[GPU] More specific note about slope-scaled depth bias
This commit is contained in:
parent
dce6938827
commit
c4a5048e62
|
@ -423,6 +423,8 @@ XE_GPU_REGISTER(0x2325, kDword, RB_SAMPLE_COUNT_ADDR)
|
|||
|
||||
// Polygon offset scales and offsets are 32-bit floating-point.
|
||||
// "slope computed in subpixels (1/12 or 1/16)" - R5xx Acceleration.
|
||||
// But the correct scale for conversion of the slope scale (FRONT_BACK/SCALE)
|
||||
// from subpixels to pixels is likely 1/16 according to:
|
||||
// https://github.com/mesa3d/mesa/blob/54ad9b444c8e73da498211870e785239ad3ff1aa/src/gallium/drivers/radeonsi/si_state.c#L946
|
||||
XE_GPU_REGISTER(0x2380, kFloat, PA_SU_POLY_OFFSET_FRONT_SCALE)
|
||||
XE_GPU_REGISTER(0x2381, kFloat, PA_SU_POLY_OFFSET_FRONT_OFFSET)
|
||||
|
|
Loading…
Reference in New Issue