timrowledge
Posts: 1151
Joined: Mon Oct 29, 2012 8:12 pm
Location: Vancouver Island
Contact: Website

Re: nuScratch is now default scratch

Mon Dec 29, 2014 11:31 pm

[quote="simplesi"@Tim Please, please, please can you get rid of the "do you want to save" when nothings happened and I'm try to open an existing file :)[/quote]

Thing is, I can't replicate this problem. I've fixed a prior issue that caused the symptom and on my machines it just works. Are you still doing weird display stuff? Maybe vnc etc causes extra damage events that make it think something has actually changed?
Making Smalltalk on ARM since 1986; making your Scratch better since 2012

timrowledge
Posts: 1151
Joined: Mon Oct 29, 2012 8:12 pm
Location: Vancouver Island
Contact: Website

Re: nuScratch is now default scratch

Mon Dec 29, 2014 11:54 pm

simplesi wrote:I'm getting very high CPU usage just by loading in a medium script without even running it

https://www.dropbox.com/s/h2zn2kir4wwuz ... ng.sb?dl=0
Interesting; profiling that shows nearly 40% of time going to repeatedly redrawing the tabs in the script editor. That certainly isn't a good thing.

On the plus side, if I can spot why & fix it , you should run a good 50% faster :-)
Making Smalltalk on ARM since 1986; making your Scratch better since 2012

timrowledge
Posts: 1151
Joined: Mon Oct 29, 2012 8:12 pm
Location: Vancouver Island
Contact: Website

Re: nuScratch is now default scratch

Tue Dec 30, 2014 12:57 am

ghp wrote:Performance results.
Fantastic work ghp. Thanks very much.

I have an ancient image with no speed up work, but no original VM. Using the old but still faster vm (it has the faster display subsystem) on this ancient image I get a whole 4 fps in Asteroid Blaster medium stage when firing, as opposed to my latest system where I get 19. In presentation mode I get a whole 4-5fps in the old system against mid-30s for the new one. I suspect with the original VM we'd see more like 3fps but I'm not sure we tested is much at the time.

If anyone still has a copy of the paeolithic package from circa january 2013 before I did any work, I'd be grateful for a copy if only as a relic.
Making Smalltalk on ARM since 1986; making your Scratch better since 2012

ghp
Posts: 1241
Joined: Wed Jun 12, 2013 12:41 pm
Location: Stuttgart Germany
Contact: Website

Re: nuScratch is now default scratch

Tue Dec 30, 2014 6:51 am

If anyone still has a copy of the paeolithic package from circa january 2013 before I did any work, I'd be grateful for a copy if only as a relic.
Hello, there is an archive of distros available http://downloads.raspberrypi.org, back to the first one 16-Jul-2012 14:58
Needs unpacking to a SD card or alike, but should give you all the vm needed.
Regards,
Gerhard

simplesi
Posts: 2327
Joined: Fri Feb 24, 2012 6:19 pm
Location: Euxton, Lancashire, UK
Contact: Website

Re: nuScratch is now default scratch

Tue Dec 30, 2014 11:03 am

timrowledge wrote:? Maybe vnc etc causes extra damage events that make it think something has actually changed?
I'll put together a non X11VNC server image and see if I can isolate what is happening then

Simon
Seeking help with Scratch and I/O stuff for Primary age children
http://cymplecy.wordpress.com/ @cymplecy on twitter

simplesi
Posts: 2327
Joined: Fri Feb 24, 2012 6:19 pm
Location: Euxton, Lancashire, UK
Contact: Website

Re: nuScratch is now default scratch

Tue Dec 30, 2014 11:14 am

Can I make a general comment here please :)

Fps increase seems to a good idea - I personally don't know what it means in the Scratch context - AFAIK - I can't measure it myself - but if its going up -- I can believe that is a good thing :)

CPU high when program doing nothing is a bad thing as it doesn't leave room for when its running but 95% running CPU load isn't necessarily a bad thing as it shows that prog is might be using using resources to a good effect - but I need some CPU time for my python handler :)

Any speed ups of any sort are good but although we use the orig Asteroid Blaster code as base test (and its a good one to show - like the original rotating coke can demos in Windows) - I'm firmly of the belief that its the editing speed boosts that young Scratchers are benefitting from as their attention span is much shorter than ours :)

JMHO

Simon
Seeking help with Scratch and I/O stuff for Primary age children
http://cymplecy.wordpress.com/ @cymplecy on twitter

simplesi
Posts: 2327
Joined: Fri Feb 24, 2012 6:19 pm
Location: Euxton, Lancashire, UK
Contact: Website

Re: nuScratch is now default scratch

Thu Jan 01, 2015 6:28 pm

After creating a new SD without X11vnc installed on it to test the File open asking to save bug, I've come up with more basic problem :(

When I create a simple test script and try and save it

1. There is no Scratch Projects folder within Documents folder

2. If I try and save in Documents - it gives Save failed: InvalidDirectoryError

Simon
Seeking help with Scratch and I/O stuff for Primary age children
http://cymplecy.wordpress.com/ @cymplecy on twitter

timrowledge
Posts: 1151
Joined: Mon Oct 29, 2012 8:12 pm
Location: Vancouver Island
Contact: Website

Re: nuScratch is now default scratch

Fri Jan 02, 2015 6:24 pm

simplesi wrote: 1. There is no Scratch Projects folder within Documents folder

2. If I try and save in Documents - it gives Save failed: InvalidDirectoryError
That's a side effect of the mistaken startup script. If you take a look at issue 122 (https://github.com/raspberrypi/scratch/issues/122) you'll notice at the end a reference to a modified version that you can try out.
Making Smalltalk on ARM since 1986; making your Scratch better since 2012

asb
Forum Moderator
Forum Moderator
Posts: 853
Joined: Fri Sep 16, 2011 7:16 pm
Contact: Website

Re: nuScratch is now default scratch

Fri Jan 30, 2015 12:22 am

I've now updated the scratch package with a new release from Tim and a fixed scratch launcher script.

To update or make sure you're running the latest version, just do the following:

Code: Select all

sudo apt-get update
sudo apt-get install nuscratch

simplesi
Posts: 2327
Joined: Fri Feb 24, 2012 6:19 pm
Location: Euxton, Lancashire, UK
Contact: Website

Re: nuScratch is now default scratch

Fri Jan 30, 2015 7:24 am

Excellent :)
[edit]passing scripts as parameters back working :) ScratchGPIO back to 100% :) Ta [/edit]
Seeking help with Scratch and I/O stuff for Primary age children
http://cymplecy.wordpress.com/ @cymplecy on twitter

timrowledge
Posts: 1151
Joined: Mon Oct 29, 2012 8:12 pm
Location: Vancouver Island
Contact: Website

Re: nuScratch is now default scratch

Tue Feb 03, 2015 5:54 am

Aaaand yes, it does run nicely on the Pi 2. Not hugely faster but definitely noticeable. The memory access seems to help a lot with opening the painter for example, and the extra cores mean better overall system performance even when scratch is busy. Running via things like rdp is much improved (though the permissions issue is untouched).
Making Smalltalk on ARM since 1986; making your Scratch better since 2012

ghp
Posts: 1241
Joined: Wed Jun 12, 2013 12:41 pm
Location: Stuttgart Germany
Contact: Website

Re: nuScratch is now default scratch

Wed Feb 04, 2015 7:02 pm

Hello,

got new hardware today, pi2 is here. What first ? No question: performancetest.
So I run on piB (the old one) the move-tests, rotate with a 'say bubble' and pingpong internal and remote.
Pingpong is simple: one script broadcasts 'ping', another does when 'ping' received then broadcast 'pong' and the third one when 'pong' received then broadcast 'ping' . First script starts the game, and then it runs as fast as the system allows. A counter stops this after some hundred iterations and takes the time.
Pingpong_remote is the same game, but a sensor network client responding to pings with pongs. So it measures remote socket throughput.

move is a diagonal move of two sprites, some watchers on screen, which was quite slow in legacy scratch.
rotate-say is a sprite which says 'hello', and gets rotated.Lots of computations needed to rotate and to find optimal position for the say-bubble.
All tests with raspbian, 1GHz cpu freq.

Results, the smaller the numbers the better

move2 16.5 --> 13.7
rotate_say 9.3 --> 6.7

pingpong 8.4 --> 11.2
pingpong_remote 8.4 -->12.1

Strange. On one side there are 17 to 28% improvement; on other tests it runs 33 to 44% slower.

Regards,
Gerhard

See other performance measurement results for nuscratch:
http://heppg.de/ikg/wordpress/?p=270

Edit: here the 'internal' pingpong sample.
performance_ping_pong.zip
(40.06 KiB) Downloaded 345 times

Return to “Scratch”