needs to be worked on.

Here’s the link to the infographic - image is too large to be posted.


(Mustang) #2

Nice graph, sadly this info is already known.

By the way if you plot as a percentile graph it’s easier to read and draw comparisons.


(Mustang) #3

Thought I’d have a go at a similar thing, and this is what I got.


Some noteworthy numbers for frame times:
0.04% of frames outside the 30fps line on Chapel
0.24% of frames outside the 60fps line on Chapel
0.03% of frames outside the 30fps line on Underground
0.14% of frames outside the 60fps line on Underground

Some noteworthy numbers for frame variance:
0.34% of frame variance outside the 144fps line on Chapel
0.28% of frame variance outside the 144fps line on Underground

In-case you’re unfamiliar with frame variance it is essentially just the time difference between sequential frames, which can be used as a measure of smoothness, or microstutter.

To be honest I found my results much better than expected, although I haven’t had any notable perceivable stutter recently either.


(FireWorks) #4

What tools are you using to gather the data?


(Mustang) #5

Fraps

#10chars