Download this project!
Project Notes
This project benchmarks three different scripts, each doing seemingly the exact same thing: Forever incrementing a variable by one.
It's interesting how the speeds differ because of parallelism and of how Scratch handles display updates within loops.
You can get some hints out of this how to boost up your scripts...
Credit goes to johnm for pointing this out to me, when we discussed performance issues relating my lists implementation.
Comments
You need to be logged in to post comments
Add a Comment
Go 1 should always be go 2 and go 3 divided by 2. I don't understand the point, what are you racing against? Because Go one is only increasing by 1 one whereas go 2 and 3 are increasing by 2.
(view all replies)Comment Reply
var2 + 3 are drawing!
(view all replies)Comment Reply
Hi all, I put an alternative version of this at: (link to project) . Mostly this is to address AdvancedScratcher's comment that the first script is being unfairly penalized. The first script should now clearly update its 'var1' twice within each outer loop... so it appears "more fair". But it actually makes the first script even slower: only one loop iteration is executed in any display cycle, so now it is running at 1/3 the rest.
(view all replies)Comment Reply
Nice,mytheory is that scratc go through code once per sprite up to down once,then go to the next sprite, your show that could b the case.
(view all replies)Comment Reply
Do you have a mac by any chance? or an iphone/ipod touch? if you do you should make some applications in xcode, it would be really awesome!
(view all replies)Comment Reply
No, fbboi999, the only thing I own from Apple is an iPod nano...
(view all replies)Comment Reply
I have a Mac! And an iPod!
(view all replies)Comment Reply
Hi Jens! Can you help us out with the type of scrolling on this project hears the (link to project) . Thanks and come join my brothers gallery by Mariojon called Mario kart gallery. Thanks! :D
(view all replies)Comment Reply
not a very fair test if you ask me. because one of them has only 1 script saying change var1 by 1, the other 2 have change var2 by 1x2 and var3 has 2 scripts saying change var3 by 1, I have also examined this and it is now known to me that var2 happens to go faster than var3, obviously var1 is the slowest because of the unfair test, but then its var3, and var2 is the fastest. good race test jens
(view all replies)Comment Reply
but this is proving that forever blocks take more time to run than another command block.
(view all replies)Comment Reply
Hi AdvancedScratcher! Can you help us out with the type of scrolling on this project hears the (link to project) . Thanks and come join my brothers gallery by Mariojon called Mario kart gallery. Thanks! :D
(view all replies)Comment Reply
Thats really cool!!!!!!!!!! You make great projects, and they help me understand scratch WAY better, causing me to make better projects!
(view all replies)Comment Reply
I downloaded it and changed the first script for var1 and made it change by 2 instead of one so all of them were equal
(view all replies)Comment Reply
Hi terminator68! Can you help us out with the type of scrolling on this project hears the (link to project) . Thanks and come join my brothers gallery by Mariojon called Mario kart gallery. Thanks! :D
(view all replies)Comment Reply
This I can use! Thanks!
(view all replies)Comment Reply
Hi Paddle2See! Can you help us out with the type of scrolling on this project hears the (link to project) . Thanks and come join my brothers gallery by Mariojon called Mario kart gallery. Thanks! :D
(view all replies)Comment Reply
I had also noticed a while ago that loop unrolling significantly sped up calculations on Scratch. I do a lot of "speed tests" on Scratch; I had a accurate sin function that would have been useful before Scratch 1.2.1, and it was only 1.5 times as slow (not much of a difference, the sin function is already pretty fast).
(view all replies)Comment Reply
Now that is interesting.
(view all replies)Comment Reply
Hi chalkmarrow! Can you help us out with the type of scrolling on this project hears the (link to project) . Thanks and come join my brothers gallery by Mariojon called Mario kart gallery. Thanks! :D
(view all replies)Comment Reply
Yes...oh I almost forgot! look at this:(link to project) I am testing the velocity of turning.
(view all replies)Comment Reply
It's also interesting how this projects measures almost the exact same speeds online and offline, even one very differently performing computers...
(view all replies)Comment Reply
ahhhhh.... I see what you mean Jens... it is just changing a vairable.
(view all replies)Comment Reply
...very interesting experiment! ....so "forever" acts also as a "wait" block?...since the first and second scripts are logically the same thing, but act differently
(view all replies)Comment Reply
Not really, the "secret" here is, that loops update the display just once on each iteration, and that updating the display is what "slows" Scratch down (of course, it's also what makes Scratch interesting in the first place)
(view all replies)Comment Reply
Ah, now I understand ^^
(view all replies)Comment Reply
Hi MyRedNeptune! Can you help us out with the type of scrolling on this project hears the (link to project) . Thanks and come join my brothers gallery by Mariojon called Mario kart gallery. Thanks! :D
(view all replies)Comment Reply
Cool i'm first to comment
(view all replies)Comment Reply
Wow, that's a lot of very complicated scripts!
(view all replies)Comment Reply
actually, redsooty9, the scripts are just about as simple as can be. All they do is forever change a variable by 1. The interesting thing is how they do that faster if you make more threads (parallelism) or put more command blocks into a loop (because loops update the display just once on each iteration). Knowing this you can speed up your projects a lot, especially those parts which are mainly doing calculations.
(view all replies)Comment Reply