Testing the speed of Scratch 3.0 on the Raspberry Pi 3 and the iPad Mini

19 November 2018

As you might know, a new version of Scratch launches in January. This new feature has a lightly redesigned interface, with the Stage on the right (where it was for Scratch 1.4) and a new ability to scroll through all the blocks to find the one you want. The most significant change, though, is that it's been rebuilt from the ground-up using HTML5. That means the same version of Scratch can now be used across a wider range of devices, compared to the previous Flash-based version of Scratch. Raspberry Pi users will be first class citizens in the Scratch community for the first time, able to use the same online version as everyone else.

Scratch 2.0 also works on the Raspberry Pi and now features in the Raspbian download. It's noticeably slower than Scratch 1.4, though, so I was wondering how Scratch 3.0 would measure up.

I created a simple benchmark program that times how long it takes for the cat to walk back and forth across the screen in 1000 movements of 10 steps each. Using a loop of 1000 movements helps to amplify any performance lag so it's easier to measure. Here's the code:

A simple speed test for Scratch

This isn't a perfect benchmark. Some operations (in particular the graphic effects) work more slowly in some versions of Scratch, for example. Because most Scratch programs use sprite movements, this code should give a useful indicator of how they will perform, though. By creating a more complicated benchmark using more features, the risk is that it doesn't capture the typical performance and that some highly optimised (or very slow) commands create a misleading average result. This test is based on the beta as at November 2018, so the launch version of Scratch 3 may be faster or slower, and different software configurations on the devices may also change performance.

I rebuilt the test script in each version of Scratch before testing. My test hardware was:

• A Raspberry Pi 3 Model B+
• An iPad Mini 2 (ME277B/A, from 2013). For Scratch 1.4, I used the Pyonkee app with its default sprite. Pyonkee brings Scratch to the iPad but does not include the trademarked sprites.
• A Windows 8 desktop PC

Obviously, using different devices and more modern devices in the case of the PC and iPad, you may see different results. You can see the code above if you want to try it on your devices, and feel free to share your observations in the comments below.

Drumroll, please! Here are the results, with execution time shown in seconds:

Performance was not always exactly the same, but it was close enough to give a fair indication.

So, the key conclusions are:

• The Raspberry Pi Foundation has invested heavily in optimising Scratch 1.4 and it shows in these results. It was the fastest of all the combinations I tested. While Scratch 3.0 will be good for simple programs, budding coders may prefer to stick with Scratch 1.4 so that their scripts and their games run faster. The Scratch 3.0 interface also runs a bit slow on the Raspberry Pi.
• On my Windows PC, Scratch 3.0 is roughly the same speed as Scratch 2.0, and a bit slower than Scratch 1.4.
• On my iPad Mini, Scratch 3.0 is very slow. The interface is pretty responsive, so you can code simple scripts okay but will probably find games are unplayably slow.

It is exciting that Scratch 3 will enable everyone to be part of the vast and growing Scratch community, running the same version of Scratch and enjoying the same features for sharing and discussing projects. There's nothing to stop anyone building their first scripts and programs using Scratch 3 on these devices. As coders progress to more advanced projects and speed-sensitive games, they may prefer to seek out a device with faster performance, depending on the device they are starting with.

For Scratch inspiration, check out my books Coder Academy, Cool Scratch Projects in Easy Steps and Scratch Programming in Easy Steps.

UPDATE: Following a discussion on Twitter, I have new results to share:

• "Result (average of 3 runs, scratch 3 beta running inside Chrome) of the benchmark running on a Pi3-based pi-top (current Polaris) was an execution time of 69.91s, so a bit slower than a vanilla Pi3/Raspbian," says Nick Morrott.
• "Tried @vivaldibrowser with beta @scratch 3.0 on Pi3B+. A little slower than your reported figure. Had to spoof the User Agent as Scratch 3 refused to run. It ran perfectly so Vivaldi should be supported," says Albert Hickey.
• "My [Windows 8] computer gets 37.88 in scratch 1 and 33 in scratch 2," adds Ryan Walmsley.

Permanent link for this post | Blog Home | Website Home | Email feedback

Does the Mission Python game work on a Mac? (and other compatibility questions)

07 November 2018

Some readers have been in touch to ask about compatibility and pass on their advice for other readers. Here's a round-up of compatibility tips and tweaks.

One of the screens you have to dash through in Mission Python.

Does Escape from Mission Python work on the Mac?

Yes! Othelo Vieira has kindly provided installation instructions for the software on his blog. David Weissman has also been in touch to confirm it's working "flawlessly" on his iMac. Many thanks to Othelo and David.

Does Escape from Mission Python work on Ubuntu?

Yes! Matt, also known as RaspberryPiSpy, is using Ubuntu 18 and had previously installed Python 3.6 and Pygame. He found he had to install Pygame Zero using pip3 install pgzero. To run the game he had to use python3 -m pgzero escape.py. Ian Brown confirms this approach using pip3 to install Pygame Zero but says you may hit permission problems, and recommends sudo pip3 install pgzero in that case. If you can't use sudo because your system is locked down, you can try pip3 install pgzero --user. Many thanks to Matt and Ian for testing and sharing their discoveries.

Can you use Thonny with Mission Python?

Yes! Reader Ian Brown says that to do so, you need to put import pgzrun at the very top of the code and then add pgzrun.go() at the very end. He is using Thonny on Ubuntu. Thank you, Ian!

The book uses IDLE, and you might find it useful to stick with IDLE because the color coding will match the colors used in the book.

Does the Mission Python game Escape work with Python 2?

No, sorry. Pygame Zero is only available for Python 3. You can easily download and install Python 3. The instructions are in the book.

Does the Mission Python game Escape work with Python 3.7?

The book recommends Python 3.6 because at the time of writing, Pygame wasn't available for easy installation on Python 3.7. Charles G Smith Jr has been in touch to say he's got the game working on Python 3.7. He writes: "Good News. I was able to get your escape game up and running using version 3.7 of Python. Here is what I did to make your code work with the latest version. I added the import statement of import pgzrun at the top of the program, and I also added the statement pgzrun.go() at the bottom of your program." Many thanks to Charles.

If you're less experienced with Python and/or don't already have it installed, I still recommend using Python 3.6 as described in the book to avoid the need to add extra code into all your listings. You can download Python 3.6 here.

Does the Mission Python game Escape work on older Raspberry Pis?

It doesn't crash, but it runs too slow to enjoy. It does need a Raspberry Pi 2 or Raspberry Pi 3 really.

What screen resolution does the Escape game from Mission Python require?

The game window has a height of 800 pixels, and I have been running it on a screen with a resolution of 1152x864 or higher. However, I did a quick edit to adjust the window height to 700 pixels so it can fit on a screen with a resolution of 1024x768. You may still need to hide the taskbar. You can download the smaller version of Escape here. You will need to rename the extension from .txt to .py, or paste the code into your code editor.

The tallest room in the game is the third room you visit (West Corridor, room 37), so in this version I have taken the height down from 15 to 13 tiles, so it matches the next-largest room sizes. I have repositioned the toxic floor tiles within the new room to avoid them being too close to the bottom exit, which makes them unfairly hard to avoid. I have commented the lines that I changed in this program. Note that in the draw() function, the red box that is used to clear the game area must be resized to avoid overwriting the air and energy bars, and those bars must also be moved up in the AIR section of the program.

Thank you to Ian and Ben Brown for identifying this issue.

What about other compatibility questions?

If you have any concerns, here's a low-risk way to test it works. You can download the free PDF sampler, which includes the installation instructions for Windows and Raspberry Pi together with instructions to get the game working, and you can download the final game code too. You can then test that the game works on your computer before buying the book.

Thank you to everyone who has supported the book so far. I'm excited to see your reviews and customisations!

Permanent link for this post | Blog Home | Website Home | Email feedback

Dip into the blog archive

June 2005 | September 2005 | January 2006 | March 2006 | April 2006 | May 2006 | June 2006 | July 2006 | August 2006 | September 2006 | October 2006 | November 2006 | December 2006 | February 2007 | March 2007 | April 2007 | May 2007 | June 2007 | July 2007 | August 2007 | September 2007 | October 2007 | November 2007 | December 2007 | January 2008 | February 2008 | March 2008 | April 2008 | May 2008 | June 2008 | July 2008 | August 2008 | September 2008 | October 2008 | November 2008 | December 2008 | January 2009 | February 2009 | March 2009 | April 2009 | May 2009 | June 2009 | July 2009 | August 2009 | September 2009 | October 2009 | November 2009 | December 2009 | January 2010 | February 2010 | March 2010 | April 2010 | May 2010 | June 2010 | August 2010 | September 2010 | October 2010 | November 2010 | December 2010 | March 2011 | April 2011 | May 2011 | June 2011 | July 2011 | August 2011 | September 2011 | October 2011 | November 2011 | December 2011 | January 2012 | February 2012 | March 2012 | June 2012 | July 2012 | August 2012 | September 2012 | October 2012 | December 2012 | January 2013 | February 2013 | March 2013 | April 2013 | June 2013 | July 2013 | August 2013 | September 2013 | October 2013 | November 2013 | December 2013 | January 2014 | February 2014 | March 2014 | April 2014 | May 2014 | June 2014 | July 2014 | August 2014 | September 2014 | October 2014 | November 2014 | December 2014 | January 2015 | February 2015 | March 2015 | April 2015 | May 2015 | June 2015 | September 2015 | October 2015 | December 2015 | January 2016 | February 2016 | March 2016 | May 2016 | July 2016 | August 2016 | September 2016 | October 2016 | November 2016 | December 2016 | January 2017 | July 2017 | August 2017 | October 2017 | November 2017 | January 2018 | February 2018 | August 2018 | October 2018 | November 2018 | December 2018 | January 2019 | March 2019 | June 2019 | August 2019 | September 2019 | October 2019 | January 2020 | February 2020 | March 2020 | April 2020 | May 2020 | June 2020 | September 2020 | October 2020 | December 2020 | January 2021 | February 2021 | May 2021 | June 2021 | October 2021 | November 2021 | December 2021 | January 2022 | February 2022 | March 2022 | May 2022 | July 2022 | August 2022 | September 2022 | December 2022 | March 2023 | April 2023 | May 2023 | June 2023 | October 2023 | November 2023 | January 2024 | February 2024 | May 2024 | June 2024 | July 2024 | September 2024 | October 2024 | November 2024 | December 2024 | Top of this page |