Limit the number of writes to improve performance #22
Conversation
| @@ -135,7 +138,7 @@ def image(self, img): | |||
| self.buf[i] = 0 | |||
| # Iterate through the pixels | |||
| for x in range(width): # yes this double loop is slow, | |||
| for y in range(height): # but these displays are small! | |||
| for y in range(height): # but these displays are small! | |||
There was a problem hiding this comment.
Was this a manual? My guess this whitespace change is what's making black angry.
There was a problem hiding this comment.
Automatically formatted by VS Code. I'm not sure what black is in reference to. I can revert this change if you wish but it's probably good cleanup.
There was a problem hiding this comment.
Yeah, it's just that the CI will continue to complain. If you want to be extra sure, you can run pre-commit yourself, and let it decide what how to format everything. Pre-commit is what the CI uses, so that's the surest way.
There was a problem hiding this comment.
black is a formatter, and our CI tool pre-commit uses it to standardize the formatting of our code.
I was able to run some experiments. Before this change I got about 1 fps on my Raspberry Pi Zero W. With this change I get about 14 fps. It's a lot better but still isn't great given the actual power of the board. It'd be great if there was an easy to install library that unlocked the whole performance of the display. Something like: https://github.com/wrobell/smemlcd |
|
Always happy to look at PRs! And I think your changes fixed the original issue, but the CI is having it's own hiccups with the new move to Python 3.11. I'm going to leave this here for just a bit to see if we can quickly resolve that but I'm pretty sure there's nothing else on your end! |
Updating https://github.com/adafruit/Adafruit_CircuitPython_CharLCD to 3.4.4 from 3.4.3: > Merge pull request adafruit/Adafruit_CircuitPython_CharLCD#73 from BiffoBear/add_typing > Update .pylintrc for v2.15.5 > Fix release CI files > Update pylint to 2.15.5 > Updated pylint version to 2.13.0 > Switching to composite actions Updating https://github.com/adafruit/Adafruit_CircuitPython_LIS3DH to 5.1.19 from 5.1.18: > Merge pull request adafruit/Adafruit_CircuitPython_LIS3DH#73 from crotwell/extra_underscore > Update .pylintrc for v2.15.5 > Fix release CI files > Update pylint to 2.15.5 > Updated pylint version to 2.13.0 > Switching to composite actions Updating https://github.com/adafruit/Adafruit_CircuitPython_SharpMemoryDisplay to 1.4.6 from 1.4.5: > Merge pull request adafruit/Adafruit_CircuitPython_SharpMemoryDisplay#22 from grouma/main > Update .pylintrc for v2.15.5 > Fix release CI files > Update pylint to 2.15.5 > Updated pylint version to 2.13.0 > Switching to composite actions Updating https://github.com/adafruit/Adafruit_CircuitPython_MiniMQTT to 6.0.0 from 5.5.2: > Merge pull request adafruit/Adafruit_CircuitPython_MiniMQTT#125 from BiffoBear/Make_is_connected_return_bool > Update .pylintrc for v2.15.5 > Fix release CI files > Update pylint to 2.15.5 > Updated pylint version to 2.13.0 > Switching to composite actions
|
Oh Hey, I am glad someone made this change. I kept meaning to get back to this but life keeps getting in the way. :) if you are running on a Raspberry Pi and have not already tried it, add Numpy to your installation. I made another small change that speeds up the image function if Numpy is present. See here for details. |
The current
dispaly.show()command is very slow and takes roughly a second to complete. I discovered from this thread that it's likely due to the Linux scheduler limiting how fast one can send data to the SPI hardware. To overcome this issue we now do a single write command. This significantly improves the performance which makes for a nearly instant screen response.This resolves #18