Game running half the speed in SU 2013 vs SU 8

Chris Fullmer

I don't think that DC's have any change in functionality or features from SU 8 to 2013.

I can't think of any reason why SU Make might act differently than SU Pro in terms of cpu processing. I'll ask around though, maybe someone has an idea. When I get a chance to test the code, I'll test it on both Pro and Make. Thanks everyone,

Chris

renderiza

@anton_s said:

I didn't meant it seriously, I just meant it the ironic funny way... Like there were topics like "Does Trimble Listen", or Or what does Trimble actually listen for? But, I do love Trimble, I love the way they upgraded SU2013 to newer window Framework. It was actually what I wanted

Understood! I do like ironic & funny comments a lot its just that sometimes I am little slow with them.

@chris fullmer said:

@Anton - yup, we're here listening! Trying to respond as best I can. thanks for pinning down the potential bottlenecks. I'll try to run some tests on them today and see if I can verify that I see the same behavior as you.

Chris Fullmer it means a lot that you are not only listening but also joining the conversation. I am very happy we got you as a Trimbler!

@driven said:

yep...
what else is taken away then???
john

Not very happy to find out that some existing features for SU8 free version are being disabled for Make.

Idea should be to limit Make with new never before seen features so that people will be all over Pro version. That way it will not seam like a cheap marketing trick to bring people over to Pro version.

There will always be bad apples working for Trimble but I have hopes since we got people like Chris Fullmer trying to do the right thing.

Cheers!

dacastror

How is configured by default "Sketchup 2013"?
I've noticed that changing this setting has strong effects on plugins that provide some kind of animation.
In Sketchup 8 an option that works best for me, is as follows

I do not know how it is in 2013

(google translate)

Chris Fullmer

Ok here is the snippet I ran from a multi-line webconsole:

model = Sketchup.active_model
sel = model.selection
gc = sel[0]
t = Geom;;Transformation.new([1.0,0.0,0.0,0.0,0.0,1.0,0.0,0.0,0.0,0.0,1.0,0.0,0.0,1.0,0.0,1.0])
puts ""
stime = Time.now
model.start_operation("no fast", false)
100000.times do
gc.transform! t#.inverse
end
model.commit_operation
puts "Transform Slow = " + (Time.now-stime).to_s

stime = Time.now
model.start_operation("fast", true)
100000.times do
gc.transform! t#.inverse
end
model.commit_operation
puts "Transform Fast = " + (Time.now-stime).to_s


stime = Time.now
model.start_operation("slow inverse", false)
1000000.times do
t.inverse
end
model.commit_operation
puts "Invert Transformation = " + (Time.now-stime).to_s


stime = Time.now
model.start_operation("slow", false)
100000.times do
gc.transform! t.inverse
end
model.commit_operation
puts "Combined Slow = " + (Time.now-stime).to_s


stime = Time.now
model.start_operation("fast", true)
100000.times do
gc.transform! t#.inverse
end
model.commit_operation
puts "Combined Fast = " + (Time.now-stime).to_s

Here are the results from 3 rounds of tests:
SU 8
` Transform Slow = 7.457
Transform Fast = 1.429
Invert Transformation = 0.55
Combined Slow = 7.442
Combined Fast = 1.437

Transform Slow = 7.391
Transform Fast = 1.446
Invert Transformation = 0.555
Combined Slow = 7.458
Combined Fast = 1.437

Transform Slow = 7.43
Transform Fast = 1.431
Invert Transformation = 0.553
Combined Slow = 7.479
Combined Fast = 1.456`

SU 2013
` Transform Slow = 0.789
Transform Fast = 0.42
Invert Transformation = 0.529
Combined Slow = 0.841
Combined Fast = 0.422

Transform Slow = 0.779
Transform Fast = 0.419
Invert Transformation = 0.53
Combined Slow = 0.841
Combined Fast = 0.419

Transform Slow = 0.806
Transform Fast = 0.421
Invert Transformation = 0.532
Combined Slow = 0.834
Combined Fast = 0.42`

So without the go_faster flag, I'm seeing SU 2013 doing the transformations about 10 times faster than it was in SU 8. Its even going about twice as fast as SU 8 with the go_faster flag.

Su 2013 appears to be about three times faster than 8 was when using the go_faster flag.

The transformation.inverse does not seem to have changed from SU 8 to 2013. I'm realizing I deleted the test from these results, but I had tested it with and without the go_faster flag and saw no difference. I could update it to do that again if it would help.

Does anyone see any holes in these tests? I could easily have overlooked something.

Chris

Anton_S

Your amazing Chris
Adding model.start_operation('KeyScene', true) when the user clicks the start button and calling model.commit_operation when user clicks the reset button seems to increase performance dramatically!

I take all my thoughts back, and do say that Trimble SU2013 is an extreme progression!!!

Thank you SketchUp Team!
Thank you Trimble!
Thank you Chris!

renderiza

@dacastror said:

How is configured by default "Sketchup 2013"?

Here is how it looks on SketchUp 2013 Make

But! If I change that default setting that option is gone and the only way I can get back to it is if I reinstall again which is little weird.
not there.jpg

@chris fullmer said:

Here are the results from 3 rounds of tests:

Thanks for does tests its really good to know 2013 is faster giving your results. I am still confused as to way in my game the SketchUp 2013 version runs very slow compared to SketchUp 8.

@anton_s said:

Adding model.start_operation('KeyScene', true) when the user clicks the start button and calling model.commit_operation when user clicks the reset button seems to increase performance dramatically!

Sounds like you made KeyScene work much faster...but was it faster than version 8?

Also I am not sure how can I add model.start_operation in Resballiza game to see if it will make it faster. Can you help me in that regards?

I am so thankful to all of you great job so far!

Anton_S

@renderiza said:

Sounds like you made KeyScene work much faster...but was it faster than version 8?

Yes, it's a little faster on SU2013 than on SU8

@renderiza said:

Also I am not sure how can I add model.start_operation in Resballiza game to see if it will make it faster. Can you help me in that regards?

I sent you a PM with a fixed copy of both KeyScene and Resballiza

driven

@Chris,

can you test on mac as well?

my results don't look so great.

EDIT: reformatted to make it clearer...
run on both versions with Sketchup.plugins_disabled? => true then false
simple cube, grouped:

%(#0000FF)[v2013 mac plugins turned off
Transform Slow = 4.051424
Transform Fast = 0.689763
Invert Transformation = 0.682962
Combined Slow = 4.149032
Combined Fast = 0.685753
nil

Sketchup.plugins_disabled?
true

v2013 mac plugins turned on
Transform Slow = 4.684443
Transform Fast = 1.141925
Invert Transformation = 0.87607
Combined Slow = 4.773361
Combined Fast = 1.152288
nil

Sketchup.plugins_disabled?
false]

%(#8000BF)[v8 mac
Transform Slow = 2.187031
Transform Fast = 0.795604
Invert Transformation = 0.809627
Combined Slow = 2.293091
Combined Fast = 0.784819
nil

Sketchup.plugins_disabled?
true

v8 mac
Transform Slow = 2.299228
Transform Fast = 0.805798
Invert Transformation = 0.985101
Combined Slow = 2.456473
Combined Fast = 0.818888
nil

Sketchup.plugins_disabled?
false]
john

renderiza

@anton_s said:

I sent you a PM with a fixed copy of both KeyScene and Resballiza

Magnificent!

Now Resballiza is way faster than before in SketchUp 2013!

SketchUp 8 runs little bit faster still but not by much...I will study ways of optimazing code to make it faster but as is right now it is playable.

Again thanks Anton_S and Chris Fullmer for all the help!

Chris Fullmer

Awesome, glad its improved! Thanks for working that into his code Anton, great!

@John - I have not yet tried it on Mac. I'll see if I get a chance here soon.

Anton_S

@renderiza said:

SketchUp 8 runs little bit faster still but not by much...I will study ways of optimizing code to make it faster but as is right now it is playable.

Sounds great, I will be the tester

@chris fullmer said:

Awesome, glad its improved! Thanks for working that into his code Anton, great!

It was interesting, I never actually knew that setting the second parameter of model.start_operation to true would actually improve the performance. That is an important parameter to know, especially when your dealing with geometry plugins.