Module ops 

GTE function-op wrappers.

Each op corresponds to one COP2 cofun instruction. The constants below are the exact 32-bit encodings PSX-SPX documents – the same bits the MIPS CPU would dispatch on hardware and the same bits the emulator’s Gte::execute decodes. On host we hand them straight to the per-thread software GTE in crate::host; on MIPS we emit them as .word instructions so LLVM’s integrated assembler can’t reject the COP2 mnemonics it doesn’t recognise.

Encoding (per PSX-SPX, “GTE Coprocessor Opcodes”):

  bits 31..26 = 0x12 (010010)      -- COP2
  bit  25     = 1                   -- `cop2 cofun` prefix
  bit  19     = sf                  -- fraction-shift
  bits 18..17 = mx                  -- MVMVA matrix select
  bits 16..15 = vx                  -- MVMVA vector select
  bits 14..13 = cv                  -- MVMVA translation select
  bit  10     = lm                  -- IR lower-bound mode
  bits 5..0   = opcode              -- function id

The 0x4A000000 base (0x12<<26 | 1<<25) is common to every op.

Functions

average_z3^⚠

Average SZ1..SZ3 weighted by ZSF3 into OTZ and MAC0 (AVSZ3).

average_z4^⚠

Average SZ0..SZ3 weighted by ZSF4 into OTZ and MAC0 (AVSZ4).

color_color^⚠

Light IR through the light color matrix and modulate by RGBC (CC).

color_depth_cue^⚠

color_color followed by a depth cue toward the far color (CDP).

depth_cue_light^⚠

Depth-cue a lit color: RGBC*IR toward FC by IR0 (DCPL).

depth_cue_single^⚠

Depth-cue RGBC toward the far color by IR0 (DPCS).

depth_cue_triple^⚠

depth_cue_single run three times against the RGB FIFO (DPCT).

interpolate_far_color^⚠

Interpolate IR toward the far color by IR0, then push the color FIFO (INTPL).

light_color_depth_single^⚠

Light, color and depth-cue one vertex normal (NCDS, sf=1, lm=0).

light_color_depth_triple^⚠

light_color_depth_single for V0, V1 and V2 (NCDT).

light_color_single^⚠

Light and color one vertex normal, no depth cue (NCCS).

light_color_triple^⚠

light_color_single for V0, V1 and V2 (NCCT).

light_single^⚠

Light one vertex normal, without the RGBC modulate (NCS).

light_triple^⚠

light_single for V0, V1 and V2 (NCT).

nccs^⚠Deprecated

Renamed to light_color_single.

ncct^⚠Deprecated

Renamed to light_color_triple.

outer_product^⚠

Outer product of IR with the rotation matrix diagonal (OP, sf=1).

project_single^⚠

Perspective-project V0 (RTPS, sf=1, lm=0).

project_triple^⚠

Perspective-project V0, V1 and V2 in sequence (RTPT, sf=1, lm=0).

rotate_translate_v0^⚠

Rotate V0 by RT and add TR, without the perspective divide (MVMVA with mx=RT, vx=V0, cv=TR, sf=1, lm=0).

rotate_v0_far_color^⚠

Rotate V0 by RT with the far color as the translation (MVMVA with mx=RT, vx=V0, cv=FC, sf=1, lm=0).

rtps^⚠Deprecated

Renamed to project_single.

rtpt^⚠Deprecated

Renamed to project_triple.

scale_vector^⚠

Scale the IR vector by IR0 into MAC, then push the color FIFO (GPF, sf=1).

scale_vector_accumulate^⚠

Add the IR vector scaled by IR0 to MAC, then push the color FIFO (GPL, sf=1).

scale_vector_unshifted^⚠

Scale the IR vector by IR0 into MAC, then push the color FIFO, without the fractional right shift (GPF, sf=0).

screen_winding^⚠

Winding of the three SXY entries (NCLIP): the Z component of (SXY1-SXY0) × (SXY2-SXY0) into MAC0.

square^⚠

Square the current IR vector into MAC1/2/3 (SQR, sf=1).

square_unshifted^⚠

Square the current IR vector into MAC1/2/3 without the fractional right shift (SQR, sf=0).