Flat span performance improvement by completely ignoring visplanes #35
Conversation
This file contains 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
… dist so it's not a division anymore
FrenkelS
reviewed
Dec 5, 2023
| #else | ||
| // Find span color per-subsector instead of per-column. - mindbleach | ||
| R_LoadColorMap( frontsector->lightlevel ); | ||
| // R_GetColorMapColor copies a colormap row to a fixed location. GBA heritage? |
There was a problem hiding this comment.
Thanks for pointing out current_colormap has a fixed location. That's weird. I've removed current_colormap.
FrenkelS
reviewed
Dec 5, 2023
|
|
||
| uint16_t l = count >> 4; | ||
|
|
||
| while( l-- ) { |
There was a problem hiding this comment.
Hey, you're right. Unrolling R_DrawColumnFlat() ourselves is faster than what gcc-ia16 produces.
Thanks.
|
I've removed all visplanes. The sky is rendered as a regular texture. |
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Outdated suggestions cannot be applied.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.
By drawing solid color directly instead of searching for / adding to / rendering out an untextured visplane, overall performance with FLAT_SPAN is 10% faster. The memory used by visplanes (and visplane functions) is no longer required. This commit only reduces the number of visplanes allocated, though.
Additionally: unrolling R_DrawColumnFlat is 5% faster.
Additionally: a naive loop writing individual bytes was faster than an unrolled loop calling R_DrawColumnPixel. An unrolled loop writing individual bytes was faster still.
These are 32-bit Open Watcom results, in DOSbox-X, acting as a 25 MHz 386DX. Nonetheless it has gone from 14 FPS to 19 FPS on demo3.