refactor: color blending function

[BharatSahlot]

Results in a small speedup of 100-200ms on most example files in 1080p. Sometimes the speedup is more visible like for eye example, the speedup was of 3secs.

[BharatSahlot]

@rodolforg @ice0 I am not sure why the Autotools run is failing, it looks like its not due to my changes.

[ice0]

@BharatSahlot

Can you add test measurements (before/after) with test files so this can be reproduced?

For example: #2831

[BharatSahlot]

@BharatSahlot

Can you add test measurements (before/after) with test files so this can be reproduced?

For example: #2831

Sure, but I don't have window and mac to test, I can only test on linux.

[ice0]

Sure, but I don't have window and mac to test, I can only test on linux.

No problems I will do tests on Windows/macOS, after you will done with Linux.

[BharatSahlot]

The performance varies between files, in the example files I saw increase anywhere between 0.7% to 26%. Below is the time taken by some example files on linux release build. A is this branch and B is the master branch. I used ./test_null.sh for testing.

Rendering files in ./files/examples. Width: 1920, Height: 1080
===================================
Testing ./files/examples/backdrop.sif
Time taken by A: 964.966
Time taken by B: 1192.06
Speed increase: 23.533
===================================
Testing ./files/examples/eye.sif
Time taken by A: 25778.8
Time taken by B: 28997.7
Speed increase: 12.486
===================================
Testing ./files/examples/pirates.sif
Time taken by A: 882.835
Time taken by B: 1033.49
Speed increase: 17.064
===================================
Testing ./files/examples/sparkle.sif
Time taken by A: 1133.12
Time taken by B: 1251.25
Speed increase: 10.425
===================================
Testing ./files/examples/z_depth_test.sif
Time taken by A: 412.215
Time taken by B: 494.878
Speed increase: 20.053

Time taken by all files can be found here: https://pastebin.com/besNtVJF

[rodolforg]

@BharatSahlot As I can see in the test_null.sh you are using different renderers (gl vs. software).
Shouldn't you compare by using the same renderer with/without this PR changes?

[BharatSahlot]

@BharatSahlot As I can see in the test_null.sh you are using different renderers (gl vs. software). Shouldn't you compare by using the same renderer with/without this PR changes?

Hi @rodolforg , when I test this branch I use the same renderer(default/software). That change was because I was testing my opengl renderer, I was just lazy and wrote it there directly.