RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-05-31 10:38:11 +00:00
Code Issues Activity Actions

LibGL: Use integer comparison for GL_EQUAL and GL_NOTEQUAL

Browse source 
This is an interesting quirk that occurs due to us using the x87 FPU
when Serenity is compiled for the i386 target. When we calculate our
depth value to be stored in the buffer, it is an 80-bit x87
floating point number, however, when stored into the DepthBuffer,
this is truncated to 32 bits. This 38 bit loss of precision means
that when x87 `FCOMP` is eventually used here the comparison fails.

This could be solved by using a `long double` for the depth buffer,
however this would take up significantly more space and is completely
overkill for a depth buffer. As such, comparing the first 32-bits of
this depth value is "good enough" that if we get a hit on it being
equal, we can pretty much guarantee that it's actually equal.
This commit is contained in:
Jesse Buhagiar 2021-08-20 21:19:42 +10:00 • committed by Ali Mohammad Pur
parent 2fe5f1528f
commit 0152f1924b
1 changed files with 19 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

19
Userland/Libraries/LibGL/SoftwareRasterizer.cpp
View file

@ -278,10 +278,29 @@ static void rasterize_triangle(const RasterizerOptions& options, Gfx::Bitmap& re
pass = z >= *depth;
break;
case GL_NOTEQUAL:
#ifdef __SSE__
pass = z != *depth;
#else
pass = bit_cast<u32>(z) != bit_cast<u32>(*depth);
#endif
break;
case GL_EQUAL:
#ifdef __SSE__
pass = z == *depth;
#else
//
// This is an interesting quirk that occurs due to us using the x87 FPU when Serenity is
// compiled for the i386 target. When we calculate our depth value to be stored in the buffer,
// it is an 80-bit x87 floating point number, however, when stored into the DepthBuffer, this is
// truncated to 32 bits. This 38 bit loss of precision means that when x87 `FCOMP` is eventually
// used here the comparison fails.
// This could be solved by using a `long double` for the depth buffer, however this would take
// up significantly more space and is completely overkill for a depth buffer. As such, comparing
// the first 32-bits of this depth value is "good enough" that if we get a hit on it being
// equal, we can pretty much guarantee that it's actually equal.
//
pass = bit_cast<u32>(z) == bit_cast<u32>(*depth);
#endif
break;
case GL_LEQUAL:
pass = z <= *depth;