AstralTransRocketries/ClassiCube
1
0
Fork 0
mirror of https://github.com/ClassiCube/ClassiCube.git synced 2025-09-07 22:34:43 -04:00
Code Issues Packages Projects Releases Wiki Activity

N64: Slightly optimise clipping path

Browse Source
This commit is contained in:
UnknownShadow200 2025-07-19 10:35:28 +10:00
parent 952824356e
commit 00a1a49405
1 changed files with 24 additions and 43 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

67
misc/n64/rsp_gpu.S
View File

@ -66,7 +66,6 @@ GL_STATE:
.align 4 .align 4
CLIP_CODE_FACTORS: .half 1, 1, GUARD_BAND_FACTOR, GUARD_BAND_FACTOR CLIP_CODE_FACTORS: .half 1, 1, GUARD_BAND_FACTOR, GUARD_BAND_FACTOR
DRAW_TRI_RA: .word 0
.bss .bss
.align 3 .align 3
@ -371,6 +370,7 @@ GPUCmd_DrawQuad:
#define v1_cflags t2 #define v1_cflags t2
#define v2_cflags t3 #define v2_cflags t3
#define v3_cflags t4 #define v3_cflags t4
// t5 is used by GL_ClipTriangle
addi src_ptr, rspq_dmem_buf_ptr, %lo(RSPQ_DMEM_BUFFER) - 64 addi src_ptr, rspq_dmem_buf_ptr, %lo(RSPQ_DMEM_BUFFER) - 64
li vtx, %lo(VERTEX_CACHE) li vtx, %lo(VERTEX_CACHE)
@ -491,12 +491,7 @@ GPUCmd_DrawQuad:
#undef vpos #undef vpos
#undef vcspos_i #undef vcspos_i
#undef vcspos_f #undef vcspos_f
#define vtx1 a1
#define vtx2 a2
#define vtx3 a3
#define vtx4 a0
// ######################## // ########################
// Trivial rejection check // Trivial rejection check
@ -534,54 +529,41 @@ GPUCmd_DrawQuad:
or tmp, v2_cflags or tmp, v2_cflags
or tmp, v3_cflags or tmp, v3_cflags
beqz tmp, DrawQuadForRDP
nop
#undef tmp
#undef v0_cflags #undef v0_cflags
#undef v1_cflags #undef v1_cflags
#undef v2_cflags #undef v2_cflags
#undef v3_cflags #undef v3_cflags
// If all 4 vertices are inside guardband, no need to clip
beqz tmp, DrawQuadForRDP
move t5, tmp // GL_ClipTriangle expects this in t5 instead
#undef tmp
// ########################### // ###########################
// Slow clipped triangle path // Slow clipped triangles path
// ########################### // ###########################
li a1, %lo(VERTEX_CACHE) + V0_OFFSET #define vtx1 a1
li a2, %lo(VERTEX_CACHE) + V1_OFFSET #define vtx2 a2
li a3, %lo(VERTEX_CACHE) + V2_OFFSET #define vtx3 a3
jal DrawClippedQuad #define vtx4 a0
li a0, %lo(VERTEX_CACHE) + V3_OFFSET
j RSPQ_Loop li vtx1, %lo(VERTEX_CACHE) + V0_OFFSET
nop li vtx2, %lo(VERTEX_CACHE) + V1_OFFSET
.endfunc li vtx3, %lo(VERTEX_CACHE) + V2_OFFSET
li vtx4, %lo(VERTEX_CACHE) + V3_OFFSET
################################################################
# DrawClippedTriangle - Breaks a triangle into one or more clipped tris
################################################################
.func DrawClippedQuad
DrawClippedQuad:
sw ra, %lo(DRAW_TRI_RA) // TODO find a register for this
lbu t0, SCREEN_VTX_CLIP_CODE(vtx1)
lbu t1, SCREEN_VTX_CLIP_CODE(vtx2)
lbu t2, SCREEN_VTX_CLIP_CODE(vtx3)
lbu t3, SCREEN_VTX_CLIP_CODE(vtx4)
or t5, t0, t1
or t5, t2
or t5, t3
// t5 = which guardband planes need to be clipped against
move s1, zero move s1, zero
jal GL_ClipTriangle jal GL_ClipTriangle
move s2, zero move s2, zero
beqz v1, gl_draw_triangle_end beqz v1, gl_draw_triangle_end
addi s2, -6 addi s2, -6
lhu s5, 0(s1) lhu s5, 0(s1)
gl_draw_clipped_triangles_loop: gl_draw_clipped_triangles_loop:
move vtx1, s5 move vtx1, s5
lhu vtx2, 2(s1) lhu vtx2, 2(s1)
lhu vtx3, 4(s1) lhu vtx3, 4(s1)
gl_draw_single_triangle: gl_draw_single_triangle:
addi vtx1, SCREEN_VTX_X addi vtx1, SCREEN_VTX_X
@ -591,17 +573,16 @@ gl_draw_single_triangle:
lhu a0, %lo(GL_TRI_CMD) lhu a0, %lo(GL_TRI_CMD)
lh v0, %lo(GL_TRI_CULL) lh v0, %lo(GL_TRI_CULL)
jal RDPQ_Triangle jal RDPQ_Triangle
li s3, %lo(RDPQ_CMD_STAGING) li s3, %lo(RDPQ_CMD_STAGING)
jal RDPQ_Send jal RDPQ_Send
li s4, %lo(RDPQ_CMD_STAGING) li s4, %lo(RDPQ_CMD_STAGING)
blt s1, s2, gl_draw_clipped_triangles_loop blt s1, s2, gl_draw_clipped_triangles_loop
addi s1, 2 addi s1, 2
gl_draw_triangle_end: gl_draw_triangle_end:
lw ra, %lo(DRAW_TRI_RA) j RSPQ_Loop
jr ra
nop nop
.endfunc .endfunc