2 * Copyright (C) 2011 Gilead Kutnick "Exophase" <exophase@gmail.com>
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public License as
6 * published by the Free Software Foundation; either version 2 of
7 * the License, or (at your option) any later version.
9 * This program is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * General Public License for more details.
17 #define MAX_BLOCKS_PER_ROW 128
19 #define psx_gpu_test_mask_offset 0
20 #define psx_gpu_uvrg_offset 16
21 #define psx_gpu_uvrg_dx_offset 32
22 #define psx_gpu_uvrg_dy_offset 48
23 #define psx_gpu_u_block_span_offset 64
24 #define psx_gpu_v_block_span_offset 80
25 #define psx_gpu_r_block_span_offset 96
26 #define psx_gpu_g_block_span_offset 112
27 #define psx_gpu_b_block_span_offset 128
29 #define psx_gpu_b_dx_offset 132
31 #define psx_gpu_b_offset 144
32 #define psx_gpu_b_dy_offset 148
33 #define psx_gpu_triangle_area_offset 152
34 #define psx_gpu_texture_window_settings_offset 156
35 #define psx_gpu_current_texture_mask_offset 160
36 #define psx_gpu_viewport_mask_offset 164
37 #define psx_gpu_dirty_textures_4bpp_mask_offset 168
38 #define psx_gpu_dirty_textures_8bpp_mask_offset 172
39 #define psx_gpu_dirty_textures_8bpp_alternate_mask_offset 176
40 #define psx_gpu_triangle_color_offset 180
41 #define psx_gpu_dither_table_offset 184
42 #define psx_gpu_render_block_handler_offset 200
43 #define psx_gpu_texture_page_ptr_offset 204
44 #define psx_gpu_texture_page_base_offset 208
45 #define psx_gpu_clut_ptr_offset 212
46 #define psx_gpu_vram_ptr_offset 216
48 #define psx_gpu_render_state_base_offset 220
49 #define psx_gpu_render_state_offset 222
50 #define psx_gpu_num_spans_offset 224
51 #define psx_gpu_num_blocks_offset 226
52 #define psx_gpu_offset_x_offset 228
53 #define psx_gpu_offset_y_offset 230
54 #define psx_gpu_clut_settings_offset 232
55 #define psx_gpu_texture_settings_offset 234
56 #define psx_gpu_viewport_start_x_offset 236
57 #define psx_gpu_viewport_start_y_offset 238
58 #define psx_gpu_viewport_end_x_offset 240
59 #define psx_gpu_viewport_end_y_offset 242
60 #define psx_gpu_mask_msb_offset 244
62 #define psx_gpu_triangle_winding_offset 246
63 #define psx_gpu_display_area_draw_enable_offset 247
64 #define psx_gpu_current_texture_page_offset 248
65 #define psx_gpu_last_8bpp_texture_page_offset 249
66 #define psx_gpu_texture_mask_width_offset 250
67 #define psx_gpu_texture_mask_height_offset 251
68 #define psx_gpu_texture_window_x_offset 252
69 #define psx_gpu_texture_window_y_offset 253
70 #define psx_gpu_primitive_type_offset 254
72 #define psx_gpu_reserved_a_offset 255
74 #define psx_gpu_blocks_offset 0x0100
75 #define psx_gpu_span_uvrg_offset_offset 0x2100
76 #define psx_gpu_span_edge_data_offset 0x4100
77 #define psx_gpu_span_b_offset_offset 0x5100
79 #define edge_data_left_x_offset 0
80 #define edge_data_num_blocks_offset 2
81 #define edge_data_right_mask_offset 4
82 #define edge_data_y_offset 6
126 #define gw_by_l gw_bx_l
146 #define uvrg_xxxx0 q3
151 #define uvrg_xxxx1 q4
156 #define uvrg_xxxx2 q5
161 #define yyyy_uvrg0 q6
166 #define yyyy_uvrg1 q7
171 #define yyyy_uvrg2 q8
198 #define uvrg_base q11
244 /* FIXME: users of this should be in psx_gpu instead */
246 #define load_pointer(register, pointer) \
247 movw register, :lower16:pointer; \
248 movt register, :upper16:pointer; \
251 #define load_pointer(register, pointer) \
252 ldr register, =pointer \
256 #define function(name) \
265 function(compute_all_gradients)
266 // First compute the triangle area reciprocal and shift. The division will
267 // happen concurrently with much of the work which follows.
268 @ r12 = psx_gpu->triangle_area
269 ldr r12, [ psx_gpu, #psx_gpu_triangle_area_offset ]
270 stmdb sp!, { r4 - r11, lr }
272 @ load exponent of 62 into upper half of double
274 clz r14, r12 @ r14 = shift
276 movt r4, #((62 + 1023) << 4)
277 mov r12, r12, lsl r14 @ r12 = triangle_area_normalized
279 @ load area normalized into lower half of double
281 vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n
283 movt r4, #((1022 + 31) << 4)
286 add r4, r4, r12, lsr #11
289 vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n
291 // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) =
292 // ( d0 * d1 ) - ( d2 * d3 ) =
293 // ( m0 ) - ( m1 ) = gradient
295 // This is split to do 12 elements at a time over three sets: a, b, and c.
296 // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so
297 // two of the slots are unused.
299 // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as
302 // First type is: uvrg bxxx xxxx
303 // Second type is: yyyy ybyy uvrg
304 // Since x_a and y_c are the same the same variable is used for both.
306 vld1.u32 { v0 }, [ v_a, : 128 ] @ v0 = { uvrg0, b0, x0, y0 }
307 ldrsh x0, [ v_a, #8 ] @ load x0
309 vld1.u32 { v1 }, [ v_b, : 128 ] @ v1 = { uvrg1, b1, x1, y1}
310 ldrh x1, [ v_b, #8 ] @ load x1
312 vld1.u32 { v2 }, [ v_c, : 128 ] @ v2 = { uvrg2, b2, x2, y2 }
313 ldrh x2, [ v_c, #8 ] @ load x2
315 vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- }
316 ldrh y0, [ v_a, #10 ] @ load y0
318 vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- }
319 ldrh y1, [ v_b, #10 ] @ load y1
321 vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- }
322 ldrh y2, [ v_c, #10 ] @ load y2
324 vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 }
325 vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 }
327 orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 }
328 pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 }
330 vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 }
331 vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 }
333 vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 }
334 vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 }
336 ldrb b2, [ v_c, #4 ] @ load b2
337 orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 }
339 ldrb b1, [ v_b, #4 ] @ load b1
340 orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 }
342 vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 }
343 vsub.s16 d0_ab, x1_ab, x0_ab
345 ldrb b0, [ v_a, #4 ] @ load b0
346 orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 }
348 vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 }
349 vsub.s16 d2_ab, x2_ab, x1_ab
351 vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 }
352 vsub.s16 d1_ab, y2_ab, y1_ab
354 orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 }
355 ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 }
357 ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 }
358 ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 }
360 vsub.s16 d3_ab, y1_ab, y0_ab
361 smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) -
362 @ ((x2 - X1) * (b1 - b0))
363 vmull.s16 ga_uvrg_x, d0_a, d1_a
364 smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) -
365 @ ((b2 - b1) * (y1 - y0))
366 vmlsl.s16 ga_uvrg_x, d2_a, d3_a
367 movs gs_bx, ga_bx, asr #31
369 vmull.s16 ga_uvrg_y, d0_b, d1_b
370 rsbmi ga_bx, ga_bx, #0
372 vmlsl.s16 ga_uvrg_y, d2_b, d3_b
373 movs gs_by, ga_by, asr #31
375 vshr.u64 d0, d30, #22
376 mov b_base, b0, lsl #16
378 rsbmi ga_by, ga_by, #0
379 vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0
381 @ r12 = psx_gpu->triangle_winding_offset
382 ldrb r12, [ psx_gpu, #psx_gpu_triangle_winding_offset ]
383 vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0
385 add b_base, b_base, #0x8000
386 rsb r12, r12, #0 @ r12 = -(triangle->winding)
388 vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w }
389 sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS)
391 vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16
392 vdup.u32 r_shift, r14 @ r_shift = { shift, shift, shift, shift }
394 vorr.u32 uvrg_base, #0x8000
395 vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x)
397 vmov area_r_s, s0 @ area_r_s = triangle_reciprocal
398 vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y)
400 vmull.u32 gw_rg_x, ga_rg_x, d0[0]
401 vmull.u32 gw_uv_x, ga_uv_x, d0[0]
402 vmull.u32 gw_rg_y, ga_rg_y, d0[0]
403 vmull.u32 gw_uv_y, ga_uv_y, d0[0]
405 vshl.u64 gw_rg_x, gw_rg_x, r_shift
406 vshl.u64 gw_uv_x, gw_uv_x, r_shift
407 vshl.u64 gw_rg_y, gw_rg_y, r_shift
408 vshl.u64 gw_uv_y, gw_uv_y, r_shift
410 veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask
411 vmovn.u64 g_uv_x, gw_uv_x
413 veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask
414 vmovn.u64 g_rg_x, gw_rg_x
416 veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
417 vmovn.u64 g_uv_y, gw_uv_y
419 vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
420 vmovn.u64 g_rg_y, gw_rg_y
422 veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
423 mov ga_bx, ga_bx, lsl #13
425 vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
426 mov ga_by, ga_by, lsl #13
429 umull gw_bx_l, gw_bx_h, ga_bx, area_r_s
431 vshl.u32 g_uvrg_x, g_uvrg_x, #4
432 vshl.u32 g_uvrg_y, g_uvrg_y, #4
434 umull gw_by_l, gw_by_h, ga_by, area_r_s
435 vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0]
437 eor gs_bx, gs_bx, r12
438 vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1
440 veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0
441 eor gs_by, gs_by, r12
443 rsb r11, r14, #0 @ r11 = negative shift for scalar lsr
444 add store_a, psx_gpu, #psx_gpu_uvrg_offset
446 sub r11, r11, #(32 - 13)
448 add store_b, store_a, #16
451 vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1
452 vst1.u32 { uvrg_base }, [ store_a, : 128 ], store_inc
454 vst1.u32 { uvrg_dx1 }, [ store_b, : 128 ], store_inc
455 mov g_bx, gw_bx_h, lsr r11
457 vst1.u32 { g_uvrg_y }, [ store_a, : 128 ], store_inc
458 mov g_by, gw_by_h, lsr r11
460 vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \
461 [ store_b, : 128 ], store_inc
462 eor g_bx, g_bx, gs_bx
464 vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \
465 [ store_b, : 128 ], store_inc
466 sub g_bx, g_bx, gs_bx
469 eor g_by, g_by, gs_by
471 mls b_base, g_bx, x0, b_base
472 sub g_by, g_by, gs_by
477 add g_bx2, g_bx, g_bx
478 add g_bx3, g_bx, g_bx2
480 stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by }
482 ldmia sp!, { r4 - r11, pc }
499 #define height_minor_a r7
500 #define height_minor_b r8
501 #define height_major r9
504 #define reciprocal_table_ptr r10
506 #define edge_alt_low r4
507 #define edge_alt_high r5
508 #define edge_dx_dy_alt r6
509 #define edge_shift_alt r10
511 #define edge_dx_dy_alt_low r4
512 #define edge_dx_dy_alt_high r5
514 #define span_edge_data r4
515 #define span_uvrg_offset r5
516 #define span_b_offset r6
524 #define alternate_x q0
525 #define alternate_dx_dy q1
526 #define alternate_x_32 q2
528 #define alternate_x_low d0
529 #define alternate_x_high d1
530 #define alternate_dx_dy_low d2
531 #define alternate_dx_dy_high d3
532 #define alternate_x_32_low d4
533 #define alternate_x_32_high d5
537 #define left_dx_dy q5
538 #define right_dx_dy q6
540 #define right_edge q8
542 #define left_x_low d6
543 #define left_x_high d7
544 #define right_x_low d8
545 #define right_x_high d9
546 #define left_dx_dy_low d10
547 #define left_dx_dy_high d11
548 #define right_dx_dy_low d12
549 #define right_dx_dy_high d13
550 #define left_edge_low d14
551 #define left_edge_high d15
552 #define right_edge_low d16
553 #define right_edge_high d17
555 #define y_mid_point d18
558 #define left_right_x_16 q11
559 #define span_shifts_y q12
562 #define span_shifts d24
567 #define left_right_x_16_low d22
568 #define left_right_x_16_high d23
573 #define alternate_x_16 d4
576 #define v_clip_low d6
578 #define right_x_32 q10
579 #define left_x_32 q11
580 #define alternate_select d24
582 #define right_x_32_low d20
583 #define right_x_32_high d21
584 #define left_x_32_low d22
585 #define left_x_32_high d23
588 #define edges_dx_dy d2
589 #define edge_shifts d3
590 #define edge_shifts_64 q2
592 #define edges_xy_left d0
593 #define edges_xy_right d1
595 #define height_reciprocals d6
603 #define heights_b d12
604 #define edges_dx_dy_64 q10
606 #define edges_dx_dy_64_left d20
607 #define edges_dx_dy_64_right d21
610 #define setup_spans_prologue() \
611 stmdb sp!, { r4 - r11, lr }; \
613 ldrsh x_a, [ v_a, #8 ]; \
614 ldrsh x_b, [ v_b, #8 ]; \
615 ldrsh x_c, [ v_c, #8 ]; \
616 ldrsh y_a, [ v_a, #10 ]; \
617 ldrsh y_b, [ v_b, #10 ]; \
618 ldrsh y_c, [ v_c, #10 ]; \
620 add temp, psx_gpu, #psx_gpu_uvrg_offset; \
621 vld1.32 { uvrg }, [ temp ]; \
622 add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \
623 vld1.32 { uvrg_dy }, [ temp ]; \
624 load_pointer(reciprocal_table_ptr, reciprocal_table); \
626 vmov.u32 c_0x01, #0x01 \
628 #define setup_spans_load_b() \
629 ldr b, [ psx_gpu, #psx_gpu_b_offset ]; \
630 ldr b_dy, [ psx_gpu, #psx_gpu_b_dy_offset ] \
632 #define setup_spans_prologue_b() \
633 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
634 add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \
636 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
637 vmov.u16 c_0x0004, #0x0004; \
639 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
640 vmov.u16 c_0x0001, #0x0001; \
642 vld1.u16 { left_edge_low[], left_edge_high[] }, [ temp ]; \
643 add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \
645 vld1.u16 { right_edge_low[], right_edge_high[] }, [ temp ]; \
646 vadd.u16 right_edge, right_edge, c_0x0001; \
648 vmov.u16 c_0x0007, #0x0007; \
649 vmvn.u16 c_0xFFFE, #0x0001 \
652 #define compute_edge_delta_x2() \
653 ldr temp, [ reciprocal_table_ptr, height, lsl #2 ]; \
655 vdup.u32 heights, height; \
656 vsub.u32 widths, x_ends, x_starts; \
658 vdup.u32 edge_shifts, temp; \
659 vsub.u32 heights_b, heights, c_0x01; \
660 vshr.u32 height_reciprocals, edge_shifts, #12; \
662 vmla.s32 heights_b, x_starts, heights; \
663 vbic.u16 edge_shifts, #0xE0; \
664 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
665 vmull.s32 edges_xy, heights_b, height_reciprocals \
668 #define height_reciprocal_alt r11
669 #define height_b_alt r12
671 #define compute_edge_delta_x3(start_c, height_a, height_b) \
672 vmov.u32 heights, height_a, height_b; \
673 ldr temp, [ reciprocal_table_ptr, height_a, lsl #2 ]; \
674 vmov.u32 edge_shifts[0], temp; \
675 ldr temp, [ reciprocal_table_ptr, height_b, lsl #2 ]; \
676 vmov.u32 edge_shifts[1], temp; \
677 ldr edge_shift_alt, [ reciprocal_table_ptr, height_minor_b, lsl #2 ]; \
679 vsub.u32 widths, x_ends, x_starts; \
680 sub width_alt, x_c, start_c; \
682 vsub.u32 heights_b, heights, c_0x01; \
683 sub height_b_alt, height_minor_b, #1; \
685 vshr.u32 height_reciprocals, edge_shifts, #12; \
686 lsr height_reciprocal_alt, edge_shift_alt, #12; \
688 vmla.s32 heights_b, x_starts, heights; \
689 mla height_b_alt, height_minor_b, start_c, height_b_alt; \
691 vbic.u16 edge_shifts, #0xE0; \
692 and edge_shift_alt, edge_shift_alt, #0x1F; \
694 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
695 mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \
697 vmull.s32 edges_xy, heights_b, height_reciprocals; \
698 smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \
701 #define setup_spans_adjust_y_up() \
702 vsub.u32 y_x4, y_x4, c_0x0004 \
704 #define setup_spans_adjust_y_down() \
705 vadd.u32 y_x4, y_x4, c_0x0004 \
707 #define setup_spans_adjust_interpolants_up() \
708 vsub.u32 uvrg, uvrg, uvrg_dy; \
711 #define setup_spans_adjust_interpolants_down() \
712 vadd.u32 uvrg, uvrg, uvrg_dy; \
716 #define setup_spans_clip_interpolants_increment() \
717 mla b, b_dy, clip, b; \
718 vmla.s32 uvrg, uvrg_dy, v_clip \
720 #define setup_spans_clip_interpolants_decrement() \
721 mls b, b_dy, clip, b; \
722 vmls.s32 uvrg, uvrg_dy, v_clip \
724 #define setup_spans_clip_alternate_yes() \
725 smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \
727 #define setup_spans_clip_alternate_no() \
729 #define setup_spans_clip(direction, alternate_active) \
730 vdup.u32 v_clip, clip; \
731 setup_spans_clip_alternate_##alternate_active(); \
732 setup_spans_clip_interpolants_##direction(); \
733 vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \
736 #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \
737 vmovl.s32 edge_shifts_64, edge_shifts; \
738 vmovl.s32 edges_dx_dy_64, edges_dx_dy; \
740 vshl.s64 edges_xy, edges_xy, edge_shifts_64; \
741 vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \
743 vmov left_x_low, edges_xy_##left_index; \
744 vmov right_x_low, edges_xy_##right_index; \
746 vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \
747 vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \
748 vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \
749 vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \
751 vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \
752 vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \
754 vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \
755 vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \
758 #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \
759 setup_spans_adjust_edges_alternate_no(left_index, right_index); \
761 vdup.u16 y_mid_point, y_b; \
762 rsb temp, edge_shift_alt, #32; \
764 lsl edge_alt_high, edge_alt_high, edge_shift_alt; \
765 orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \
766 lsl edge_alt_low, edge_alt_low, edge_shift_alt; \
767 vmov alternate_x_low, edge_alt_low, edge_alt_high; \
769 asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \
770 lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \
771 vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \
772 vmov alternate_dx_dy_high, alternate_dx_dy_low; \
774 vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \
775 vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \
778 #define setup_spans_y_select_up() \
779 vclt.s16 alternate_select, y_x4, y_mid_point \
781 #define setup_spans_y_select_down() \
782 vcgt.s16 alternate_select, y_x4, y_mid_point \
785 #define setup_spans_alternate_select_left() \
786 vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \
788 #define setup_spans_alternate_select_right() \
789 vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \
792 #define setup_spans_set_x4_alternate_yes(alternate, direction) \
793 vshrn.s64 alternate_x_32_low, alternate_x, #32; \
794 vshrn.s64 left_x_32_low, left_x, #32; \
795 vshrn.s64 right_x_32_low, right_x, #32; \
797 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
798 vadd.u64 left_x, left_x, left_dx_dy; \
799 vadd.u64 right_x, right_x, right_dx_dy; \
801 vshrn.s64 alternate_x_32_high, alternate_x, #32; \
802 vshrn.s64 left_x_32_high, left_x, #32; \
803 vshrn.s64 right_x_32_high, right_x, #32; \
805 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
806 vadd.u64 left_x, left_x, left_dx_dy; \
807 vadd.u64 right_x, right_x, right_dx_dy; \
809 vmovn.u32 alternate_x_16, alternate_x_32; \
810 setup_spans_y_select_##direction(); \
811 vmovn.u32 left_right_x_16_low, left_x_32; \
813 vmovn.u32 left_right_x_16_high, right_x_32; \
814 setup_spans_alternate_select_##alternate(); \
816 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
817 str b, [ span_b_offset ], #4; \
818 setup_spans_adjust_interpolants_##direction(); \
820 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
822 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
823 str b, [ span_b_offset ], #4; \
824 setup_spans_adjust_interpolants_##direction(); \
826 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
828 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
829 str b, [ span_b_offset ], #4; \
830 setup_spans_adjust_interpolants_##direction(); \
832 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
833 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
834 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
836 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
837 str b, [ span_b_offset ], #4; \
838 setup_spans_adjust_interpolants_##direction(); \
840 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
841 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
843 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
845 setup_spans_adjust_y_##direction() \
848 #define setup_spans_set_x4_alternate_no(alternate, direction) \
849 vshrn.s64 left_x_32_low, left_x, #32; \
850 vshrn.s64 right_x_32_low, right_x, #32; \
852 vadd.u64 left_x, left_x, left_dx_dy; \
853 vadd.u64 right_x, right_x, right_dx_dy; \
855 vshrn.s64 left_x_32_high, left_x, #32; \
856 vshrn.s64 right_x_32_high, right_x, #32; \
858 vadd.u64 left_x, left_x, left_dx_dy; \
859 vadd.u64 right_x, right_x, right_dx_dy; \
861 vmovn.u32 left_right_x_16_low, left_x_32; \
862 vmovn.u32 left_right_x_16_high, right_x_32; \
864 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
865 str b, [ span_b_offset ], #4; \
866 setup_spans_adjust_interpolants_##direction(); \
868 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
870 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
871 str b, [ span_b_offset ], #4; \
872 setup_spans_adjust_interpolants_##direction(); \
874 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
876 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
877 str b, [ span_b_offset ], #4; \
878 setup_spans_adjust_interpolants_##direction(); \
880 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
881 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
882 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
884 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
885 str b, [ span_b_offset ], #4; \
886 setup_spans_adjust_interpolants_##direction(); \
888 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
889 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
891 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
893 setup_spans_adjust_y_##direction() \
896 #define edge_adjust_low r11
897 #define edge_adjust_high r12
899 #define setup_spans_alternate_adjust_yes() \
900 smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \
901 subs edge_alt_low, edge_alt_low, edge_adjust_low; \
902 sbc edge_alt_high, edge_alt_high, edge_adjust_high \
904 #define setup_spans_alternate_adjust_no() \
907 #define setup_spans_down(left_index, right_index, alternate, alternate_active) \
908 setup_spans_alternate_adjust_##alternate_active(); \
909 setup_spans_load_b(); \
911 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
912 subs y_c, y_c, temp; \
913 subgt height, height, y_c; \
914 addgt height, height, #1; \
916 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
917 subs clip, temp, y_a; \
920 sub height, height, clip; \
921 add y_a, y_a, clip; \
922 setup_spans_clip(increment, alternate_active); \
928 orr temp, y_a, y_a, lsl #16; \
929 add temp, temp, #(1 << 16); \
931 add y_a, y_a, #(2 << 16); \
932 vmov.u32 y_x4, temp, y_a; \
934 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
936 setup_spans_prologue_b(); \
938 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
941 setup_spans_set_x4_alternate_##alternate_active(alternate, down); \
942 subs height, height, #4; \
948 #define setup_spans_alternate_pre_increment_yes() \
949 adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \
950 adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \
952 #define setup_spans_alternate_pre_increment_no() \
955 #define setup_spans_up_decrement_yes() \
956 suble height, height, #1 \
958 #define setup_spans_up_decrement_no() \
961 #define setup_spans_up(left_index, right_index, alternate, alternate_active) \
962 setup_spans_alternate_adjust_##alternate_active(); \
963 setup_spans_load_b(); \
966 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
967 subs temp, temp, y_c; \
968 subgt height, height, temp; \
969 setup_spans_up_decrement_##alternate_active(); \
971 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
972 subs clip, y_a, temp; \
975 sub height, height, clip; \
976 sub y_a, y_a, clip; \
977 setup_spans_clip(decrement, alternate_active); \
983 orr temp, y_a, y_a, lsl #16; \
984 sub temp, temp, #(1 << 16); \
986 sub y_a, y_a, #(2 << 16); \
987 vmov.u32 y_x4, temp, y_a; \
989 vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \
991 setup_spans_alternate_pre_increment_##alternate_active(); \
992 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
994 setup_spans_adjust_interpolants_up(); \
995 setup_spans_prologue_b(); \
997 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1000 setup_spans_set_x4_alternate_##alternate_active(alternate, up); \
1001 subs height, height, #4; \
1007 #define setup_spans_epilogue() \
1008 ldmia sp!, { r4 - r11, pc } \
1011 #define setup_spans_up_up(minor, major) \
1012 setup_spans_prologue(); \
1013 sub height_minor_a, y_a, y_b; \
1014 sub height_minor_b, y_b, y_c; \
1015 sub height, y_a, y_c; \
1017 vdup.u32 x_starts, x_a; \
1018 vmov.u32 x_ends, x_c, x_b; \
1020 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
1021 setup_spans_up(major, minor, minor, yes); \
1022 setup_spans_epilogue() \
1024 function(setup_spans_up_left)
1025 setup_spans_up_up(left, right)
1027 function(setup_spans_up_right)
1028 setup_spans_up_up(right, left)
1032 #define setup_spans_down_down(minor, major) \
1033 setup_spans_prologue(); \
1034 sub height_minor_a, y_b, y_a; \
1035 sub height_minor_b, y_c, y_b; \
1036 sub height, y_c, y_a; \
1038 vdup.u32 x_starts, x_a; \
1039 vmov.u32 x_ends, x_c, x_b; \
1041 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
1042 setup_spans_down(major, minor, minor, yes); \
1043 setup_spans_epilogue() \
1045 function(setup_spans_down_left)
1046 setup_spans_down_down(left, right)
1048 function(setup_spans_down_right)
1049 setup_spans_down_down(right, left)
1052 #define setup_spans_up_flat() \
1053 sub height, y_a, y_c; \
1055 compute_edge_delta_x2(); \
1056 setup_spans_up(left, right, none, no); \
1057 setup_spans_epilogue() \
1059 function(setup_spans_up_a)
1060 setup_spans_prologue()
1062 vmov.u32 x_starts, x_a, x_b
1063 vdup.u32 x_ends, x_c
1065 setup_spans_up_flat()
1067 function(setup_spans_up_b)
1068 setup_spans_prologue()
1070 vdup.u32 x_starts, x_a
1071 vmov.u32 x_ends, x_b, x_c
1073 setup_spans_up_flat()
1075 #define setup_spans_down_flat() \
1076 sub height, y_c, y_a; \
1078 compute_edge_delta_x2(); \
1079 setup_spans_down(left, right, none, no); \
1080 setup_spans_epilogue() \
1082 function(setup_spans_down_a)
1083 setup_spans_prologue()
1085 vmov.u32 x_starts, x_a, x_b
1086 vdup.u32 x_ends, x_c
1088 setup_spans_down_flat()
1090 function(setup_spans_down_b)
1091 setup_spans_prologue()
1093 vdup.u32 x_starts, x_a
1094 vmov.u32 x_ends, x_b, x_c
1096 setup_spans_down_flat()
1101 #define edges_xy_b q11
1102 #define edges_dx_dy_b d26
1103 #define edge_shifts_b d27
1104 #define edges_dx_dy_and_shifts_b q13
1105 #define height_increment d20
1107 #define edges_dx_dy_and_shifts q1
1109 #define edges_xy_b_left d22
1110 #define edges_xy_b_right d23
1112 #define setup_spans_up_down_load_edge_set_b() \
1113 vmov edges_xy, edges_xy_b; \
1114 vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \
1117 function(setup_spans_up_down)
1118 setup_spans_prologue()
1120 // s32 middle_y = y_a;
1121 sub height_minor_a, y_a, y_b
1122 sub height_minor_b, y_c, y_a
1123 sub height_major, y_c, y_b
1125 vmov.u32 x_starts, x_a, x_c
1126 vdup.u32 x_ends, x_b
1128 compute_edge_delta_x3(x_a, height_minor_a, height_major)
1131 vmov.u32 height_increment, temp, height_minor_b
1132 vmlal.s32 edges_xy, edges_dx_dy, height_increment
1134 vmov edges_xy_b_left, edge_alt_low, edge_alt_high
1135 vmov edges_xy_b_right, edges_xy_right
1137 vmov edge_shifts_b, edge_shifts
1138 vmov.u32 edge_shifts_b[0], edge_shift_alt
1140 vneg.s32 edges_dx_dy_b, edges_dx_dy
1141 vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt
1145 setup_spans_load_b()
1148 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1149 subs temp, temp, y_b
1150 subgt height_minor_a, height_minor_a, temp
1152 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1153 subs clip, y_a, temp
1156 sub height_minor_a, height_minor_a, clip
1158 setup_spans_clip(decrement, no)
1161 cmp height_minor_a, #0
1164 orr temp, y_a, y_a, lsl #16
1165 sub temp, temp, #(1 << 16)
1167 sub y_a, y_a, #(2 << 16)
1168 vmov.u32 y_x4, temp, y_a
1170 vaddw.s32 edges_xy, edges_xy, edges_dx_dy
1172 strh height_minor_a, [ psx_gpu, #psx_gpu_num_spans_offset ]
1174 setup_spans_adjust_edges_alternate_no(left, right);
1175 setup_spans_adjust_interpolants_up()
1176 setup_spans_up_down_load_edge_set_b()
1178 setup_spans_prologue_b()
1182 setup_spans_set_x4_alternate_no(none, up)
1183 subs height_minor_a, height_minor_a, #4
1186 add span_edge_data, span_edge_data, height_minor_a, lsl #3
1187 add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4
1188 add span_b_offset, span_b_offset, height_minor_a, lsl #2
1191 add temp, psx_gpu, #psx_gpu_uvrg_offset
1192 vld1.32 { uvrg }, [ temp ]
1195 setup_spans_load_b()
1197 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1199 subgt height_minor_b, height_minor_b, y_c
1200 addgt height_minor_b, height_minor_b, #1
1202 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1203 subs clip, temp, y_a
1206 sub height_minor_b, height_minor_b, clip
1208 setup_spans_clip(increment, no)
1211 cmp height_minor_b, #0
1214 orr temp, y_a, y_a, lsl #16
1215 add temp, temp, #(1 << 16)
1217 add y_a, y_a, #(2 << 16)
1218 vmov.u32 y_x4, temp, y_a
1220 setup_spans_adjust_edges_alternate_no(left, right)
1222 ldrh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1223 add temp, temp, height_minor_b
1224 strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1227 setup_spans_set_x4_alternate_no(none, down)
1228 subs height_minor_b, height_minor_b, #4
1232 setup_spans_epilogue()
1235 setup_spans_up_down_load_edge_set_b()
1236 setup_spans_prologue_b()
1241 #undef span_uvrg_offset
1242 #undef span_edge_data
1243 #undef span_b_offset
1248 #define num_spans r1
1249 #define span_uvrg_offset r2
1250 #define span_edge_data r3
1251 #define span_b_offset r4
1253 #define span_num_blocks r6
1257 #define dither_offset_ptr r10
1258 #define block_ptr_a r11
1260 #define num_blocks r14
1262 #define uvrg_dx_ptr r2
1263 #define texture_mask_ptr r3
1264 #define dither_shift r8
1265 #define dither_row r10
1270 #define block_ptr_b r10
1272 #define block_span_ptr r10
1273 #define right_mask r8
1292 #define b_whole_8 d14
1293 #define fb_mask_ptrs d15
1306 #define u_whole_low d16
1307 #define u_whole_high d17
1308 #define v_whole_low d18
1309 #define v_whole_high d19
1310 #define r_whole_low d20
1311 #define r_whole_high d21
1312 #define g_whole_low d22
1313 #define g_whole_high d23
1314 #define b_whole_low d24
1315 #define b_whole_high d25
1320 #define u_whole_8 d26
1321 #define v_whole_8 d27
1322 #define u_whole_8b d24
1323 #define r_whole_8 d24
1324 #define g_whole_8 d25
1326 #define uv_whole_8 q13
1327 #define uv_whole_8b q14
1329 #define dither_offsets q14
1330 #define texture_mask q15
1331 #define texture_mask_u d30
1332 #define texture_mask_v d31
1334 #define dither_offsets_short d28
1338 #define block_span q10
1343 #define draw_mask q1
1344 #define draw_mask_edge q13
1345 #define test_mask q0
1351 #define setup_blocks_texture_swizzled() \
1352 vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \
1353 vsli.u8 u_whole_8, v_whole_8, #4; \
1354 vsri.u8 v_whole_8, u_whole_8b, #4 \
1356 #define setup_blocks_texture_unswizzled() \
1359 #define setup_blocks_shaded_textured_builder(swizzling) \
1362 function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \
1363 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1364 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1366 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1367 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1369 cmp num_spans, #0; \
1372 stmdb sp!, { r4 - r11, r14 }; \
1373 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1375 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
1376 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1378 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1379 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1381 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1382 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1384 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
1385 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1387 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1390 vmov.u8 fb_mask_ptrs, #0; \
1392 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1393 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1395 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1396 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
1398 cmp span_num_blocks, #0; \
1401 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1402 add num_blocks, span_num_blocks, num_blocks; \
1404 cmp num_blocks, #MAX_BLOCKS; \
1408 ldr b, [ span_b_offset ]; \
1409 add fb_ptr, fb_ptr, y, lsl #11; \
1411 vdup.u32 v_left_x, left_x; \
1414 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1415 add fb_ptr, fb_ptr, left_x, lsl #1; \
1417 mla b, b_dx, left_x, b; \
1418 and dither_shift, left_x, #0x03; \
1420 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1421 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1423 mov dither_shift, dither_shift, lsl #3; \
1424 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1427 subs span_num_blocks, span_num_blocks, #1; \
1429 mov dither_row, dither_row, ror dither_shift; \
1430 mov b_dx4, b_dx, lsl #2; \
1432 vdup.u32 dither_offsets_short, dither_row; \
1433 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1435 vdup.u32 b_block, b; \
1436 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1438 vdup.u32 u_block, uv[0]; \
1439 mov b_dx8, b_dx, lsl #3; \
1441 vdup.u32 v_block, uv[1]; \
1442 vdup.u32 r_block, rg[0]; \
1443 vdup.u32 g_block, rg[1]; \
1445 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1447 vadd.u32 u_block, u_block, block_span; \
1448 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1450 vadd.u32 v_block, v_block, block_span; \
1451 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1453 vadd.u32 r_block, r_block, block_span; \
1454 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1456 vadd.u32 g_block, g_block, block_span; \
1457 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
1459 vadd.u32 b_block, b_block, block_span; \
1460 add block_ptr_b, block_ptr_a, #16; \
1462 vshrn.u32 u_whole_low, u_block, #16; \
1463 vshrn.u32 v_whole_low, v_block, #16; \
1464 vshrn.u32 r_whole_low, r_block, #16; \
1465 vshrn.u32 g_whole_low, g_block, #16; \
1467 vdup.u32 dx4, uv_dx4[0]; \
1468 vshrn.u32 b_whole_low, b_block, #16; \
1470 vaddhn.u32 u_whole_high, u_block, dx4; \
1471 vdup.u32 dx4, uv_dx4[1]; \
1473 vaddhn.u32 v_whole_high, v_block, dx4; \
1474 vdup.u32 dx4, rg_dx4[0]; \
1476 vaddhn.u32 r_whole_high, r_block, dx4; \
1477 vdup.u32 dx4, rg_dx4[1]; \
1479 vaddhn.u32 g_whole_high, g_block, dx4; \
1480 vdup.u32 dx4, b_dx4; \
1482 vaddhn.u32 b_whole_high, b_block, dx4; \
1483 vdup.u32 dx8, uv_dx8[0]; \
1485 vadd.u32 u_block, u_block, dx8; \
1486 vdup.u32 dx8, uv_dx8[1]; \
1488 vadd.u32 v_block, v_block, dx8; \
1489 vdup.u32 dx8, rg_dx8[0]; \
1491 vadd.u32 r_block, r_block, dx8; \
1492 vdup.u32 dx8, rg_dx8[1]; \
1494 vadd.u32 g_block, g_block, dx8; \
1495 vdup.u32 dx8, b_dx8; \
1497 vadd.u32 b_block, b_block, dx8; \
1498 vmovn.u16 u_whole_8, u_whole; \
1500 vmovn.u16 v_whole_8, v_whole; \
1502 vmovn.u16 b_whole_8, b_whole; \
1504 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1506 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1507 setup_blocks_texture_##swizzling(); \
1509 vmovn.u16 r_whole_8, r_whole; \
1513 vmovn.u16 g_whole_8, g_whole; \
1514 vshrn.u32 u_whole_low, u_block, #16; \
1516 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1517 vshrn.u32 v_whole_low, v_block, #16; \
1519 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1520 vshrn.u32 r_whole_low, r_block, #16; \
1522 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1523 vshrn.u32 g_whole_low, g_block, #16; \
1525 vdup.u32 dx4, uv_dx4[0]; \
1526 vshrn.u32 b_whole_low, b_block, #16; \
1528 vaddhn.u32 u_whole_high, u_block, dx4; \
1529 vdup.u32 dx4, uv_dx4[1]; \
1531 vaddhn.u32 v_whole_high, v_block, dx4; \
1532 vdup.u32 dx4, rg_dx4[0]; \
1534 vaddhn.u32 r_whole_high, r_block, dx4; \
1535 vdup.u32 dx4, rg_dx4[1]; \
1537 vaddhn.u32 g_whole_high, g_block, dx4; \
1538 vdup.u32 dx4, b_dx4; \
1540 vaddhn.u32 b_whole_high, b_block, dx4; \
1541 vdup.u32 dx8, uv_dx8[0]; \
1543 vadd.u32 u_block, u_block, dx8; \
1544 vdup.u32 dx8, uv_dx8[1]; \
1546 vadd.u32 v_block, v_block, dx8; \
1547 vdup.u32 dx8, rg_dx8[0]; \
1549 vadd.u32 r_block, r_block, dx8; \
1550 vdup.u32 dx8, rg_dx8[1]; \
1552 vadd.u32 g_block, g_block, dx8; \
1553 vdup.u32 dx8, b_dx8; \
1555 vadd.u32 b_block, b_block, dx8; \
1556 vmovn.u16 u_whole_8, u_whole; \
1558 add fb_ptr, fb_ptr, #16; \
1559 vmovn.u16 v_whole_8, v_whole; \
1561 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1562 vmovn.u16 b_whole_8, b_whole; \
1566 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1567 subs span_num_blocks, span_num_blocks, #1; \
1569 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1570 setup_blocks_texture_##swizzling(); \
1572 vmovn.u16 r_whole_8, r_whole; \
1576 vmovn.u16 g_whole_8, g_whole; \
1577 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1579 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1580 vdup.u8 draw_mask, right_mask; \
1582 vmov.u32 fb_mask_ptrs[0], right_mask; \
1583 vtst.u16 draw_mask, draw_mask, test_mask; \
1584 vzip.u8 u_whole_8, v_whole_8; \
1586 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1587 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1588 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1589 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1590 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1593 add span_uvrg_offset, span_uvrg_offset, #16; \
1594 add span_b_offset, span_b_offset, #4; \
1596 add span_edge_data, span_edge_data, #8; \
1597 subs num_spans, num_spans, #1; \
1599 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1602 ldmia sp!, { r4 - r11, pc }; \
1605 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1606 vpush { texture_mask }; \
1607 vpush { uvrg_dx4 }; \
1609 stmdb sp!, { r0 - r3, r12, r14 }; \
1610 bl flush_render_block_buffer; \
1611 ldmia sp!, { r0 - r3, r12, r14 }; \
1613 vpop { uvrg_dx4 }; \
1614 vpop { texture_mask }; \
1616 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1617 vmov.u8 fb_mask_ptrs, #0; \
1619 mov num_blocks, span_num_blocks; \
1620 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1624 setup_blocks_shaded_textured_builder(swizzled)
1625 setup_blocks_shaded_textured_builder(unswizzled)
1628 #define setup_blocks_unshaded_textured_builder(swizzling) \
1631 function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \
1632 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1633 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1635 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1636 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1638 cmp num_spans, #0; \
1641 stmdb sp!, { r4 - r11, r14 }; \
1642 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1644 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1646 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1647 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1649 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1650 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1652 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1654 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1657 vmov.u8 fb_mask_ptrs, #0; \
1659 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1660 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1662 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1663 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
1665 cmp span_num_blocks, #0; \
1668 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1669 add num_blocks, span_num_blocks, num_blocks; \
1671 cmp num_blocks, #MAX_BLOCKS; \
1675 add fb_ptr, fb_ptr, y, lsl #11; \
1677 vdup.u32 v_left_x, left_x; \
1680 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1681 add fb_ptr, fb_ptr, left_x, lsl #1; \
1683 and dither_shift, left_x, #0x03; \
1685 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1686 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1688 mov dither_shift, dither_shift, lsl #3; \
1689 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1692 subs span_num_blocks, span_num_blocks, #1; \
1694 mov dither_row, dither_row, ror dither_shift; \
1696 vdup.u32 dither_offsets_short, dither_row; \
1697 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1699 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1701 vdup.u32 u_block, uv[0]; \
1703 vdup.u32 v_block, uv[1]; \
1704 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1706 vadd.u32 u_block, u_block, block_span; \
1707 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1709 vadd.u32 v_block, v_block, block_span; \
1710 add block_ptr_b, block_ptr_a, #16; \
1712 vshrn.u32 u_whole_low, u_block, #16; \
1713 vshrn.u32 v_whole_low, v_block, #16; \
1715 vdup.u32 dx4, uv_dx4[0]; \
1717 vaddhn.u32 u_whole_high, u_block, dx4; \
1718 vdup.u32 dx4, uv_dx4[1]; \
1720 vaddhn.u32 v_whole_high, v_block, dx4; \
1721 vdup.u32 dx8, uv_dx8[0]; \
1723 vadd.u32 u_block, u_block, dx8; \
1724 vdup.u32 dx8, uv_dx8[1]; \
1726 vadd.u32 v_block, v_block, dx8; \
1727 vmovn.u16 u_whole_8, u_whole; \
1729 vmovn.u16 v_whole_8, v_whole; \
1732 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1734 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1735 setup_blocks_texture_##swizzling(); \
1740 vshrn.u32 u_whole_low, u_block, #16; \
1742 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1743 vshrn.u32 v_whole_low, v_block, #16; \
1745 add block_ptr_b, block_ptr_b, #32; \
1746 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1748 vdup.u32 dx4, uv_dx4[0]; \
1749 vaddhn.u32 u_whole_high, u_block, dx4; \
1750 vdup.u32 dx4, uv_dx4[1]; \
1752 vaddhn.u32 v_whole_high, v_block, dx4; \
1753 vdup.u32 dx8, uv_dx8[0]; \
1755 vadd.u32 u_block, u_block, dx8; \
1756 vdup.u32 dx8, uv_dx8[1]; \
1758 vadd.u32 v_block, v_block, dx8; \
1759 vmovn.u16 u_whole_8, u_whole; \
1761 add fb_ptr, fb_ptr, #16; \
1762 vmovn.u16 v_whole_8, v_whole; \
1764 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1767 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1768 subs span_num_blocks, span_num_blocks, #1; \
1770 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1771 setup_blocks_texture_##swizzling(); \
1776 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1778 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1779 vdup.u8 draw_mask, right_mask; \
1781 vmov.u32 fb_mask_ptrs[0], right_mask; \
1782 vtst.u16 draw_mask, draw_mask, test_mask; \
1783 vzip.u8 u_whole_8, v_whole_8; \
1785 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1786 add block_ptr_b, block_ptr_b, #32; \
1787 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1788 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1789 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1792 add span_uvrg_offset, span_uvrg_offset, #16; \
1793 add span_edge_data, span_edge_data, #8; \
1794 subs num_spans, num_spans, #1; \
1796 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1799 ldmia sp!, { r4 - r11, pc }; \
1802 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1803 vpush { texture_mask }; \
1804 vpush { uvrg_dx4 }; \
1806 stmdb sp!, { r0 - r3, r12, r14 }; \
1807 bl flush_render_block_buffer; \
1808 ldmia sp!, { r0 - r3, r12, r14 }; \
1810 vpop { uvrg_dx4 }; \
1811 vpop { texture_mask }; \
1813 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1814 vmov.u8 fb_mask_ptrs, #0; \
1816 mov num_blocks, span_num_blocks; \
1817 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1821 setup_blocks_unshaded_textured_builder(swizzled)
1822 setup_blocks_unshaded_textured_builder(unswizzled)
1827 function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect)
1828 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1829 veor.u32 draw_mask, draw_mask, draw_mask
1834 stmdb sp!, { r4 - r11, r14 }
1835 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1837 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1839 ubfx color_r, color, #3, #5
1840 ubfx color_g, color, #11, #5
1841 ubfx color_b, color, #19, #5
1843 orr color, color_r, color_b, lsl #10
1844 orr color, color, color_g, lsl #5
1846 vdup.u16 colors, color
1848 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1849 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1851 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1852 add block_ptr_a, block_ptr_a, num_blocks, lsl #6
1855 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1856 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1858 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
1860 cmp span_num_blocks, #0
1863 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1864 add num_blocks, span_num_blocks, num_blocks
1866 cmp num_blocks, #MAX_BLOCKS
1870 add fb_ptr, fb_ptr, y, lsl #11
1873 add fb_ptr, fb_ptr, left_x, lsl #1
1876 subs span_num_blocks, span_num_blocks, #1
1878 add block_ptr_b, block_ptr_a, #16
1881 vmov.u32 fb_mask_ptrs[1], fb_ptr
1885 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_32
1886 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1887 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1889 add fb_ptr, fb_ptr, #16
1890 add block_ptr_b, block_ptr_b, #32
1894 vmov.u32 fb_mask_ptrs[1], fb_ptr
1895 subs span_num_blocks, span_num_blocks, #1
1900 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1902 vdup.u8 draw_mask_edge, right_mask
1903 vtst.u16 draw_mask_edge, draw_mask_edge, test_mask
1905 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1906 vst1.u32 { draw_mask_edge }, [ block_ptr_a, :128 ], c_32
1907 add block_ptr_b, block_ptr_b, #32
1908 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1911 add span_edge_data, span_edge_data, #8
1912 subs num_spans, num_spans, #1
1914 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1917 ldmia sp!, { r4 - r11, pc }
1922 stmdb sp!, { r0 - r3, r12, r14 }
1923 bl flush_render_block_buffer
1924 ldmia sp!, { r0 - r3, r12, r14 }
1928 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1929 veor.u32 draw_mask, draw_mask, draw_mask
1931 mov num_blocks, span_num_blocks
1932 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1936 #define mask_msb_scalar r14
1938 #define msb_mask q15
1940 #define pixels_low d16
1942 #define msb_mask_low d30
1943 #define msb_mask_high d31
1948 function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct)
1949 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1954 stmdb sp!, { r4 - r11, r14 }
1956 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1958 ubfx color_r, color, #3, #5
1959 ubfx color_g, color, #11, #5
1961 ldrh mask_msb_scalar, [ psx_gpu, #psx_gpu_mask_msb_offset ]
1962 ubfx color_b, color, #19, #5
1964 orr color, color_r, color_b, lsl #10
1965 orr color, color, color_g, lsl #5
1966 orr color, color, mask_msb_scalar
1968 vdup.u16 colors, color
1970 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1971 orr color, color, lsl #16
1975 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1976 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1978 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
1980 cmp span_num_blocks, #0
1983 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1985 add fb_ptr, fb_ptr, y, lsl #11
1986 subs span_num_blocks, span_num_blocks, #1
1988 add fb_ptr, fb_ptr, left_x, lsl #1
1992 vst1.u32 { colors }, [ fb_ptr ]!
1993 subs span_num_blocks, span_num_blocks, #1
1998 ldrb right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
2000 cmp right_mask, #0x0
2003 tst right_mask, #0xF
2004 streq color, [ fb_ptr ], #4
2005 moveq right_mask, right_mask, lsr #4
2006 streq color, [ fb_ptr ], #4
2008 tst right_mask, #0x3
2009 streq color, [ fb_ptr ], #4
2010 moveq right_mask, right_mask, lsr #2
2012 tst right_mask, #0x1
2013 streqh color, [ fb_ptr ]
2016 add span_edge_data, span_edge_data, #8
2017 subs num_spans, num_spans, #1
2020 ldmia sp!, { r4 - r11, pc }
2023 vst1.u32 { colors }, [ fb_ptr ]
2030 #define rg_dx_ptr r2
2048 #undef dither_offsets
2068 #define r_whole_low d6
2069 #define r_whole_high d7
2070 #define g_whole_low d8
2071 #define g_whole_high d9
2072 #define b_whole_low d10
2073 #define b_whole_high d11
2075 #define gb_whole_8 q6
2077 #define g_whole_8 d12
2078 #define b_whole_8 d13
2080 #define r_whole_8 d14
2092 #define block_span q5
2100 #define d128_0x7 q13
2104 #define dither_offsets q14
2105 #define draw_mask q15
2107 #define dither_offsets_low d28
2110 #define test_mask q10
2113 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2114 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2115 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2117 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2118 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2119 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2121 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2123 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2126 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2129 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2130 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2131 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2133 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2135 cmp num_spans, #0; \
2138 stmdb sp!, { r4 - r11, r14 }; \
2139 vshl.u32 rg_dx4, rg_dx, #2; \
2141 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2142 vshl.u32 rg_dx8, rg_dx, #3; \
2144 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2146 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2147 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2149 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2150 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2152 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2153 vmov.u8 d64_1, #1; \
2155 vmov.u8 d128_4, #4; \
2156 vmov.u8 d64_128, #128; \
2158 vmov.u8 d128_0x7, #0x7; \
2161 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2162 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2164 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2165 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
2167 cmp span_num_blocks, #0; \
2170 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2171 add num_blocks, span_num_blocks, num_blocks; \
2173 cmp num_blocks, #MAX_BLOCKS; \
2177 ldr b, [ span_b_offset ]; \
2178 add fb_ptr, fb_ptr, y, lsl #11; \
2180 vdup.u32 v_left_x, left_x; \
2183 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2184 add fb_ptr, fb_ptr, left_x, lsl #1; \
2186 mla b, b_dx, left_x, b; \
2187 and dither_shift, left_x, #0x03; \
2189 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2190 vshr.u32 rg_dx, rg_dx4, #2; \
2192 mov dither_shift, dither_shift, lsl #3; \
2193 vmla.u32 rg, rg_dx, v_left_x; \
2196 subs span_num_blocks, span_num_blocks, #1; \
2198 mov dither_row, dither_row, ror dither_shift; \
2199 mov b_dx4, b_dx, lsl #2; \
2201 vdup.u32 dither_offsets, dither_row; \
2202 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2204 vdup.u32 b_block, b; \
2205 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2207 mov b_dx8, b_dx, lsl #3; \
2208 vdup.u32 r_block, rg[0]; \
2209 vdup.u32 g_block, rg[1]; \
2211 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2213 vadd.u32 r_block, r_block, block_span; \
2214 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2216 vadd.u32 g_block, g_block, block_span; \
2217 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2219 vadd.u32 b_block, b_block, block_span; \
2220 add block_ptr_b, block_ptr_a, #16; \
2222 vshrn.u32 r_whole_low, r_block, #16; \
2223 vshrn.u32 g_whole_low, g_block, #16; \
2224 vshrn.u32 b_whole_low, b_block, #16; \
2225 vdup.u32 dx4, rg_dx4[0]; \
2227 vaddhn.u32 r_whole_high, r_block, dx4; \
2228 vdup.u32 dx4, rg_dx4[1]; \
2230 vaddhn.u32 g_whole_high, g_block, dx4; \
2231 vdup.u32 dx4, b_dx4; \
2233 vaddhn.u32 b_whole_high, b_block, dx4; \
2234 vdup.u32 dx8, rg_dx8[0]; \
2236 vadd.u32 r_block, r_block, dx8; \
2237 vdup.u32 dx8, rg_dx8[1]; \
2239 vadd.u32 g_block, g_block, dx8; \
2240 vdup.u32 dx8, b_dx8; \
2242 vadd.u32 b_block, b_block, dx8; \
2244 vmovn.u16 r_whole_8, r_whole; \
2245 vmovn.u16 g_whole_8, g_whole; \
2246 vmovn.u16 b_whole_8, b_whole; \
2249 veor.u32 draw_mask, draw_mask, draw_mask; \
2252 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2253 vshrn.u32 r_whole_low, r_block, #16; \
2255 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2256 vshrn.u32 g_whole_low, g_block, #16; \
2258 vshrn.u32 b_whole_low, b_block, #16; \
2259 str fb_ptr, [ block_ptr_a, #44 ]; \
2261 vdup.u32 dx4, rg_dx4[0]; \
2262 vshr.u8 r_whole_8, r_whole_8, #3; \
2263 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2265 vaddhn.u32 r_whole_high, r_block, dx4; \
2266 vdup.u32 dx4, rg_dx4[1]; \
2268 vaddhn.u32 g_whole_high, g_block, dx4; \
2269 vdup.u32 dx4, b_dx4; \
2271 vaddhn.u32 b_whole_high, b_block, dx4; \
2272 vdup.u32 dx8, rg_dx8[0]; \
2274 vmull.u8 pixels, r_whole_8, d64_1; \
2275 vmlal.u8 pixels, g_whole_8, d64_4; \
2276 vmlal.u8 pixels, b_whole_8, d64_128; \
2278 vadd.u32 r_block, r_block, dx8; \
2279 vdup.u32 dx8, rg_dx8[1]; \
2281 vadd.u32 g_block, g_block, dx8; \
2282 vdup.u32 dx8, b_dx8; \
2284 vadd.u32 b_block, b_block, dx8; \
2285 add fb_ptr, fb_ptr, #16; \
2287 vmovn.u16 r_whole_8, r_whole; \
2288 vmovn.u16 g_whole_8, g_whole; \
2289 vmovn.u16 b_whole_8, b_whole; \
2291 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2292 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2296 subs span_num_blocks, span_num_blocks, #1; \
2300 str fb_ptr, [ block_ptr_a, #44 ]; \
2301 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2303 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2304 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2306 vshr.u8 r_whole_8, r_whole_8, #3; \
2307 vdup.u8 draw_mask, right_mask; \
2309 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2310 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
2312 vtst.u16 draw_mask, draw_mask, test_mask; \
2314 vmull.u8 pixels, r_whole_8, d64_1; \
2315 vmlal.u8 pixels, g_whole_8, d64_4; \
2316 vmlal.u8 pixels, b_whole_8, d64_128; \
2318 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2319 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2322 add span_uvrg_offset, span_uvrg_offset, #16; \
2323 add span_b_offset, span_b_offset, #4; \
2325 add span_edge_data, span_edge_data, #8; \
2326 subs num_spans, num_spans, #1; \
2328 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2331 ldmia sp!, { r4 - r11, pc }; \
2334 /* TODO: Load from psx_gpu instead of saving/restoring these */\
2337 stmdb sp!, { r0 - r3, r12, r14 }; \
2338 bl flush_render_block_buffer; \
2339 ldmia sp!, { r0 - r3, r12, r14 }; \
2343 vmov.u8 d64_1, #1; \
2344 vmov.u8 d128_4, #4; \
2345 vmov.u8 d64_128, #128; \
2346 vmov.u8 d128_0x7, #0x7; \
2348 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2350 mov num_blocks, span_num_blocks; \
2351 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2355 setup_blocks_shaded_untextured_indirect_builder(undithered)
2356 setup_blocks_shaded_untextured_indirect_builder(dithered)
2361 #define mask_msb_ptr r14
2363 #define draw_mask q0
2364 #define pixels_low d16
2365 #define pixels_high d17
2369 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2372 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2373 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2374 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2376 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2378 cmp num_spans, #0; \
2381 stmdb sp!, { r4 - r11, r14 }; \
2382 vshl.u32 rg_dx4, rg_dx, #2; \
2384 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2385 vshl.u32 rg_dx8, rg_dx, #3; \
2387 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2388 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2390 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2391 vmov.u8 d64_1, #1; \
2393 vmov.u8 d128_4, #4; \
2394 vmov.u8 d64_128, #128; \
2396 vmov.u8 d128_0x7, #0x7; \
2397 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2398 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
2401 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2402 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2404 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2405 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
2407 cmp span_num_blocks, #0; \
2410 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2411 add fb_ptr, fb_ptr, y, lsl #11; \
2413 ldr b, [ span_b_offset ]; \
2414 vdup.u32 v_left_x, left_x; \
2417 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2418 add fb_ptr, fb_ptr, left_x, lsl #1; \
2420 mla b, b_dx, left_x, b; \
2421 and dither_shift, left_x, #0x03; \
2423 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2424 vshr.u32 rg_dx, rg_dx4, #2; \
2426 mov dither_shift, dither_shift, lsl #3; \
2427 vmla.u32 rg, rg_dx, v_left_x; \
2429 subs span_num_blocks, span_num_blocks, #1; \
2431 mov dither_row, dither_row, ror dither_shift; \
2432 mov b_dx4, b_dx, lsl #2; \
2434 vdup.u32 dither_offsets, dither_row; \
2435 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2437 vdup.u32 b_block, b; \
2438 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2440 mov b_dx8, b_dx, lsl #3; \
2441 vdup.u32 r_block, rg[0]; \
2442 vdup.u32 g_block, rg[1]; \
2444 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2446 vadd.u32 r_block, r_block, block_span; \
2447 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2449 vadd.u32 g_block, g_block, block_span; \
2450 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2452 vadd.u32 b_block, b_block, block_span; \
2453 add block_ptr_b, block_ptr_a, #16; \
2455 vshrn.u32 r_whole_low, r_block, #16; \
2456 vshrn.u32 g_whole_low, g_block, #16; \
2457 vshrn.u32 b_whole_low, b_block, #16; \
2458 vdup.u32 dx4, rg_dx4[0]; \
2460 vaddhn.u32 r_whole_high, r_block, dx4; \
2461 vdup.u32 dx4, rg_dx4[1]; \
2463 vaddhn.u32 g_whole_high, g_block, dx4; \
2464 vdup.u32 dx4, b_dx4; \
2466 vaddhn.u32 b_whole_high, b_block, dx4; \
2467 vdup.u32 dx8, rg_dx8[0]; \
2469 vadd.u32 r_block, r_block, dx8; \
2470 vdup.u32 dx8, rg_dx8[1]; \
2472 vadd.u32 g_block, g_block, dx8; \
2473 vdup.u32 dx8, b_dx8; \
2475 vadd.u32 b_block, b_block, dx8; \
2477 vmovn.u16 r_whole_8, r_whole; \
2478 vmovn.u16 g_whole_8, g_whole; \
2479 vmovn.u16 b_whole_8, b_whole; \
2484 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2485 vshrn.u32 r_whole_low, r_block, #16; \
2487 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2488 vshrn.u32 g_whole_low, g_block, #16; \
2490 vshrn.u32 b_whole_low, b_block, #16; \
2492 vdup.u32 dx4, rg_dx4[0]; \
2493 vshr.u8 r_whole_8, r_whole_8, #3; \
2494 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2496 vaddhn.u32 r_whole_high, r_block, dx4; \
2497 vdup.u32 dx4, rg_dx4[1]; \
2499 vmov pixels, msb_mask; \
2500 vaddhn.u32 g_whole_high, g_block, dx4; \
2501 vdup.u32 dx4, b_dx4; \
2503 vaddhn.u32 b_whole_high, b_block, dx4; \
2504 vdup.u32 dx8, rg_dx8[0]; \
2506 vmlal.u8 pixels, r_whole_8, d64_1; \
2507 vmlal.u8 pixels, g_whole_8, d64_4; \
2508 vmlal.u8 pixels, b_whole_8, d64_128; \
2510 vadd.u32 r_block, r_block, dx8; \
2511 vdup.u32 dx8, rg_dx8[1]; \
2513 vadd.u32 g_block, g_block, dx8; \
2514 vdup.u32 dx8, b_dx8; \
2516 vadd.u32 b_block, b_block, dx8; \
2518 vmovn.u16 r_whole_8, r_whole; \
2519 vmovn.u16 g_whole_8, g_whole; \
2520 vmovn.u16 b_whole_8, b_whole; \
2522 vst1.u32 { pixels }, [ fb_ptr ]!; \
2523 subs span_num_blocks, span_num_blocks, #1; \
2527 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2529 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2530 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2532 vshr.u8 r_whole_8, r_whole_8, #3; \
2533 rbit right_mask, right_mask; \
2534 vmov pixels, msb_mask; \
2535 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2536 clz right_mask, right_mask; \
2538 vmlal.u8 pixels, r_whole_8, d64_1; \
2539 vmlal.u8 pixels, g_whole_8, d64_4; \
2540 vmlal.u8 pixels, b_whole_8, d64_128; \
2542 ldr pc, [ pc, right_mask, lsl #2 ]; \
2555 vst1.u16 { pixels_low[0] }, [ fb_ptr ]; \
2559 vst1.u32 { pixels_low[0] }, [ fb_ptr ]; \
2563 vst1.u32 { pixels_low[0] }, [ fb_ptr ]!; \
2564 vst1.u16 { pixels_low[2] }, [ fb_ptr ]; \
2568 vst1.u32 { pixels_low }, [ fb_ptr ]; \
2572 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2573 vst1.u16 { pixels_high[0] }, [ fb_ptr ]; \
2577 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2578 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2582 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2583 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2584 vst1.u16 { pixels_high[2] }, [ fb_ptr ]; \
2588 vst1.u32 { pixels }, [ fb_ptr ]; \
2592 add span_uvrg_offset, span_uvrg_offset, #16; \
2593 add span_b_offset, span_b_offset, #4; \
2595 add span_edge_data, span_edge_data, #8; \
2596 subs num_spans, num_spans, #1; \
2600 ldmia sp!, { r4 - r11, pc } \
2602 setup_blocks_shaded_untextured_direct_builder(undithered)
2603 setup_blocks_shaded_untextured_direct_builder(dithered)
2612 #define block_ptr r1
2613 #define num_blocks r2
2626 #define texture_ptr r11
2640 #define pixels_d r10
2644 #define clut_ptr r12
2645 #define current_texture_mask r5
2646 #define dirty_textures_mask r6
2650 #define clut_low_a d2
2651 #define clut_low_b d3
2652 #define clut_high_a d4
2653 #define clut_high_b d5
2658 #define texels_low d6
2659 #define texels_high d7
2663 function(texture_blocks_untextured)
2669 function(texture_blocks_4bpp)
2670 stmdb sp!, { r3 - r11, r14 }
2671 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2673 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2674 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2676 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2677 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]
2679 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2680 vuzp.u8 clut_a, clut_b
2682 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
2683 tst dirty_textures_mask, current_texture_mask
2689 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2691 uxtah uv_0, texture_ptr, uv_01
2692 uxtah uv_1, texture_ptr, uv_01, ror #16
2694 uxtah uv_2, texture_ptr, uv_23
2695 uxtah uv_3, texture_ptr, uv_23, ror #16
2697 uxtah uv_4, texture_ptr, uv_45
2698 ldrb pixel_0, [ uv_0 ]
2700 uxtah uv_5, texture_ptr, uv_45, ror #16
2701 ldrb pixel_1, [ uv_1 ]
2703 uxtah uv_6, texture_ptr, uv_67
2704 ldrb pixel_2, [ uv_2 ]
2706 uxtah uv_7, texture_ptr, uv_67, ror #16
2707 ldrb pixel_3, [ uv_3 ]
2709 ldrb pixel_4, [ uv_4 ]
2710 subs num_blocks, num_blocks, #1
2712 ldrb pixel_5, [ uv_5 ]
2713 orr pixels_a, pixel_0, pixel_1, lsl #8
2715 ldrb pixel_6, [ uv_6 ]
2716 orr pixels_b, pixel_4, pixel_5, lsl #8
2718 ldrb pixel_7, [ uv_7 ]
2719 orr pixels_a, pixels_a, pixel_2, lsl #16
2721 orr pixels_b, pixels_b, pixel_6, lsl #16
2722 orr pixels_a, pixels_a, pixel_3, lsl #24
2724 orr pixels_b, pixels_b, pixel_7, lsl #24
2725 vmov.u32 texels, pixels_a, pixels_b
2727 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2728 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2730 vst2.u8 { texels_low, texels_high }, [ block_ptr, :128 ], c_64
2733 ldmia sp!, { r3 - r11, pc }
2736 stmdb sp!, { r1 - r2 }
2737 bl update_texture_4bpp_cache
2740 ldmia sp!, { r1 - r2 }
2746 function(texture_blocks_8bpp)
2747 stmdb sp!, { r3 - r11, r14 }
2748 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2750 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2751 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2753 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2754 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2756 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]
2757 tst dirty_textures_mask, current_texture_mask
2763 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2765 uxtah uv_0, texture_ptr, uv_01
2766 uxtah uv_1, texture_ptr, uv_01, ror #16
2768 uxtah uv_2, texture_ptr, uv_23
2769 uxtah uv_3, texture_ptr, uv_23, ror #16
2771 uxtah uv_4, texture_ptr, uv_45
2772 ldrb pixel_0, [ uv_0 ]
2774 uxtah uv_5, texture_ptr, uv_45, ror #16
2775 ldrb pixel_1, [ uv_1 ]
2777 uxtah uv_6, texture_ptr, uv_67
2778 ldrb pixel_2, [ uv_2 ]
2780 uxtah uv_7, texture_ptr, uv_67, ror #16
2781 ldrb pixel_3, [ uv_3 ]
2783 ldrb pixel_4, [ uv_4 ]
2784 add pixel_0, pixel_0, pixel_0
2786 ldrb pixel_5, [ uv_5 ]
2787 add pixel_1, pixel_1, pixel_1
2789 ldrb pixel_6, [ uv_6 ]
2790 add pixel_2, pixel_2, pixel_2
2792 ldrb pixel_7, [ uv_7 ]
2793 add pixel_3, pixel_3, pixel_3
2795 ldrh pixel_0, [ clut_ptr, pixel_0 ]
2796 add pixel_4, pixel_4, pixel_4
2798 ldrh pixel_1, [ clut_ptr, pixel_1 ]
2799 add pixel_5, pixel_5, pixel_5
2801 ldrh pixel_2, [ clut_ptr, pixel_2 ]
2802 add pixel_6, pixel_6, pixel_6
2804 ldrh pixel_3, [ clut_ptr, pixel_3 ]
2805 add pixel_7, pixel_7, pixel_7
2807 ldrh pixel_4, [ clut_ptr, pixel_4 ]
2808 orr pixels_a, pixel_0, pixel_1, lsl #16
2810 ldrh pixel_5, [ clut_ptr, pixel_5 ]
2811 orr pixels_c, pixel_2, pixel_3, lsl #16
2813 ldrh pixel_6, [ clut_ptr, pixel_6 ]
2814 subs num_blocks, num_blocks, #1
2816 ldrh pixel_7, [ clut_ptr, pixel_7 ]
2817 orr pixels_b, pixel_4, pixel_5, lsl #16
2819 orr pixels_d, pixel_6, pixel_7, lsl #16
2820 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2822 add block_ptr, block_ptr, #64
2825 ldmia sp!, { r3 - r11, pc }
2828 stmdb sp!, { r1 - r2, r12 }
2830 bl update_texture_8bpp_cache
2832 ldmia sp!, { r1 - r2, r12 }
2862 #define block_ptr r1
2863 #define num_blocks r2
2907 #define texture_ptr r12
2912 function(texture_blocks_16bpp)
2913 stmdb sp!, { r3 - r11, r14 }
2914 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2916 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2917 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2920 ldrh uv_0, [ block_ptr ]
2921 subs num_blocks, num_blocks, #1
2923 ldrh uv_1, [ block_ptr, #2 ]
2925 and v_0, uv_0, #0xFF00
2926 and v_1, uv_1, #0xFF00
2928 and u_0, uv_0, #0xFF
2929 and u_1, uv_1, #0xFF
2931 add uv_0, u_0, v_0, lsl #2
2932 ldrh uv_2, [ block_ptr, #4 ]
2934 add uv_1, u_1, v_1, lsl #2
2935 ldrh uv_3, [ block_ptr, #6 ]
2937 add uv_0, uv_0, uv_0
2938 add uv_1, uv_1, uv_1
2940 and v_2, uv_2, #0xFF00
2941 and v_3, uv_3, #0xFF00
2943 and u_2, uv_2, #0xFF
2944 and u_3, uv_3, #0xFF
2946 add uv_2, u_2, v_2, lsl #2
2947 ldrh uv_4, [ block_ptr, #8 ]
2949 add uv_3, u_3, v_3, lsl #2
2950 ldrh uv_5, [ block_ptr, #10 ]
2952 add uv_2, uv_2, uv_2
2953 add uv_3, uv_3, uv_3
2955 and v_4, uv_4, #0xFF00
2956 and v_5, uv_5, #0xFF00
2958 and u_4, uv_4, #0xFF
2959 and u_5, uv_5, #0xFF
2961 add uv_4, u_4, v_4, lsl #2
2962 ldrh uv_6, [ block_ptr, #12 ]
2964 add uv_5, u_5, v_5, lsl #2
2965 ldrh uv_7, [ block_ptr, #14 ]
2967 add uv_4, uv_4, uv_4
2968 ldrh pixel_0, [ texture_ptr, uv_0 ]
2970 add uv_5, uv_5, uv_5
2971 ldrh pixel_1, [ texture_ptr, uv_1 ]
2973 and v_6, uv_6, #0xFF00
2974 ldrh pixel_2, [ texture_ptr, uv_2 ]
2976 and v_7, uv_7, #0xFF00
2977 ldrh pixel_3, [ texture_ptr, uv_3 ]
2979 and u_6, uv_6, #0xFF
2980 ldrh pixel_4, [ texture_ptr, uv_4 ]
2982 and u_7, uv_7, #0xFF
2983 ldrh pixel_5, [ texture_ptr, uv_5 ]
2985 add uv_6, u_6, v_6, lsl #2
2986 add uv_7, u_7, v_7, lsl #2
2988 add uv_6, uv_6, uv_6
2989 add uv_7, uv_7, uv_7
2991 orr pixels_a, pixel_0, pixel_1, lsl #16
2992 orr pixels_b, pixel_2, pixel_3, lsl #16
2994 ldrh pixel_6, [ texture_ptr, uv_6 ]
2995 orr pixels_c, pixel_4, pixel_5, lsl #16
2997 ldrh pixel_7, [ texture_ptr, uv_7 ]
2998 orr pixels_d, pixel_6, pixel_7, lsl #16
3000 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
3001 add block_ptr, block_ptr, #64
3005 ldmia sp!, { r3 - r11, pc }
3020 #undef msb_mask_high
3028 #define num_blocks r1
3029 #define color_ptr r2
3030 #define colors_scalar r2
3031 #define colors_scalar_compare r3
3032 #define mask_msb_ptr r2
3034 #define block_ptr_load_a r0
3035 #define block_ptr_store r3
3036 #define block_ptr_load_b r12
3041 #define draw_mask_bits_scalar r5
3043 #define d128_0x07 q0
3044 #define d128_0x1F q1
3045 #define d128_0x8000 q2
3046 #define test_mask q3
3048 #define colors_rg q5
3049 #define colors_b_dm_bits q6
3050 #define texels_rg q7
3053 #define pixels_b q10
3055 #define zero_mask q4
3056 #define draw_mask q12
3057 #define msb_mask q13
3059 #define fb_pixels q8
3061 #define pixels_gb_low q9
3063 #define colors_r d10
3064 #define colors_g d11
3065 #define colors_b d12
3066 #define draw_mask_bits d13
3067 #define texels_r d14
3068 #define texels_g d15
3069 #define pixels_r_low d16
3070 #define pixels_g_low d18
3071 #define pixels_b_low d19
3072 #define msb_mask_low d26
3073 #define msb_mask_high d27
3078 #define texels_b d31
3080 #define shade_blocks_textured_modulated_prologue_indirect() \
3082 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3084 #define shade_blocks_textured_modulated_prologue_direct() \
3085 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3086 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ] \
3089 #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \
3091 #define shade_blocks_textured_false_modulation_check_undithered(target) \
3092 ldr colors_scalar, [ psx_gpu, #psx_gpu_triangle_color_offset ]; \
3093 movw colors_scalar_compare, #0x8080; \
3095 movt colors_scalar_compare, #0x80; \
3096 cmp colors_scalar, colors_scalar_compare; \
3097 beq shade_blocks_textured_unmodulated_##target \
3099 #define shade_blocks_textured_false_modulation_check_dithered(target) \
3101 #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \
3102 shade_blocks_textured_false_modulation_check_##dithering(target); \
3103 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3104 vld1.u32 { colors_r[] }, [ color_ptr, :32 ]; \
3105 vdup.u8 colors_g, colors_r[1]; \
3106 vdup.u8 colors_b, colors_r[2]; \
3107 vdup.u8 colors_r, colors_r[0] \
3110 #define shade_blocks_textured_modulated_load_dithered(target) \
3111 vld1.u32 { target }, [ block_ptr_load_b, :128 ] \
3113 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3114 vld1.u32 { target }, [ block_ptr_load_b, :128 ], c_32 \
3116 #define shade_blocks_textured_modulated_load_undithered(target) \
3118 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3119 add block_ptr_load_b, block_ptr_load_b, #32 \
3121 #define shade_blocks_textured_modulate_dithered(channel) \
3122 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3124 #define shade_blocks_textured_modulate_undithered(channel) \
3125 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3128 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3129 vst1.u32 { draw_mask }, [ block_ptr_store, :128 ]! \
3131 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3132 ldr fb_ptr, [ block_ptr_load_b, #(offset - 64) ]; \
3133 vld1.u32 { fb_pixels }, [ fb_ptr ]; \
3134 vbit.u16 pixels, fb_pixels, draw_mask \
3136 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3137 vst1.u32 { pixels }, [ block_ptr_store, :128 ], c_48 \
3139 #define shade_blocks_textured_modulated_store_pixels_direct() \
3140 vst1.u32 { pixels }, [ fb_ptr ] \
3143 #define shade_blocks_textured_modulated_load_rg_shaded() \
3144 vld1.u32 { colors_r, colors_g }, [ block_ptr_load_b, :128 ], c_32 \
3146 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3147 add block_ptr_load_b, block_ptr_load_b, #32 \
3149 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3150 vld1.u32 { colors_b, draw_mask_bits }, [ block_ptr_load_a, :128 ], c_32 \
3152 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3153 ldr draw_mask_bits_scalar, [ block_ptr_load_a, #8 ]; \
3154 add block_ptr_load_a, block_ptr_load_a, #32 \
3156 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3157 vdup.u16 draw_mask, draw_mask_bits[0] \
3159 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3160 vdup.u16 draw_mask, draw_mask_bits_scalar \
3163 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3165 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3166 vorr.u16 pixels, pixels, msb_mask \
3169 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3172 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3173 shade_blocks_textured_modulated_prologue_##shading(dithering, target); \
3174 stmdb sp!, { r4 - r5, lr }; \
3175 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3177 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
3179 shade_blocks_textured_modulated_prologue_##target(); \
3181 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3184 add block_ptr_load_b, block_ptr_load_a, #16; \
3185 vmov.u8 d64_1, #1; \
3186 vmov.u8 d64_4, #4; \
3187 vmov.u8 d64_128, #128; \
3189 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3190 vmov.u8 d128_0x07, #0x07; \
3192 shade_blocks_textured_modulated_load_rg_##shading(); \
3193 vmov.u8 d128_0x1F, #0x1F; \
3195 shade_blocks_textured_modulated_load_bdm_##shading(); \
3196 vmov.u16 d128_0x8000, #0x8000; \
3198 vmovn.u16 texels_r, texels; \
3199 vshrn.u16 texels_g, texels, #5; \
3201 vshrn.u16 texels_b, texels, #7; \
3202 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3204 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3205 vtst.u16 draw_mask, draw_mask, test_mask; \
3207 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3208 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3210 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3211 vshr.u8 texels_b, texels_b, #3; \
3213 shade_blocks_textured_modulate_##dithering(r); \
3214 shade_blocks_textured_modulate_##dithering(g); \
3215 shade_blocks_textured_modulate_##dithering(b); \
3217 vand.u16 pixels, texels, d128_0x8000; \
3218 vceq.u16 zero_mask, texels, #0; \
3220 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3221 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3222 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3224 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3225 vorr.u16 draw_mask, draw_mask, zero_mask; \
3226 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3227 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3229 subs num_blocks, num_blocks, #1; \
3235 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3236 shade_blocks_textured_modulated_load_rg_##shading(); \
3237 vshrn.u16 texels_g, texels, #5; \
3239 shade_blocks_textured_modulated_load_bdm_##shading(); \
3240 vshrn.u16 texels_b, texels, #7; \
3242 vmovn.u16 texels_r, texels; \
3243 vmlal.u8 pixels, pixels_r_low, d64_1; \
3245 vmlal.u8 pixels, pixels_g_low, d64_4; \
3246 vmlal.u8 pixels, pixels_b_low, d64_128; \
3247 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3249 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3250 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3252 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3253 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3255 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3256 vtst.u16 draw_mask, draw_mask, test_mask; \
3258 shade_blocks_textured_modulated_store_pixels_##target(); \
3259 vshr.u8 texels_b, texels_b, #3; \
3261 shade_blocks_textured_modulate_##dithering(r); \
3262 shade_blocks_textured_modulate_##dithering(g); \
3263 shade_blocks_textured_modulate_##dithering(b); \
3265 vand.u16 pixels, texels, d128_0x8000; \
3266 vceq.u16 zero_mask, texels, #0; \
3268 subs num_blocks, num_blocks, #1; \
3270 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3271 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3272 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3274 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3275 vorr.u16 draw_mask, draw_mask, zero_mask; \
3276 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3277 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3282 vmlal.u8 pixels, pixels_r_low, d64_1; \
3283 vmlal.u8 pixels, pixels_g_low, d64_4; \
3284 vmlal.u8 pixels, pixels_b_low, d64_128; \
3286 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3287 shade_blocks_textured_modulated_store_pixels_##target(); \
3289 ldmia sp!, { r4 - r5, pc } \
3292 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3293 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3294 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3295 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3297 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3298 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3299 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3300 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3318 #undef msb_mask_high
3321 #define num_blocks r1
3322 #define mask_msb_ptr r2
3323 #define color_ptr r3
3325 #define block_ptr_load r0
3326 #define draw_mask_store_ptr r3
3327 #define draw_mask_bits_ptr r12
3328 #define draw_mask_ptr r12
3329 #define pixel_store_ptr r14
3331 #define fb_ptr_cmp r4
3334 #define fb_ptr_next r14
3338 #define test_mask q0
3340 #define draw_mask q2
3341 #define zero_mask q3
3342 #define draw_mask_combined q4
3343 #define fb_pixels q5
3344 #define fb_pixels_next q6
3347 #define draw_mask_low d4
3348 #define draw_mask_high d5
3349 #define msb_mask_low d14
3350 #define msb_mask_high d15
3353 function(shade_blocks_textured_unmodulated_indirect)
3354 str r14, [ sp, #-4 ]
3355 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3357 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3358 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3360 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3361 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3364 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3366 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3367 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3368 [ draw_mask_bits_ptr, :16 ], c_64
3369 vceq.u16 zero_mask, pixels, #0
3371 vtst.u16 draw_mask, draw_mask, test_mask
3372 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3374 subs num_blocks, num_blocks, #1
3378 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3379 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3381 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3382 [ draw_mask_bits_ptr, :16 ], c_64
3383 vceq.u16 zero_mask, pixels, #0
3385 vtst.u16 draw_mask, draw_mask, test_mask
3386 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3388 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3389 subs num_blocks, num_blocks, #1
3394 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3395 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3402 function(shade_blocks_textured_unmodulated_direct)
3403 stmdb sp!, { r4, r14 }
3404 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3406 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3407 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3409 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3412 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3413 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3415 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3416 [ draw_mask_bits_ptr, :16 ], c_64
3417 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3419 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3420 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3421 vceq.u16 zero_mask, pixels, #0
3422 vtst.u16 draw_mask, draw_mask, test_mask
3424 subs num_blocks, num_blocks, #1
3428 mov fb_ptr, fb_ptr_next
3429 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3431 vorr.u16 pixels, pixels, msb_mask
3433 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3434 vmov fb_pixels, fb_pixels_next
3436 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3437 [ draw_mask_bits_ptr, :16 ], c_64
3438 vbif.u16 fb_pixels, pixels, draw_mask_combined
3440 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3442 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3443 add fb_ptr_cmp, fb_ptr_cmp, #14
3447 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3448 vceq.u16 zero_mask, pixels, #0
3450 vst1.u16 { fb_pixels }, [ fb_ptr ]
3451 vtst.u16 draw_mask, draw_mask, test_mask
3454 subs num_blocks, num_blocks, #1
3458 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3459 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3461 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3463 ldmia sp!, { r4, pc }
3466 vst1.u16 { fb_pixels }, [ fb_ptr ]
3467 vceq.u16 zero_mask, pixels, #0
3469 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3470 vtst.u16 draw_mask, draw_mask, test_mask
3475 function(shade_blocks_unshaded_untextured_indirect)
3480 function(shade_blocks_unshaded_untextured_direct)
3481 stmdb sp!, { r4, r14 }
3482 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3484 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3485 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3487 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3488 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3490 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3491 vld1.u16 { pixels }, [ color_ptr, :128 ]
3494 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3496 vorr.u16 pixels, pixels, msb_mask
3497 subs num_blocks, num_blocks, #1
3499 ldr fb_ptr_next, [ block_ptr_load ], #64
3501 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3505 vmov fb_pixels, fb_pixels_next
3506 mov fb_ptr, fb_ptr_next
3507 ldr fb_ptr_next, [ block_ptr_load ], #64
3509 vbif.u16 fb_pixels, pixels, draw_mask
3510 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3512 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3513 add fb_ptr_cmp, fb_ptr_cmp, #14
3517 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3518 vst1.u16 { fb_pixels }, [ fb_ptr ]
3521 subs num_blocks, num_blocks, #1
3525 vbif.u16 fb_pixels_next, pixels, draw_mask
3526 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3528 ldmia sp!, { r4, pc }
3531 vst1.u16 { fb_pixels }, [ fb_ptr ]
3532 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3536 #undef draw_mask_ptr
3543 #define num_blocks r1
3544 #define msb_mask_ptr r2
3545 #define pixel_ptr r3
3546 #define draw_mask_ptr r0
3549 #define fb_ptr_next r14
3550 #define fb_ptr_cmp r4
3558 #undef msb_mask_high
3559 #undef draw_mask_next
3562 #undef fb_pixels_next
3565 #define draw_mask q1
3567 #define fb_pixels q3
3568 #define blend_pixels q4
3569 #define pixels_no_msb q5
3570 #define blend_mask q6
3571 #define fb_pixels_no_msb q7
3572 #define d128_0x8000 q8
3573 #define d128_0x0421 q9
3574 #define fb_pixels_next q10
3575 #define blend_pixels_next q11
3576 #define pixels_next q12
3577 #define draw_mask_next q13
3578 #define write_mask q14
3580 #define pixels_rb q5
3581 #define pixels_mg q7
3583 #define d128_0x7C1F q8
3584 #define d128_0x03E0 q9
3585 #define fb_pixels_rb q10
3586 #define fb_pixels_g q11
3587 #define fb_pixels_masked q11
3588 #define d128_0x83E0 q15
3589 #define pixels_fourth q7
3590 #define d128_0x1C07 q12
3591 #define d128_0x00E0 q13
3592 #define d128_0x80E0 q13
3594 #define msb_mask_low d0
3595 #define msb_mask_high d1
3597 #define blend_blocks_average_set_blend_mask_textured(source) \
3598 vclt.s16 blend_mask, source, #0 \
3600 #define blend_blocks_average_set_stp_bit_textured() \
3601 vorr.u16 blend_pixels, #0x8000 \
3603 #define blend_blocks_average_combine_textured(source) \
3604 vbif.u16 blend_pixels, source, blend_mask \
3606 #define blend_blocks_average_set_blend_mask_untextured(source) \
3608 #define blend_blocks_average_set_stp_bit_untextured() \
3610 #define blend_blocks_average_combine_untextured(source) \
3612 #define blend_blocks_average_mask_set_on() \
3613 vclt.s16 write_mask, fb_pixels_next, #0 \
3615 #define blend_blocks_average_mask_copy_on() \
3616 vorr.u16 draw_mask, draw_mask_next, write_mask \
3618 #define blend_blocks_average_mask_copy_b_on() \
3619 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3621 #define blend_blocks_average_mask_set_off() \
3623 #define blend_blocks_average_mask_copy_off() \
3624 vmov draw_mask, draw_mask_next \
3626 #define blend_blocks_average_mask_copy_b_off() \
3628 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3631 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3632 stmdb sp!, { r4, r14 }; \
3633 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3634 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3636 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3637 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3639 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3642 vmov.u16 d128_0x8000, #0x8000; \
3643 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3644 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3646 vmov.u16 d128_0x0421, #0x0400; \
3647 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3649 vorr.u16 d128_0x0421, #0x0021; \
3650 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3652 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3653 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3654 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3655 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3656 blend_blocks_average_mask_set_##mask_evaluate(); \
3657 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3659 subs num_blocks, num_blocks, #1; \
3663 mov fb_ptr, fb_ptr_next; \
3664 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3666 vmov pixels, pixels_next; \
3667 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3669 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3671 blend_blocks_average_mask_copy_##mask_evaluate(); \
3672 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3674 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3675 blend_blocks_average_set_stp_bit_##texturing(); \
3676 vmov fb_pixels, fb_pixels_next; \
3677 blend_blocks_average_combine_##texturing(pixels); \
3679 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3680 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3681 cmp fb_ptr_cmp, #28; \
3684 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3685 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3687 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3688 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3690 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3691 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3693 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3694 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3695 blend_blocks_average_mask_set_##mask_evaluate(); \
3696 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3699 subs num_blocks, num_blocks, #1; \
3703 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3704 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3706 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3707 blend_blocks_average_set_stp_bit_##texturing(); \
3708 blend_blocks_average_combine_##texturing(pixels_next); \
3710 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3711 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3712 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3714 ldmia sp!, { r4, pc }; \
3717 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3718 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3719 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3721 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3722 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3723 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3724 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3725 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3726 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3730 blend_blocks_average_builder(textured, off)
3731 blend_blocks_average_builder(untextured, off)
3732 blend_blocks_average_builder(textured, on)
3733 blend_blocks_average_builder(untextured, on)
3736 #define blend_blocks_add_mask_set_on() \
3737 vclt.s16 write_mask, fb_pixels, #0 \
3739 #define blend_blocks_add_mask_copy_on() \
3740 vorr.u16 draw_mask, draw_mask, write_mask \
3742 #define blend_blocks_add_mask_set_off() \
3744 #define blend_blocks_add_mask_copy_off() \
3747 #define blend_blocks_add_textured_builder(mask_evaluate) \
3750 function(blend_blocks_textured_add_##mask_evaluate) \
3751 stmdb sp!, { r4, r14 }; \
3752 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3753 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3755 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3756 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3758 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3761 vmov.u16 d128_0x7C1F, #0x7C00; \
3762 vmov.u16 d128_0x03E0, #0x0300; \
3763 vmov.u16 d128_0x83E0, #0x8000; \
3764 vorr.u16 d128_0x03E0, #0x00E0; \
3765 vorr.u16 d128_0x7C1F, #0x001F; \
3766 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3768 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3769 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3770 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3771 vclt.s16 blend_mask, pixels, #0; \
3772 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3773 blend_blocks_add_mask_set_##mask_evaluate(); \
3774 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3776 blend_blocks_add_mask_copy_##mask_evaluate(); \
3777 vorr.u16 pixels, pixels, msb_mask; \
3778 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3779 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3780 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3781 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3782 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3783 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3784 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3785 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3787 subs num_blocks, num_blocks, #1; \
3791 mov fb_ptr, fb_ptr_next; \
3793 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3795 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3796 vclt.s16 blend_mask, pixels, #0; \
3798 vorr.u16 pixels, pixels, msb_mask; \
3799 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3800 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3802 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3803 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3805 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3806 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3807 cmp fb_ptr_cmp, #28; \
3810 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3811 blend_blocks_add_mask_set_##mask_evaluate(); \
3812 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3813 blend_blocks_add_mask_copy_##mask_evaluate(); \
3814 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3815 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3816 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3819 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3820 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3821 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3822 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3823 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3825 subs num_blocks, num_blocks, #1; \
3829 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3830 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3831 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3833 ldmia sp!, { r4, pc }; \
3836 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3837 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3839 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3840 blend_blocks_add_mask_set_##mask_evaluate(); \
3841 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3842 blend_blocks_add_mask_copy_##mask_evaluate(); \
3843 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3847 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3850 function(blend_blocks_untextured_add_##mask_evaluate) \
3851 stmdb sp!, { r4, r14 }; \
3852 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3853 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3855 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3856 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3858 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3861 vmov.u16 d128_0x7C1F, #0x7C00; \
3862 vmov.u16 d128_0x03E0, #0x0300; \
3863 vorr.u16 d128_0x7C1F, #0x001F; \
3864 vorr.u16 d128_0x03E0, #0x00E0; \
3866 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3867 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3868 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3869 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3870 blend_blocks_add_mask_set_##mask_evaluate(); \
3871 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3873 blend_blocks_add_mask_copy_##mask_evaluate(); \
3874 vand.u16 pixels_g, pixels, d128_0x03E0; \
3875 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3876 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3877 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3878 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3879 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3880 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3882 subs num_blocks, num_blocks, #1; \
3886 mov fb_ptr, fb_ptr_next; \
3888 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3890 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3892 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3893 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3894 vand.u16 pixels_g, pixels, d128_0x03E0; \
3896 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3897 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3899 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3900 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3901 cmp fb_ptr_cmp, #28; \
3904 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3905 blend_blocks_add_mask_set_##mask_evaluate(); \
3906 blend_blocks_add_mask_copy_##mask_evaluate(); \
3907 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3908 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3909 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3912 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3913 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3914 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3915 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3916 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3918 subs num_blocks, num_blocks, #1; \
3922 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3923 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3924 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3925 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3927 ldmia sp!, { r4, pc }; \
3930 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3931 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3933 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3934 blend_blocks_add_mask_set_##mask_evaluate(); \
3935 blend_blocks_add_mask_copy_##mask_evaluate(); \
3936 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3940 blend_blocks_add_textured_builder(off)
3941 blend_blocks_add_textured_builder(on)
3942 blend_blocks_add_untextured_builder(off)
3943 blend_blocks_add_untextured_builder(on)
3945 #define blend_blocks_subtract_set_blend_mask_textured() \
3946 vclt.s16 blend_mask, pixels_next, #0 \
3948 #define blend_blocks_subtract_combine_textured() \
3949 vbif.u16 blend_pixels, pixels, blend_mask \
3951 #define blend_blocks_subtract_set_stb_textured() \
3952 vorr.u16 blend_pixels, #0x8000 \
3954 #define blend_blocks_subtract_msb_mask_textured() \
3955 vorr.u16 pixels, pixels_next, msb_mask \
3957 #define blend_blocks_subtract_set_blend_mask_untextured() \
3959 #define blend_blocks_subtract_combine_untextured() \
3961 #define blend_blocks_subtract_set_stb_untextured() \
3962 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3964 #define blend_blocks_subtract_msb_mask_untextured() \
3967 #define blend_blocks_subtract_mask_set_on() \
3968 vclt.s16 write_mask, fb_pixels, #0 \
3970 #define blend_blocks_subtract_mask_copy_on() \
3971 vorr.u16 draw_mask, draw_mask_next, write_mask \
3973 #define blend_blocks_subtract_mask_set_off() \
3975 #define blend_blocks_subtract_mask_copy_off() \
3976 vmov draw_mask, draw_mask_next \
3979 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
3982 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
3983 stmdb sp!, { r4, r14 }; \
3984 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3985 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3987 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3988 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3990 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3993 vmov.u16 d128_0x7C1F, #0x7C00; \
3994 vmov.u16 d128_0x03E0, #0x0300; \
3995 vorr.u16 d128_0x7C1F, #0x001F; \
3996 vorr.u16 d128_0x03E0, #0x00E0; \
3998 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3999 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4000 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
4001 blend_blocks_subtract_set_blend_mask_##texturing(); \
4002 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4003 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4004 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4006 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4007 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4008 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4009 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4010 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4012 subs num_blocks, num_blocks, #1; \
4016 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4017 mov fb_ptr, fb_ptr_next; \
4018 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4020 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
4021 blend_blocks_subtract_msb_mask_##texturing(); \
4023 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
4024 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4025 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4026 blend_blocks_subtract_set_stb_##texturing(); \
4027 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4028 blend_blocks_subtract_combine_##texturing(); \
4029 blend_blocks_subtract_set_blend_mask_##texturing(); \
4030 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4032 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4033 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4034 cmp fb_ptr_cmp, #28; \
4037 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4038 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4039 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4040 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4041 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4042 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4043 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4046 subs num_blocks, num_blocks, #1; \
4050 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4052 blend_blocks_subtract_msb_mask_##texturing(); \
4053 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4054 blend_blocks_subtract_set_stb_##texturing(); \
4055 blend_blocks_subtract_combine_##texturing(); \
4056 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4057 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4059 ldmia sp!, { r4, pc }; \
4062 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4063 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4064 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4065 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4066 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4067 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4068 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4072 blend_blocks_subtract_builder(textured, off)
4073 blend_blocks_subtract_builder(textured, on)
4074 blend_blocks_subtract_builder(untextured, off)
4075 blend_blocks_subtract_builder(untextured, on)
4078 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4081 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4082 stmdb sp!, { r4, r14 }; \
4083 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4084 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4086 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4087 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4089 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4092 vmov.u16 d128_0x7C1F, #0x7C00; \
4093 vmov.u16 d128_0x03E0, #0x0300; \
4094 vmov.u16 d128_0x83E0, #0x8300; \
4095 vmov.u16 d128_0x1C07, #0x1C00; \
4096 vmov.u16 d128_0x80E0, #0x8000; \
4097 vorr.u16 d128_0x7C1F, #0x001F; \
4098 vorr.u16 d128_0x03E0, #0x00E0; \
4099 vorr.u16 d128_0x83E0, #0x00E0; \
4100 vorr.u16 d128_0x1C07, #0x0007; \
4101 vorr.u16 d128_0x80E0, #0x00E0; \
4103 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4104 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4105 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4106 vclt.s16 blend_mask, pixels, #0; \
4107 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4108 blend_blocks_add_mask_set_##mask_evaluate(); \
4109 vshr.s16 pixels_fourth, pixels, #2; \
4111 blend_blocks_add_mask_copy_##mask_evaluate(); \
4112 vorr.u16 pixels, pixels, msb_mask; \
4113 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4114 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
4115 vand.u16 pixels_mg, pixels_fourth, d128_0x80E0; \
4116 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
4117 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
4118 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4119 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
4120 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4121 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
4123 subs num_blocks, num_blocks, #1; \
4127 mov fb_ptr, fb_ptr_next; \
4129 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4131 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4132 vclt.s16 blend_mask, pixels, #0; \
4134 vshr.s16 pixels_fourth, pixels, #2; \
4135 vorr.u16 pixels, pixels, msb_mask; \
4136 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4137 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4139 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4140 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4142 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4143 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4144 cmp fb_ptr_cmp, #28; \
4147 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4148 blend_blocks_add_mask_set_##mask_evaluate(); \
4149 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
4150 blend_blocks_add_mask_copy_##mask_evaluate(); \
4151 vand.u16 pixels_mg, pixels_fourth, d128_0x80E0; \
4152 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
4153 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4156 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
4157 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4158 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
4159 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4160 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
4162 subs num_blocks, num_blocks, #1; \
4166 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4167 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4168 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4170 ldmia sp!, { r4, pc }; \
4173 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4174 vand.u16 pixels_mg, pixels_fourth, d128_0x80E0; \
4176 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4177 blend_blocks_add_mask_set_##mask_evaluate(); \
4178 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
4179 blend_blocks_add_mask_copy_##mask_evaluate(); \
4180 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
4184 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4187 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4188 stmdb sp!, { r4, r14 }; \
4189 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4190 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4192 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4193 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4195 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4198 vmov.u16 d128_0x7C1F, #0x7C00; \
4199 vmov.u16 d128_0x03E0, #0x0300; \
4200 vmov.u16 d128_0x83E0, #0x8300; \
4201 vmov.u16 d128_0x1C07, #0x1C00; \
4202 vmov.u16 d128_0x00E0, #0x00E0; \
4203 vorr.u16 d128_0x7C1F, #0x001F; \
4204 vorr.u16 d128_0x03E0, #0x00E0; \
4205 vorr.u16 d128_0x83E0, #0x00E0; \
4206 vorr.u16 d128_0x1C07, #0x0007; \
4208 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4209 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4210 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4211 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4212 blend_blocks_add_mask_set_##mask_evaluate(); \
4213 vshr.s16 pixels_fourth, pixels, #2; \
4214 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4216 blend_blocks_add_mask_copy_##mask_evaluate(); \
4217 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4218 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4219 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4220 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4221 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4222 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4223 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4225 subs num_blocks, num_blocks, #1; \
4229 mov fb_ptr, fb_ptr_next; \
4231 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4233 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4235 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4236 vshr.s16 pixels_fourth, pixels, #2; \
4237 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4238 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4240 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4241 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4243 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4244 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4245 cmp fb_ptr_cmp, #28; \
4248 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4249 blend_blocks_add_mask_set_##mask_evaluate(); \
4250 blend_blocks_add_mask_copy_##mask_evaluate(); \
4251 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4252 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4253 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4256 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4257 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4258 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4259 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4260 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4262 subs num_blocks, num_blocks, #1; \
4266 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4267 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4268 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4269 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4271 ldmia sp!, { r4, pc }; \
4274 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4275 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4277 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4278 blend_blocks_add_mask_set_##mask_evaluate(); \
4279 blend_blocks_add_mask_copy_##mask_evaluate(); \
4280 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4284 blend_blocks_add_fourth_textured_builder(off)
4285 blend_blocks_add_fourth_textured_builder(on)
4286 blend_blocks_add_fourth_untextured_builder(off)
4287 blend_blocks_add_fourth_untextured_builder(on)
4289 // TODO: Optimize this more. Need a scene that actually uses it for
4294 function(blend_blocks_textured_unblended_on)
4295 stmdb sp!, { r4, r14 }
4296 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4297 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
4299 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4300 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
4302 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4305 ldr fb_ptr, [ pixel_ptr, #28 ]
4306 vld1.u16 { fb_pixels }, [ fb_ptr ]
4307 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4308 vclt.s16 write_mask, fb_pixels, #0
4309 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4311 subs num_blocks, num_blocks, #1
4315 vorr.u16 draw_mask, draw_mask, write_mask
4316 vbif.u16 fb_pixels, pixels, draw_mask
4317 vst1.u16 { fb_pixels }, [ fb_ptr ]
4319 ldr fb_ptr, [ pixel_ptr, #28 ]
4320 vld1.u16 { fb_pixels }, [ fb_ptr ]
4321 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4322 vclt.s16 write_mask, fb_pixels, #0
4323 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4325 subs num_blocks, num_blocks, #1
4329 vorr.u16 draw_mask, draw_mask, write_mask
4330 vbif.u16 fb_pixels, pixels, draw_mask
4331 vst1.u16 { fb_pixels }, [ fb_ptr ]
4333 ldmia sp!, { r4, pc }
4336 function(blend_blocks_textured_unblended_off)
4346 vld1.u32 { u_whole_8, v_whole_8 }, [ r1, :128 ], r3
4366 #define num_width r12
4376 function(render_block_fill_body)
4377 vdup.u16 colors_a, color
4380 vmov colors_b, colors_a
4381 sub pitch, pitch, width, lsl #1
4383 mov num_width, width
4386 vst1.u32 { colors_a, colors_b }, [ vram_ptr, :256 ]!
4388 subs num_width, num_width, #2
4391 add vram_ptr, vram_ptr, pitch
4392 mov num_width, width
4394 subs height, height, #1
4408 #undef dirty_textures_mask
4410 #undef current_texture_mask
4421 #define offset_u_right r10
4422 #define width_rounded r11
4423 #define height_rounded r12
4425 #define texture_offset_base r1
4426 #define tile_width r2
4427 #define tile_height r3
4428 #define num_blocks r4
4430 #define sub_tile_height r6
4432 #define texture_mask r8
4433 #define column_data r9
4434 #define texture_offset r10
4435 #define tiles_remaining r11
4436 #define fb_ptr_advance_column r12
4437 #define texture_block_ptr r14
4439 #define texture_page_ptr r3
4440 #define left_block_mask r4
4441 #define right_block_mask r5
4442 #define texture_mask_rev r10
4443 #define control_mask r11
4445 #define dirty_textures_mask r4
4447 #define current_texture_mask r6
4461 #define draw_masks_fb_ptrs q1
4463 #define draw_mask_fb_ptr_left d2
4464 #define draw_mask_fb_ptr_right d3
4466 #define clut_low_a d4
4467 #define clut_low_b d5
4468 #define clut_high_a d6
4469 #define clut_high_b d7
4471 #define block_masks d8
4472 #define block_masks_shifted d9
4477 #define texels_low d10
4478 #define texels_high d11
4481 setup_sprite_flush_blocks_single:
4484 stmdb sp!, { r0 - r3, r12, r14 }
4485 bl flush_render_block_buffer
4486 ldmia sp!, { r0 - r3, r12, r14 }
4490 add block, psx_gpu, #psx_gpu_blocks_offset
4492 mov num_blocks, sub_tile_height
4496 setup_sprite_flush_blocks_double:
4499 stmdb sp!, { r0 - r3, r12, r14 }
4500 bl flush_render_block_buffer
4501 ldmia sp!, { r0 - r3, r12, r14 }
4505 add block, psx_gpu, #psx_gpu_blocks_offset
4507 mov num_blocks, sub_tile_height, lsl #1
4511 setup_sprite_update_texture_4bpp_cache:
4512 stmdb sp!, { r0 - r3, r14 }
4513 bl update_texture_4bpp_cache
4514 ldmia sp!, { r0 - r3, pc }
4517 setup_sprite_update_texture_8bpp_cache:
4518 stmdb sp!, { r0 - r3, r14 }
4519 bl update_texture_8bpp_cache
4520 ldmia sp!, { r0 - r3, pc }
4523 #define setup_sprite_tiled_initialize_4bpp() \
4524 ldr dirty_textures_mask, \
4525 [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]; \
4526 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4528 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4529 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4531 tst current_texture_mask, dirty_textures_mask; \
4532 vuzp.u8 clut_a, clut_b; \
4534 blne setup_sprite_update_texture_4bpp_cache \
4536 #define setup_sprite_tiled_initialize_8bpp() \
4537 ldr dirty_textures_mask, \
4538 [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]; \
4539 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4541 tst current_texture_mask, dirty_textures_mask; \
4542 blne setup_sprite_update_texture_8bpp_cache \
4545 #define setup_sprite_tile_setup_block_no(side, offset, texture_mode) \
4547 #define setup_sprite_block_count_single() \
4550 #define setup_sprite_block_count_double() \
4551 sub_tile_height, lsl #1 \
4553 #define setup_sprite_tile_add_blocks(type) \
4554 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4555 cmp num_blocks, #MAX_BLOCKS; \
4557 blgt setup_sprite_flush_blocks_##type \
4560 #define setup_sprite_tile_full_4bpp(edge) \
4561 setup_sprite_tile_add_blocks(double); \
4564 and texture_block_ptr, texture_offset, texture_mask; \
4565 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4568 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4569 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4571 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4572 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4574 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4575 add texture_block_ptr, texture_offset, #8; \
4577 and texture_block_ptr, texture_block_ptr, texture_mask; \
4578 add block, block, #40; \
4580 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4581 add fb_ptr, fb_ptr, #16; \
4583 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4584 add block, block, #24; \
4586 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4587 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4590 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4591 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4593 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4594 add block, block, #40; \
4596 add texture_offset, texture_offset, #0x10; \
4597 add fb_ptr, fb_ptr, #(2048 - 16); \
4599 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4600 add block, block, #24; \
4602 subs sub_tile_height, sub_tile_height, #1; \
4605 add texture_offset, texture_offset, #0xF00; \
4606 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4609 #define setup_sprite_tile_half_4bpp(edge) \
4610 setup_sprite_tile_add_blocks(single); \
4613 and texture_block_ptr, texture_offset, texture_mask; \
4614 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4617 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4618 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4620 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4621 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4623 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4624 add block, block, #40; \
4626 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4627 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4629 add block, block, #24; \
4630 add texture_offset, texture_offset, #0x10; \
4632 add fb_ptr, fb_ptr, #2048; \
4633 subs sub_tile_height, sub_tile_height, #1; \
4637 add texture_offset, texture_offset, #0xF00; \
4638 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4641 #define setup_sprite_tile_full_8bpp(edge) \
4642 setup_sprite_tile_add_blocks(double); \
4643 add block, block, #16; \
4646 and texture_block_ptr, texture_offset, texture_mask; \
4647 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4650 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4651 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4653 add texture_block_ptr, texture_offset, #8; \
4654 vst1.u32 { texels }, [ block, :64 ]; \
4656 and texture_block_ptr, texture_block_ptr, texture_mask; \
4657 add block, block, #24; \
4659 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4661 add fb_ptr, fb_ptr, #16; \
4662 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4664 add block, block, #40; \
4665 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4668 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4669 vst1.u32 { texels }, [ block, :64 ]; \
4670 add block, block, #24; \
4672 add texture_offset, texture_offset, #0x10; \
4673 add fb_ptr, fb_ptr, #(2048 - 16); \
4675 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4676 add block, block, #40; \
4678 subs sub_tile_height, sub_tile_height, #1; \
4681 sub block, block, #16; \
4682 add texture_offset, texture_offset, #0xF00; \
4683 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4686 #define setup_sprite_tile_half_8bpp(edge) \
4687 setup_sprite_tile_add_blocks(single); \
4688 add block, block, #16; \
4691 and texture_block_ptr, texture_offset, texture_mask; \
4692 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4695 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4696 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4698 vst1.u32 { texels }, [ block, :64 ]; \
4699 add block, block, #24; \
4701 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4702 add block, block, #40; \
4704 add texture_offset, texture_offset, #0x10; \
4705 add fb_ptr, fb_ptr, #2048; \
4707 subs sub_tile_height, sub_tile_height, #1; \
4710 sub block, block, #16; \
4711 add texture_offset, texture_offset, #0xF00; \
4712 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4715 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4716 add texture_offset, texture_offset_base, #8; \
4717 add fb_ptr, fb_ptr, #16 \
4719 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4720 mov texture_offset, texture_offset_base \
4722 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4723 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4725 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4726 mov texture_offset, texture_offset_base \
4728 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4729 sub fb_ptr, fb_ptr, #16 \
4731 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4733 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4734 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4736 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4739 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode) \
4740 mov sub_tile_height, column_data; \
4741 setup_sprite_tile_column_edge_pre_adjust_##edge_mode(edge); \
4742 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4743 setup_sprite_tile_column_edge_post_adjust_##edge_mode(edge) \
4745 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode) \
4746 and sub_tile_height, column_data, #0xFF; \
4747 mov tiles_remaining, column_data, lsr #16; \
4748 setup_sprite_tile_column_edge_pre_adjust_##edge_mode(edge); \
4749 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4751 subs tiles_remaining, tiles_remaining, #1; \
4755 mov sub_tile_height, #16; \
4756 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4757 subs tiles_remaining, tiles_remaining, #1; \
4761 uxtb sub_tile_height, column_data, ror #8; \
4762 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4763 setup_sprite_tile_column_edge_post_adjust_##edge_mode(edge) \
4766 #define setup_sprite_column_data_single() \
4767 mov column_data, height; \
4768 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] \
4770 #define setup_sprite_column_data_multi() \
4771 and height_rounded, height_rounded, #0xF; \
4772 rsb column_data, offset_v, #16; \
4774 add height_rounded, height_rounded, #1; \
4775 sub tile_height, tile_height, #1; \
4777 orr column_data, column_data, tile_height, lsl #16; \
4778 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]; \
4780 orr column_data, column_data, height_rounded, lsl #8 \
4782 #define setup_sprite_tile_column_width_single(texture_mode, multi_height, \
4784 setup_sprite_##texture_mode##_single_##multi_height##_##edge_mode##_##edge: \
4785 setup_sprite_column_data_##multi_height(); \
4786 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4787 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4788 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4789 vdup.u8 draw_mask_fb_ptr_right, block_masks[1]; \
4791 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, \
4793 ldmia sp!, { r4 - r11, pc } \
4795 #define setup_sprite_tiled_advance_column() \
4796 add texture_offset_base, texture_offset_base, #0x100; \
4797 tst texture_offset_base, #0xF00; \
4798 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4800 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4802 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode: \
4803 setup_sprite_column_data_##multi_height(); \
4804 mov fb_ptr_advance_column, #32; \
4806 sub fb_ptr_advance_column, height, lsl #11; \
4807 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4809 vdup.u8 draw_mask_fb_ptr_right, block_masks[1]; \
4810 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm); \
4812 subs tile_width, tile_width, #2; \
4813 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4815 vmov.u8 draw_masks_fb_ptrs, #0; \
4819 setup_sprite_tiled_advance_column(); \
4820 setup_sprite_tile_column_height_##multi_height(full, none, tm); \
4821 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4822 subs tile_width, tile_width, #1; \
4826 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4827 vdup.u8 draw_mask_fb_ptr_right, block_masks[5]; \
4829 setup_sprite_tiled_advance_column(); \
4830 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm); \
4831 ldmia sp!, { r4 - r11, pc } \
4839 // [ sp + 4 ]: width
4840 // [ sp + 8 ]: height
4841 // [ sp + 12 ]: color (unused)
4843 #define setup_sprite_tiled_builder(texture_mode) \
4845 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full); \
4846 setup_sprite_tile_column_width_single(texture_mode, multi, full, none); \
4847 setup_sprite_tile_column_width_multi(texture_mode, single, full, full); \
4848 setup_sprite_tile_column_width_single(texture_mode, single, full, none); \
4849 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full); \
4850 setup_sprite_tile_column_width_single(texture_mode, multi, half, right); \
4851 setup_sprite_tile_column_width_multi(texture_mode, single, half, full); \
4852 setup_sprite_tile_column_width_single(texture_mode, single, half, right); \
4853 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half); \
4854 setup_sprite_tile_column_width_single(texture_mode, multi, half, left); \
4855 setup_sprite_tile_column_width_multi(texture_mode, single, full, half); \
4856 setup_sprite_tile_column_width_single(texture_mode, single, half, left); \
4857 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half); \
4858 setup_sprite_tile_column_width_multi(texture_mode, single, half, half); \
4862 function(setup_sprite_##texture_mode) \
4863 stmdb sp!, { r4 - r11, r14 }; \
4864 setup_sprite_tiled_initialize_##texture_mode(); \
4866 ldr v, [ sp, #36 ]; \
4867 and offset_u, u, #0xF; \
4869 ldr width, [ sp, #40 ]; \
4870 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
4872 ldr height, [ sp, #44 ]; \
4873 add fb_ptr, fb_ptr, y, lsl #11; \
4875 add fb_ptr, fb_ptr, x, lsl #1; \
4876 and offset_v, v, #0xF; \
4878 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4879 add width_rounded, offset_u, width; \
4881 add height_rounded, offset_v, height; \
4882 add width_rounded, width_rounded, #15; \
4884 add height_rounded, height_rounded, #15; \
4885 mov tile_width, width_rounded, lsr #4; \
4887 /* texture_offset_base = VH-VL-00-00 */\
4888 mov texture_offset_base, v, lsl #8; \
4889 and offset_u_right, width_rounded, #0xF; \
4891 /* texture_offset_base = VH-UH-UL-00 */\
4892 bfi texture_offset_base, u, #4, #8; \
4893 movw right_block_mask, #0xFFFE; \
4895 /* texture_offset_base = VH-UH-VL-00 */\
4896 bfi texture_offset_base, v, #4, #4; \
4897 movw left_block_mask, #0xFFFF; \
4899 mov tile_height, height_rounded, lsr #4; \
4900 mvn left_block_mask, left_block_mask, lsl offset_u; \
4902 /* texture_mask = HH-HL-WH-WL */\
4903 ldrh texture_mask, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]; \
4904 mov right_block_mask, right_block_mask, lsl offset_u_right; \
4906 /* texture_mask_rev = WH-WL-HH-HL */\
4907 rev16 texture_mask_rev, texture_mask; \
4908 vmov block_masks, left_block_mask, right_block_mask; \
4910 /* texture_mask = HH-HL-HL-WL */\
4911 bfi texture_mask, texture_mask_rev, #4, #4; \
4912 /* texture_mask_rev = 00-00-00-WH */\
4913 mov texture_mask_rev, texture_mask_rev, lsr #12; \
4915 /* texture_mask = HH-WH-HL-WL */\
4916 bfi texture_mask, texture_mask_rev, #8, #4; \
4917 and left_block_mask, left_block_mask, #0xFF; \
4919 mov control_mask, #0; \
4920 cmp left_block_mask, #0xFF; \
4922 uxtb right_block_mask, right_block_mask, ror #8; \
4923 orreq control_mask, control_mask, #0x4; \
4925 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4926 cmp right_block_mask, #0xFF; \
4928 orreq control_mask, control_mask, #0x8; \
4929 cmp tile_width, #1; \
4931 add block, psx_gpu, #psx_gpu_blocks_offset; \
4932 orreq control_mask, control_mask, #0x1; \
4934 cmp tile_height, #1; \
4935 add block, block, num_blocks, lsl #6; \
4937 orreq control_mask, control_mask, #0x2; \
4938 ldr pc, [ pc, control_mask, lsl #2 ]; \
4941 .word setup_sprite_##texture_mode##_multi_multi_full_full; \
4942 .word setup_sprite_##texture_mode##_single_multi_full_none; \
4943 .word setup_sprite_##texture_mode##_multi_single_full_full; \
4944 .word setup_sprite_##texture_mode##_single_single_full_none; \
4945 .word setup_sprite_##texture_mode##_multi_multi_half_full; \
4946 .word setup_sprite_##texture_mode##_single_multi_half_right; \
4947 .word setup_sprite_##texture_mode##_multi_single_half_full; \
4948 .word setup_sprite_##texture_mode##_single_single_half_right; \
4949 .word setup_sprite_##texture_mode##_multi_multi_full_half; \
4950 .word setup_sprite_##texture_mode##_single_multi_half_left; \
4951 .word setup_sprite_##texture_mode##_multi_single_full_half; \
4952 .word setup_sprite_##texture_mode##_single_single_half_left; \
4953 .word setup_sprite_##texture_mode##_multi_multi_half_half; \
4955 .word setup_sprite_##texture_mode##_multi_single_half_half \
4958 setup_sprite_tiled_builder(4bpp);
4959 setup_sprite_tiled_builder(8bpp);
4967 #define block_ptr r0
4968 #define num_blocks r1
4970 #define texel_shift_mask r3
4971 #define block_pixels_a r4
4972 #define block_pixels_b r5
4981 #define texels_01 r6
4982 #define texels_23 r7
4983 #define texels_45 r8
4984 #define texels_67 r9
4986 function(texture_sprite_blocks_8bpp)
4987 stmdb sp!, { r4 - r11, r14 }
4988 movw texel_shift_mask, #(0xFF << 1)
4990 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
4991 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
4993 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
4994 ldr block_pixels_a, [ block_ptr, #16 ]
4997 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
4998 ldr block_pixels_b, [ block_ptr, #20 ]
5000 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
5001 ldrh texel_0, [ clut_ptr, texel_0 ]
5003 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
5004 ldrh texel_1, [ clut_ptr, texel_1 ]
5006 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
5007 ldr block_pixels_a, [ block_ptr, #(64 + 16) ]
5009 ldrh texel_2, [ clut_ptr, texel_2 ]
5010 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
5012 ldrh texel_3, [ clut_ptr, texel_3 ]
5013 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
5015 ldrh texel_4, [ clut_ptr, texel_4 ]
5016 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
5018 ldrh texel_5, [ clut_ptr, texel_5 ]
5019 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
5021 ldrh texel_6, [ clut_ptr, texel_6 ]
5022 orr texels_01, texel_0, texel_1, lsl #16
5024 ldrh texel_7, [ clut_ptr, texel_7 ]
5025 orr texels_23, texel_2, texel_3, lsl #16
5027 orr texels_45, texel_4, texel_5, lsl #16
5028 str texels_01, [ block_ptr, #0 ]
5030 orr texels_67, texel_6, texel_7, lsl #16
5031 str texels_23, [ block_ptr, #4 ]
5033 subs num_blocks, num_blocks, #1
5034 str texels_45, [ block_ptr, #8 ]
5036 str texels_67, [ block_ptr, #12 ]
5037 add block_ptr, block_ptr, #64
5041 ldmia sp!, { r4 - r11, pc }
5044 #undef width_rounded
5047 #undef texture_offset
5056 #define left_offset r8
5057 #define width_rounded r9
5058 #define right_width r10
5059 #define block_width r11
5061 #define texture_offset_base r1
5062 #define texture_mask r2
5063 #define texture_page_ptr r3
5064 #define num_blocks r4
5067 #define texture_offset r8
5068 #define blocks_remaining r9
5069 #define fb_ptr_pitch r12
5070 #define texture_block_ptr r14
5072 #define texture_mask_width r2
5073 #define texture_mask_height r3
5074 #define left_mask_bits r4
5075 #define right_mask_bits r5
5079 #undef block_masks_shifted
5082 #define block_masks d0
5083 #define block_masks_shifted d1
5084 #define draw_mask_fb_ptr d2
5088 setup_sprites_16bpp_flush_single:
5091 stmdb sp!, { r0 - r3, r12, r14 }
5092 bl flush_render_block_buffer
5093 ldmia sp!, { r0 - r3, r12, r14 }
5097 add block, psx_gpu, #psx_gpu_blocks_offset
5102 setup_sprites_16bpp_flush_row:
5105 stmdb sp!, { r0 - r3, r12, r14 }
5106 bl flush_render_block_buffer
5107 ldmia sp!, { r0 - r3, r12, r14 }
5111 add block, psx_gpu, #psx_gpu_blocks_offset
5112 mov num_blocks, block_width
5116 function(setup_sprite_16bpp)
5117 stmdb sp!, { r4 - r11, r14 }
5118 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5121 add fb_ptr, fb_ptr, y, lsl #11
5123 ldr width, [ sp, #40 ]
5124 add fb_ptr, fb_ptr, x, lsl #1
5126 ldr height, [ sp, #44 ]
5127 and left_offset, u, #0x7
5129 add texture_offset_base, u, u
5130 add width_rounded, width, #7
5132 add texture_offset_base, v, lsl #11
5133 mov left_mask_bits, #0xFF
5135 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5136 add width_rounded, width_rounded, left_offset
5138 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5139 sub fb_ptr, fb_ptr, left_offset, lsl #1
5141 add texture_mask, texture_mask_width, texture_mask_width
5142 mov right_mask_bits, #0xFE
5144 and right_width, width_rounded, #0x7
5145 mvn left_mask_bits, left_mask_bits, lsl left_offset
5147 add texture_mask, texture_mask_height, lsl #11
5148 mov block_width, width_rounded, lsr #3
5150 mov right_mask_bits, right_mask_bits, lsl right_width
5151 movw fb_ptr_pitch, #(2048 + 16)
5153 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5154 vmov block_masks, left_mask_bits, right_mask_bits
5156 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5157 add block, psx_gpu, #psx_gpu_blocks_offset
5159 bic texture_offset_base, texture_offset_base, #0xF
5162 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5163 add block, block, num_blocks, lsl #6
5167 vext.32 block_masks_shifted, block_masks, block_masks, #1
5168 vorr.u32 block_masks, block_masks, block_masks_shifted
5169 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5172 add num_blocks, num_blocks, #1
5173 cmp num_blocks, #MAX_BLOCKS
5174 blgt setup_sprites_16bpp_flush_single
5176 and texture_block_ptr, texture_offset_base, texture_mask
5177 subs height, height, #1
5179 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5180 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5182 vst1.u32 { texels }, [ block, :128 ]
5183 add block, block, #40
5185 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5188 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5190 add block, block, #24
5191 add texture_offset_base, texture_offset_base, #2048
5192 add fb_ptr, fb_ptr, #2048
5193 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5196 ldmia sp!, { r4 - r11, pc }
5199 add num_blocks, num_blocks, block_width
5200 mov texture_offset, texture_offset_base
5202 cmp num_blocks, #MAX_BLOCKS
5203 blgt setup_sprites_16bpp_flush_row
5205 add texture_offset_base, texture_offset_base, #2048
5206 and texture_block_ptr, texture_offset, texture_mask
5208 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5209 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5211 vst1.u32 { texels }, [ block, :128 ]
5212 add block, block, #40
5214 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5215 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5218 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5219 subs blocks_remaining, block_width, #2
5221 add texture_offset, texture_offset, #16
5222 add fb_ptr, fb_ptr, #16
5224 vmov.u8 draw_mask_fb_ptr, #0
5226 add block, block, #24
5230 and texture_block_ptr, texture_offset, texture_mask
5231 subs blocks_remaining, blocks_remaining, #1
5233 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5234 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5236 vst1.u32 { texels }, [ block, :128 ]
5237 add block, block, #40
5239 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5242 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5244 add texture_offset, texture_offset, #16
5245 add fb_ptr, fb_ptr, #16
5247 add block, block, #24
5251 and texture_block_ptr, texture_offset, texture_mask
5252 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5254 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5255 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5257 vst1.u32 { texels }, [ block, :128 ]
5258 add block, block, #40
5260 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5261 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5263 add block, block, #24
5264 subs height, height, #1
5266 add fb_ptr, fb_ptr, fb_ptr_pitch
5267 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5271 ldmia sp!, { r4 - r11, pc }
5274 #undef texture_page_ptr
5276 #undef dirty_textures_mask
5277 #undef current_texture_mask
5280 #define current_texture_page r1
5281 #define texture_page_ptr r2
5282 #define vram_ptr_a r3
5283 #define current_texture_page_x r12
5284 #define current_texture_page_y r4
5285 #define dirty_textures_mask r5
5289 #define current_texture_mask r9
5291 #define vram_ptr_b r11
5293 #define texel_block_a d0
5294 #define texel_block_b d1
5295 #define texel_block_expanded_a q1
5296 #define texel_block_expanded_b q2
5297 #define texel_block_expanded_ab q2
5298 #define texel_block_expanded_c q3
5299 #define texel_block_expanded_d q4
5300 #define texel_block_expanded_cd q3
5302 function(update_texture_4bpp_cache)
5303 stmdb sp!, { r4 - r11, r14 }
5306 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5308 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5309 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5311 and current_texture_page_x, current_texture_page, #0xF
5312 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
5314 mov current_texture_page_y, current_texture_page, lsr #4
5315 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5317 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5320 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5321 bic dirty_textures_mask, current_texture_mask
5324 str dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5329 add vram_ptr_b, vram_ptr_a, #2048
5332 vld1.u32 { texel_block_a }, [ vram_ptr_a, :64 ], c_4096
5333 vld1.u32 { texel_block_b }, [ vram_ptr_b, :64 ], c_4096
5335 vmovl.u8 texel_block_expanded_a, texel_block_a
5336 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5337 vmovl.u8 texel_block_expanded_c, texel_block_b
5338 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5340 vbic.u16 texel_block_expanded_a, #0x00F0
5341 vbic.u16 texel_block_expanded_b, #0x00F0
5342 vbic.u16 texel_block_expanded_c, #0x00F0
5343 vbic.u16 texel_block_expanded_d, #0x00F0
5345 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
5346 texel_block_expanded_b
5347 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
5348 texel_block_expanded_d
5350 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
5351 [ texture_page_ptr, :256 ]!
5353 subs sub_y, sub_y, #1
5357 add vram_ptr_a, vram_ptr_a, #8
5358 add vram_ptr_b, vram_ptr_b, #8
5360 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5361 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5363 subs tile_x, tile_x, #1
5367 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5368 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5370 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5371 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5373 subs tile_y, tile_y, #1
5377 ldmia sp!, { r4 - r11, pc }
5380 #undef current_texture_page
5383 #define texture_page r1
5384 #define texture_page_ptr r2
5385 #define vram_ptr_a r3
5386 #define texture_page_x r12
5387 #define texture_page_y r4
5388 #define current_texture_page r5
5393 #define vram_ptr_b r11
5405 function(update_texture_8bpp_cache_slice)
5406 stmdb sp!, { r4 - r11, r14 }
5409 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5410 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5412 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5415 and texture_page_x, texture_page, #0xF
5416 mov texture_page_y, texture_page, lsr #4
5418 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
5421 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
5422 eor current_texture_page, current_texture_page, texture_page
5424 ands current_texture_page, current_texture_page, #0x1
5427 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
5430 add vram_ptr_b, vram_ptr_a, #2048
5433 vld1.u32 { texels_a }, [ vram_ptr_a, :128 ], c_4096
5434 vld1.u32 { texels_b }, [ vram_ptr_b, :128 ], c_4096
5435 vld1.u32 { texels_c }, [ vram_ptr_a, :128 ], c_4096
5436 vld1.u32 { texels_d }, [ vram_ptr_b, :128 ], c_4096
5438 vst1.u32 { texels_a, texels_b }, [ texture_page_ptr, :256 ]!
5439 vst1.u32 { texels_c, texels_d }, [ texture_page_ptr, :256 ]!
5441 subs sub_y, sub_y, #1
5446 add vram_ptr_a, vram_ptr_a, #16
5447 add vram_ptr_b, vram_ptr_b, #16
5449 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5450 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5452 subs tile_x, tile_x, #1
5457 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5458 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5460 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5461 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5463 subs tile_y, tile_y, #1
5464 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
5469 ldmia sp!, { r4 - r11, pc }