64-Bit Distance Interpolation - Intel i86W Manual

FLOATING-POINT INSTRUCTIONS
Normally, two
faddz
instructions are executed consecutively. The shifting of MERGE
causes the results of consecutive
faddz
instructions to be accumulated in the MERGE
register.
32-bit Z-buffers can use 32-bit or 64-bit distance interpolation. For 32-bit interpolation,
no special instructions are required. Two 32-bit adds can be performed as an 64-bit add
instruction. The fact that data is carried from the low-order 32-bits into the high-order
32-bits may introduce an insignificant distortion into the interpolation.
For 32-bit Z-buffers, 64-bit distance interpolation is implemented (as Figure 6-10 shows)
with two 64-bit
fiadd
instructions. The merging is implemented with the 32-bit move
fmov.ss
fsrcl, fdest.
63 31
0
63 31
0
I
INTEGER
FRACTION
I I
INTEGER
!
FRACTION
++++++
fladd.dd ++++++
++++++
fladd.dd
++++++
63
0
63
0
I~EGER FRACTION
INTEGER
FRACTION
o
o
INTEGER
FRACTION
INTEGER
FRACTION
63
o
INTEGER
INTEGER
240329i
Figure 6-10. 64-Bit Distance Interpolation
6-37