FLC Import Performance #31
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
- Minor things in ReadFlic (ToString implementation), Game.cs (move the start-up timer to the wrong place so I get metrics for just the FLC), and project.godot (the console can now handle reams of print statements, useful when I was printing all the images in all the FLCs to get an idea of the overall size of the task) - Micro-optimizations in LegacyUnit.cs that don't make much of a difference. Basically, replacing expensive CPU instructions (multiply, divide, modulus) with cheap ones (addition, shift logical left).
…els takes half the time. Obviously we don't want to merge this, but it feels good to follow the one commit, one change ideal when it so clearly demonstrates a difference.
…ching from SetPixel to LoadBmpFromBuffer. SetPixel was taking up half the time before, about 350 ms out of just over 700. This new method reduces that time by 80%. One major call-out about this code is that it uses an unsafe helper function to process the bits, including pointers and pointer arithmetic. When thinking "how do we create a byte buffer where there are various length primitives included in it?" this seemed like the way to go, as it allows us to treat shorts and ints as shorts and ints by pointer casting. We might be able to work around it without an unsafe block by tossing the ints/shorts in the header in as multiple bytes, based on their constituent bytes, and using bytes from the palette as-is. But that introduces a different level of complex reasoning. Regardless of what we decide about the specifics of this function, it shows that there are real and significant gains to be had by using byte buffers instead of SetPixel, and it was fun to dive back into the BMP format and learn how to write some low-level C-style C#.
We've added enough stuff to slow this down again over the past 5 days, so it's now slower than before (2.4 seconds). But we're drawing more stuff, and importing a SAV, and haven't tried to optimize any of that. Not going to try to make everything lickity-split in this branch.
…r amount if enabled, though are useful for figuring out the scope of what we import.
QuintillusCFC
commented
Nov 29, 2021
|
|
||
| [network] | ||
|
|
||
| limits/debugger_stdout/max_chars_per_second=16384 |
Member
Author
There was a problem hiding this comment.
This lets the console print more stuff. When I was printing out all the pictures we were importing, the console couldn't handle it until I added this. Might be too much on really slow systems, but it seems like a generally good thing and didn't cause any problems on my 9.9-year-old system.
QuintillusCFC
commented
Nov 29, 2021
Member
Author
QuintillusCFC
left a comment
QuintillusCFC
left a comment
There was a problem hiding this comment.
Looks like what I expected. Removed a few debugging sysouts.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Boosts FLC import performance by 40% (what used to take 1 second now takes 0.6 seconds).