Performance testing and sign-off

[jbphet]

As part of the code review process for the release of the phet-io version of this sim, performance should be reviewed and approved by the designer(s).

[jbphet]

@KatieWoe and I just discussed how best to approach doing a performance test for this sim, and here's what we are currently thinking:

• @KatieWoe feels like collecting data for this is a reasonable task for QA
• I would like to collect a combination of both quantitative and subjective information on this, so I'll identify what I believe to be the worst case computational and rendering situation for each of the three screens and QA will profile them and also provide a subjective assessment
• I'll build a dev version and link it in this issue
• I'll add a table for the data, and QA can edit it as they collect the information
• I'll fill out the first row of the table with the information from my machine

[jbphet]

Worst cases for each of the three screens:

• "Waves" screen - Max greenhouse gas level, all the checkboxes checked, allow it to run for at least 30 seconds. Screenshot:
  
• "Photons" screen - Max greenhouse gas level, all the checkboxes checked, let it run at least 30 seconds. Screenshot:
  
• "Layer Mode" screen - All checkboxes checked, max solar intensity, min surface albedo, max layers, max IR absorbance, let it run at least 30 seconds:
  

[jbphet]

The dev version has been built and can be found at https://phet-dev.colorado.edu/html/greenhouse-effect/1.2.0-dev.1/phet/greenhouse-effect_all_phet.html.

[jbphet]

The table for collecting the performance data is below (in the next comment). Here are some notes:

• Use the query parameter profiler to get the performance info to show up on the screen, and use the FPS value on the far left. It will look something like this:
  
• WC stands for "Worst Case", meaning the cases described above
• In the headings, I shortened "Layer Model" to "Layers" for the 3rd screen
• For the "Usable" heading, use only "yes", "no", or "marginal". This is the tester's subjective assessment.
• The first row is from my development machine.

[jbphet]

Device	Browser	Waves WC FPS	Waves WC Usable?	Photons WC FPS	Photons WC Usable?	Layers WC FPS	Layers WC Usable?
@jbphet's Dell Precision	Chrome	60	yes	60	yes	60	yes
Device X	?	?	?	?	?	?	?

[jbphet]

Assigning to @KatieWoe and @Nancy-Salpepi for the next steps.

[Nancy-Salpepi]

Device	Browser	Waves WC FPS	Waves WC Usable?	Photons WC FPS	Photons WC Usable?	Layers WC FPS	Layers WC Usable?
@jbphet's Dell Precision	Chrome	60	yes	60	yes	60	yes
macbook air	safari	60	yes	60	yes	60	yes
2003 mac desktop	safari	60	yes	60	yes	60	yes
iPad (9th gen)	safari	30	yes	30	yes	30	yes
chromebook	chrome	*20	yes	*14	yes	*23	yes

• I waited 30 seconds and then took an average of the first 5 numbers for FPS.

[KatieWoe]

Device	Browser	Waves WC FPS	Waves WC Usable?	Photons WC FPS	Photons WC Usable?	Layers WC FPS	Layers WC Usable?
@jbphet's Dell Precision	Chrome	60	yes	60	yes	60	yes
Samsung Win 11	Firefox	60	yes	60	yes	50-60	yes
Chromebook	Chrome	40-50	yes	45-55	yes	30	yes
iPad Pro	Safari	60	yes	55	yes	35	yes

[KatieWoe]

Between @Nancy-Salpepi and I, I think think this is enough data. @jbphet back to you

[jbphet]

Thanks QA team!

Based on the data above, it looks like QA found the worst case situations for each of the three screens usable on all test devices. @arouinfar - as the designer, I think it is up to you to ultimately sign off on this, though you're welcome to bring in @kathy-phet if you like.

[arouinfar]

Thanks @Nancy-Salpepi @KatieWoe!

Performance is generally looking good, with lower-performing devices still being acceptable. I'm happy to close @jbphet.