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_primitive_color_offset 184
42 #define psx_gpu_dither_table_offset 188
43 #define psx_gpu_render_block_handler_offset 204
44 #define psx_gpu_texture_page_ptr_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 #define function(name) \
253 function(compute_all_gradients)
254 // First compute the triangle area reciprocal and shift. The division will
255 // happen concurrently with much of the work which follows.
256 @ r12 = psx_gpu->triangle_area
257 ldr r12, [ psx_gpu, #psx_gpu_triangle_area_offset ]
258 stmdb sp!, { r4 - r11, lr }
260 @ load exponent of 62 into upper half of double
262 clz r14, r12 @ r14 = shift
264 movt r4, #((62 + 1023) << 4)
265 mov r12, r12, lsl r14 @ r12 = triangle_area_normalized
267 @ load area normalized into lower half of double
269 vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n
271 movt r4, #((1022 + 31) << 4)
274 add r4, r4, r12, lsr #11
277 vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n
279 // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) =
280 // ( d0 * d1 ) - ( d2 * d3 ) =
281 // ( m0 ) - ( m1 ) = gradient
283 // This is split to do 12 elements at a time over three sets: a, b, and c.
284 // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so
285 // two of the slots are unused.
287 // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as
290 // First type is: uvrg bxxx xxxx
291 // Second type is: yyyy ybyy uvrg
292 // Since x_a and y_c are the same the same variable is used for both.
294 vld1.u32 { v0 }, [ v_a, : 128 ] @ v0 = { uvrg0, b0, x0, y0 }
295 ldrsh x0, [ v_a, #8 ] @ load x0
297 vld1.u32 { v1 }, [ v_b, : 128 ] @ v1 = { uvrg1, b1, x1, y1}
298 ldrh x1, [ v_b, #8 ] @ load x1
300 vld1.u32 { v2 }, [ v_c, : 128 ] @ v2 = { uvrg2, b2, x2, y2 }
301 ldrh x2, [ v_c, #8 ] @ load x2
303 vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- }
304 ldrh y0, [ v_a, #10 ] @ load y0
306 vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- }
307 ldrh y1, [ v_b, #10 ] @ load y1
309 vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- }
310 ldrh y2, [ v_c, #10 ] @ load y2
312 vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 }
313 vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 }
315 orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 }
316 pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 }
318 vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 }
319 vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 }
321 vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 }
322 vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 }
324 ldrb b2, [ v_c, #4 ] @ load b2
325 orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 }
327 ldrb b1, [ v_b, #4 ] @ load b1
328 orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 }
330 vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 }
331 vsub.s16 d0_ab, x1_ab, x0_ab
333 ldrb b0, [ v_a, #4 ] @ load b0
334 orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 }
336 vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 }
337 vsub.s16 d2_ab, x2_ab, x1_ab
339 vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 }
340 vsub.s16 d1_ab, y2_ab, y1_ab
342 orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 }
343 ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 }
345 ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 }
346 ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 }
348 vsub.s16 d3_ab, y1_ab, y0_ab
349 smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) -
350 @ ((x2 - X1) * (b1 - b0))
351 vmull.s16 ga_uvrg_x, d0_a, d1_a
352 smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) -
353 @ ((b2 - b1) * (y1 - y0))
354 vmlsl.s16 ga_uvrg_x, d2_a, d3_a
355 movs gs_bx, ga_bx, asr #31
357 vmull.s16 ga_uvrg_y, d0_b, d1_b
358 rsbmi ga_bx, ga_bx, #0
360 vmlsl.s16 ga_uvrg_y, d2_b, d3_b
361 movs gs_by, ga_by, asr #31
363 vshr.u64 d0, d30, #22
364 mov b_base, b0, lsl #16
366 rsbmi ga_by, ga_by, #0
367 vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0
369 @ r12 = psx_gpu->triangle_winding_offset
370 ldrb r12, [ psx_gpu, #psx_gpu_triangle_winding_offset ]
371 vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0
373 add b_base, b_base, #0x8000
374 rsb r12, r12, #0 @ r12 = -(triangle->winding)
376 vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w }
377 sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS)
379 vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16
380 vdup.u32 r_shift, r14 @ r_shift = { shift, shift, shift, shift }
382 vorr.u32 uvrg_base, #0x8000
383 vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x)
385 vmov area_r_s, s0 @ area_r_s = triangle_reciprocal
386 vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y)
388 vmull.u32 gw_rg_x, ga_rg_x, d0[0]
389 vmull.u32 gw_uv_x, ga_uv_x, d0[0]
390 vmull.u32 gw_rg_y, ga_rg_y, d0[0]
391 vmull.u32 gw_uv_y, ga_uv_y, d0[0]
393 vshl.u64 gw_rg_x, gw_rg_x, r_shift
394 vshl.u64 gw_uv_x, gw_uv_x, r_shift
395 vshl.u64 gw_rg_y, gw_rg_y, r_shift
396 vshl.u64 gw_uv_y, gw_uv_y, r_shift
398 veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask
399 vmovn.u64 g_uv_x, gw_uv_x
401 veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask
402 vmovn.u64 g_rg_x, gw_rg_x
404 veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
405 vmovn.u64 g_uv_y, gw_uv_y
407 vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
408 vmovn.u64 g_rg_y, gw_rg_y
410 veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
411 mov ga_bx, ga_bx, lsl #13
413 vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
414 mov ga_by, ga_by, lsl #13
417 umull gw_bx_l, gw_bx_h, ga_bx, area_r_s
419 vshl.u32 g_uvrg_x, g_uvrg_x, #4
420 vshl.u32 g_uvrg_y, g_uvrg_y, #4
422 umull gw_by_l, gw_by_h, ga_by, area_r_s
423 vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0]
425 eor gs_bx, gs_bx, r12
426 vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1
428 veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0
429 eor gs_by, gs_by, r12
431 rsb r11, r14, #0 @ r11 = negative shift for scalar lsr
432 add store_a, psx_gpu, #psx_gpu_uvrg_offset
434 sub r11, r11, #(32 - 13)
436 add store_b, store_a, #16
439 vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1
440 vst1.u32 { uvrg_base }, [ store_a, : 128 ], store_inc
442 vst1.u32 { uvrg_dx1 }, [ store_b, : 128 ], store_inc
443 mov g_bx, gw_bx_h, lsr r11
445 vst1.u32 { g_uvrg_y }, [ store_a, : 128 ], store_inc
446 mov g_by, gw_by_h, lsr r11
448 vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \
449 [ store_b, : 128 ], store_inc
450 eor g_bx, g_bx, gs_bx
452 vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \
453 [ store_b, : 128 ], store_inc
454 sub g_bx, g_bx, gs_bx
457 eor g_by, g_by, gs_by
459 mls b_base, g_bx, x0, b_base
460 sub g_by, g_by, gs_by
465 add g_bx2, g_bx, g_bx
466 add g_bx3, g_bx, g_bx2
468 stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by }
470 ldmia sp!, { r4 - r11, pc }
487 #define height_minor_a r7
488 #define height_minor_b r8
489 #define height_major r9
492 #define reciprocal_table_ptr r10
494 #define edge_alt_low r4
495 #define edge_alt_high r5
496 #define edge_dx_dy_alt r6
497 #define edge_shift_alt r10
499 #define edge_dx_dy_alt_low r4
500 #define edge_dx_dy_alt_high r5
502 #define span_edge_data r4
503 #define span_uvrg_offset r5
504 #define span_b_offset r6
512 #define alternate_x q0
513 #define alternate_dx_dy q1
514 #define alternate_x_32 q2
516 #define alternate_x_low d0
517 #define alternate_x_high d1
518 #define alternate_dx_dy_low d2
519 #define alternate_dx_dy_high d3
520 #define alternate_x_32_low d4
521 #define alternate_x_32_high d5
525 #define left_dx_dy q5
526 #define right_dx_dy q6
528 #define right_edge q8
530 #define left_x_low d6
531 #define left_x_high d7
532 #define right_x_low d8
533 #define right_x_high d9
534 #define left_dx_dy_low d10
535 #define left_dx_dy_high d11
536 #define right_dx_dy_low d12
537 #define right_dx_dy_high d13
538 #define left_edge_low d14
539 #define left_edge_high d15
540 #define right_edge_low d16
541 #define right_edge_high d17
543 #define y_mid_point d18
546 #define left_right_x_16 q11
547 #define span_shifts_y q12
550 #define span_shifts d24
555 #define left_right_x_16_low d22
556 #define left_right_x_16_high d23
561 #define alternate_x_16 d4
564 #define v_clip_low d6
566 #define right_x_32 q10
567 #define left_x_32 q11
568 #define alternate_select d24
570 #define right_x_32_low d20
571 #define right_x_32_high d21
572 #define left_x_32_low d22
573 #define left_x_32_high d23
576 #define edges_dx_dy d2
577 #define edge_shifts d3
578 #define edge_shifts_64 q2
580 #define edges_xy_left d0
581 #define edges_xy_right d1
583 #define height_reciprocals d6
591 #define heights_b d12
592 #define edges_dx_dy_64 q10
594 #define edges_dx_dy_64_left d20
595 #define edges_dx_dy_64_right d21
598 #define setup_spans_prologue() \
599 stmdb sp!, { r4 - r11, lr }; \
601 ldrsh x_a, [ v_a, #8 ]; \
602 ldrsh x_b, [ v_b, #8 ]; \
603 ldrsh x_c, [ v_c, #8 ]; \
604 ldrsh y_a, [ v_a, #10 ]; \
605 ldrsh y_b, [ v_b, #10 ]; \
606 ldrsh y_c, [ v_c, #10 ]; \
608 add temp, psx_gpu, #psx_gpu_uvrg_offset; \
609 vld1.32 { uvrg }, [ temp ]; \
610 add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \
611 vld1.32 { uvrg_dy }, [ temp ]; \
612 movw reciprocal_table_ptr, :lower16:reciprocal_table; \
613 movt reciprocal_table_ptr, :upper16:reciprocal_table; \
615 vmov.u32 c_0x01, #0x01 \
617 #define setup_spans_load_b() \
618 ldr b, [ psx_gpu, #psx_gpu_b_offset ]; \
619 ldr b_dy, [ psx_gpu, #psx_gpu_b_dy_offset ] \
621 #define setup_spans_prologue_b() \
622 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
623 add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \
625 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
626 vmov.u16 c_0x0004, #0x0004; \
628 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
629 vmov.u16 c_0x0001, #0x0001; \
631 vld1.u16 { left_edge_low[], left_edge_high[] }, [ temp ]; \
632 add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \
634 vld1.u16 { right_edge_low[], right_edge_high[] }, [ temp ]; \
635 vadd.u16 right_edge, right_edge, c_0x0001; \
637 vmov.u16 c_0x0007, #0x0007; \
638 vmvn.u16 c_0xFFFE, #0x0001 \
641 #define compute_edge_delta_x2() \
642 ldr temp, [ reciprocal_table_ptr, height, lsl #2 ]; \
644 vdup.u32 heights, height; \
645 vsub.u32 widths, x_ends, x_starts; \
647 vdup.u32 edge_shifts, temp; \
648 vsub.u32 heights_b, heights, c_0x01; \
649 vshr.u32 height_reciprocals, edge_shifts, #12; \
651 vmla.s32 heights_b, x_starts, heights; \
652 vbic.u16 edge_shifts, #0xE0; \
653 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
654 vmull.s32 edges_xy, heights_b, height_reciprocals \
657 #define height_reciprocal_alt r11
658 #define height_b_alt r12
660 #define compute_edge_delta_x3(start_c, height_a, height_b) \
661 vmov.u32 heights, height_a, height_b; \
662 ldr temp, [ reciprocal_table_ptr, height_a, lsl #2 ]; \
663 vmov.u32 edge_shifts[0], temp; \
664 ldr temp, [ reciprocal_table_ptr, height_b, lsl #2 ]; \
665 vmov.u32 edge_shifts[1], temp; \
666 ldr edge_shift_alt, [ reciprocal_table_ptr, height_minor_b, lsl #2 ]; \
668 vsub.u32 widths, x_ends, x_starts; \
669 sub width_alt, x_c, start_c; \
671 vsub.u32 heights_b, heights, c_0x01; \
672 sub height_b_alt, height_minor_b, #1; \
674 vshr.u32 height_reciprocals, edge_shifts, #12; \
675 lsr height_reciprocal_alt, edge_shift_alt, #12; \
677 vmla.s32 heights_b, x_starts, heights; \
678 mla height_b_alt, height_minor_b, start_c, height_b_alt; \
680 vbic.u16 edge_shifts, #0xE0; \
681 and edge_shift_alt, edge_shift_alt, #0x1F; \
683 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
684 mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \
686 vmull.s32 edges_xy, heights_b, height_reciprocals; \
687 smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \
690 #define setup_spans_adjust_y_up() \
691 vsub.u32 y_x4, y_x4, c_0x0004 \
693 #define setup_spans_adjust_y_down() \
694 vadd.u32 y_x4, y_x4, c_0x0004 \
696 #define setup_spans_adjust_interpolants_up() \
697 vsub.u32 uvrg, uvrg, uvrg_dy; \
700 #define setup_spans_adjust_interpolants_down() \
701 vadd.u32 uvrg, uvrg, uvrg_dy; \
705 #define setup_spans_clip_interpolants_increment() \
706 mla b, b_dy, clip, b; \
707 vmla.s32 uvrg, uvrg_dy, v_clip \
709 #define setup_spans_clip_interpolants_decrement() \
710 mls b, b_dy, clip, b; \
711 vmls.s32 uvrg, uvrg_dy, v_clip \
713 #define setup_spans_clip_alternate_yes() \
714 smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \
716 #define setup_spans_clip_alternate_no() \
718 #define setup_spans_clip(direction, alternate_active) \
719 vdup.u32 v_clip, clip; \
720 setup_spans_clip_alternate_##alternate_active(); \
721 setup_spans_clip_interpolants_##direction(); \
722 vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \
725 #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \
726 vmovl.s32 edge_shifts_64, edge_shifts; \
727 vmovl.s32 edges_dx_dy_64, edges_dx_dy; \
729 vshl.s64 edges_xy, edges_xy, edge_shifts_64; \
730 vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \
732 vmov left_x_low, edges_xy_##left_index; \
733 vmov right_x_low, edges_xy_##right_index; \
735 vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \
736 vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \
737 vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \
738 vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \
740 vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \
741 vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \
743 vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \
744 vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \
747 #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \
748 setup_spans_adjust_edges_alternate_no(left_index, right_index); \
750 vdup.u16 y_mid_point, y_b; \
751 rsb temp, edge_shift_alt, #32; \
753 lsl edge_alt_high, edge_alt_high, edge_shift_alt; \
754 orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \
755 lsl edge_alt_low, edge_alt_low, edge_shift_alt; \
756 vmov alternate_x_low, edge_alt_low, edge_alt_high; \
758 asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \
759 lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \
760 vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \
761 vmov alternate_dx_dy_high, alternate_dx_dy_low; \
763 vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \
764 vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \
767 #define setup_spans_y_select_up() \
768 vclt.s16 alternate_select, y_x4, y_mid_point \
770 #define setup_spans_y_select_down() \
771 vcgt.s16 alternate_select, y_x4, y_mid_point \
774 #define setup_spans_alternate_select_left() \
775 vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \
777 #define setup_spans_alternate_select_right() \
778 vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \
781 #define setup_spans_set_x4_alternate_yes(alternate, direction) \
782 vshrn.s64 alternate_x_32_low, alternate_x, #32; \
783 vshrn.s64 left_x_32_low, left_x, #32; \
784 vshrn.s64 right_x_32_low, right_x, #32; \
786 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
787 vadd.u64 left_x, left_x, left_dx_dy; \
788 vadd.u64 right_x, right_x, right_dx_dy; \
790 vshrn.s64 alternate_x_32_high, alternate_x, #32; \
791 vshrn.s64 left_x_32_high, left_x, #32; \
792 vshrn.s64 right_x_32_high, right_x, #32; \
794 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
795 vadd.u64 left_x, left_x, left_dx_dy; \
796 vadd.u64 right_x, right_x, right_dx_dy; \
798 vmovn.u32 alternate_x_16, alternate_x_32; \
799 setup_spans_y_select_##direction(); \
800 vmovn.u32 left_right_x_16_low, left_x_32; \
802 vmovn.u32 left_right_x_16_high, right_x_32; \
803 setup_spans_alternate_select_##alternate(); \
805 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
806 str b, [ span_b_offset ], #4; \
807 setup_spans_adjust_interpolants_##direction(); \
809 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
811 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
812 str b, [ span_b_offset ], #4; \
813 setup_spans_adjust_interpolants_##direction(); \
815 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
817 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
818 str b, [ span_b_offset ], #4; \
819 setup_spans_adjust_interpolants_##direction(); \
821 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
822 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
823 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
825 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
826 str b, [ span_b_offset ], #4; \
827 setup_spans_adjust_interpolants_##direction(); \
829 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
830 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
832 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
834 setup_spans_adjust_y_##direction() \
837 #define setup_spans_set_x4_alternate_no(alternate, direction) \
838 vshrn.s64 left_x_32_low, left_x, #32; \
839 vshrn.s64 right_x_32_low, right_x, #32; \
841 vadd.u64 left_x, left_x, left_dx_dy; \
842 vadd.u64 right_x, right_x, right_dx_dy; \
844 vshrn.s64 left_x_32_high, left_x, #32; \
845 vshrn.s64 right_x_32_high, right_x, #32; \
847 vadd.u64 left_x, left_x, left_dx_dy; \
848 vadd.u64 right_x, right_x, right_dx_dy; \
850 vmovn.u32 left_right_x_16_low, left_x_32; \
851 vmovn.u32 left_right_x_16_high, right_x_32; \
853 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
854 str b, [ span_b_offset ], #4; \
855 setup_spans_adjust_interpolants_##direction(); \
857 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
859 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
860 str b, [ span_b_offset ], #4; \
861 setup_spans_adjust_interpolants_##direction(); \
863 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
865 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
866 str b, [ span_b_offset ], #4; \
867 setup_spans_adjust_interpolants_##direction(); \
869 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
870 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
871 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
873 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
874 str b, [ span_b_offset ], #4; \
875 setup_spans_adjust_interpolants_##direction(); \
877 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
878 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
880 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
882 setup_spans_adjust_y_##direction() \
885 #define edge_adjust_low r11
886 #define edge_adjust_high r12
888 #define setup_spans_alternate_adjust_yes() \
889 smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \
890 subs edge_alt_low, edge_alt_low, edge_adjust_low; \
891 sbc edge_alt_high, edge_alt_high, edge_adjust_high \
893 #define setup_spans_alternate_adjust_no() \
896 #define setup_spans_down(left_index, right_index, alternate, alternate_active) \
897 setup_spans_alternate_adjust_##alternate_active(); \
898 setup_spans_load_b(); \
900 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
901 subs y_c, y_c, temp; \
902 subgt height, height, y_c; \
903 addgt height, height, #1; \
905 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
906 subs clip, temp, y_a; \
909 sub height, height, clip; \
910 add y_a, y_a, clip; \
911 setup_spans_clip(increment, alternate_active); \
917 orr temp, y_a, y_a, lsl #16; \
918 add temp, temp, #(1 << 16); \
920 add y_a, y_a, #(2 << 16); \
921 vmov.u32 y_x4, temp, y_a; \
923 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
925 setup_spans_prologue_b(); \
927 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
930 setup_spans_set_x4_alternate_##alternate_active(alternate, down); \
931 subs height, height, #4; \
937 #define setup_spans_alternate_pre_increment_yes() \
938 adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \
939 adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \
941 #define setup_spans_alternate_pre_increment_no() \
944 #define setup_spans_up_decrement_yes() \
945 suble height, height, #1 \
947 #define setup_spans_up_decrement_no() \
950 #define setup_spans_up(left_index, right_index, alternate, alternate_active) \
951 setup_spans_alternate_adjust_##alternate_active(); \
952 setup_spans_load_b(); \
955 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
956 subs temp, temp, y_c; \
957 subgt height, height, temp; \
958 setup_spans_up_decrement_##alternate_active(); \
960 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
961 subs clip, y_a, temp; \
964 sub height, height, clip; \
965 sub y_a, y_a, clip; \
966 setup_spans_clip(decrement, alternate_active); \
972 orr temp, y_a, y_a, lsl #16; \
973 sub temp, temp, #(1 << 16); \
975 sub y_a, y_a, #(2 << 16); \
976 vmov.u32 y_x4, temp, y_a; \
978 vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \
980 setup_spans_alternate_pre_increment_##alternate_active(); \
981 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
983 setup_spans_adjust_interpolants_up(); \
984 setup_spans_prologue_b(); \
986 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
989 setup_spans_set_x4_alternate_##alternate_active(alternate, up); \
990 subs height, height, #4; \
996 #define setup_spans_epilogue() \
997 ldmia sp!, { r4 - r11, pc } \
1000 #define setup_spans_up_up(minor, major) \
1001 setup_spans_prologue(); \
1002 sub height_minor_a, y_a, y_b; \
1003 sub height_minor_b, y_b, y_c; \
1004 sub height, y_a, y_c; \
1006 vdup.u32 x_starts, x_a; \
1007 vmov.u32 x_ends, x_c, x_b; \
1009 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
1010 setup_spans_up(major, minor, minor, yes); \
1011 setup_spans_epilogue() \
1013 function(setup_spans_up_left)
1014 setup_spans_up_up(left, right)
1016 function(setup_spans_up_right)
1017 setup_spans_up_up(right, left)
1020 #define setup_spans_down_down(minor, major) \
1021 setup_spans_prologue(); \
1022 sub height_minor_a, y_b, y_a; \
1023 sub height_minor_b, y_c, y_b; \
1024 sub height, y_c, y_a; \
1026 vdup.u32 x_starts, x_a; \
1027 vmov.u32 x_ends, x_c, x_b; \
1029 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
1030 setup_spans_down(major, minor, minor, yes); \
1031 setup_spans_epilogue() \
1033 function(setup_spans_down_left)
1034 setup_spans_down_down(left, right)
1036 function(setup_spans_down_right)
1037 setup_spans_down_down(right, left)
1040 #define setup_spans_up_flat() \
1041 sub height, y_a, y_c; \
1043 compute_edge_delta_x2(); \
1044 setup_spans_up(left, right, none, no); \
1045 setup_spans_epilogue() \
1047 function(setup_spans_up_a)
1048 setup_spans_prologue()
1050 vmov.u32 x_starts, x_a, x_b
1051 vdup.u32 x_ends, x_c
1053 setup_spans_up_flat()
1055 function(setup_spans_up_b)
1056 setup_spans_prologue()
1058 vdup.u32 x_starts, x_a
1059 vmov.u32 x_ends, x_b, x_c
1061 setup_spans_up_flat()
1063 #define setup_spans_down_flat() \
1064 sub height, y_c, y_a; \
1066 compute_edge_delta_x2(); \
1067 setup_spans_down(left, right, none, no); \
1068 setup_spans_epilogue() \
1070 function(setup_spans_down_a)
1071 setup_spans_prologue()
1073 vmov.u32 x_starts, x_a, x_b
1074 vdup.u32 x_ends, x_c
1076 setup_spans_down_flat()
1078 function(setup_spans_down_b)
1079 setup_spans_prologue()
1081 vdup.u32 x_starts, x_a
1082 vmov.u32 x_ends, x_b, x_c
1084 setup_spans_down_flat()
1089 #define edges_xy_b q11
1090 #define edges_dx_dy_b d26
1091 #define edge_shifts_b d27
1092 #define edges_dx_dy_and_shifts_b q13
1093 #define height_increment d20
1095 #define edges_dx_dy_and_shifts q1
1097 #define edges_xy_b_left d22
1098 #define edges_xy_b_right d23
1100 #define setup_spans_up_down_load_edge_set_b() \
1101 vmov edges_xy, edges_xy_b; \
1102 vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \
1105 function(setup_spans_up_down)
1106 setup_spans_prologue()
1108 // s32 middle_y = y_a;
1109 sub height_minor_a, y_a, y_b
1110 sub height_minor_b, y_c, y_a
1111 sub height_major, y_c, y_b
1113 vmov.u32 x_starts, x_a, x_c
1114 vdup.u32 x_ends, x_b
1116 compute_edge_delta_x3(x_a, height_minor_a, height_major)
1119 vmov.u32 height_increment, temp, height_minor_b
1120 vmlal.s32 edges_xy, edges_dx_dy, height_increment
1122 vmov edges_xy_b_left, edge_alt_low, edge_alt_high
1123 vmov edges_xy_b_right, edges_xy_right
1125 vmov edge_shifts_b, edge_shifts
1126 vmov.u32 edge_shifts_b[0], edge_shift_alt
1128 vneg.s32 edges_dx_dy_b, edges_dx_dy
1129 vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt
1133 setup_spans_load_b()
1136 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1137 subs temp, temp, y_b
1138 subgt height_minor_a, height_minor_a, temp
1140 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1141 subs clip, y_a, temp
1144 sub height_minor_a, height_minor_a, clip
1146 setup_spans_clip(decrement, no)
1149 cmp height_minor_a, #0
1152 orr temp, y_a, y_a, lsl #16
1153 sub temp, temp, #(1 << 16)
1155 sub y_a, y_a, #(2 << 16)
1156 vmov.u32 y_x4, temp, y_a
1158 vaddw.s32 edges_xy, edges_xy, edges_dx_dy
1160 strh height_minor_a, [ psx_gpu, #psx_gpu_num_spans_offset ]
1162 setup_spans_adjust_edges_alternate_no(left, right);
1163 setup_spans_adjust_interpolants_up()
1164 setup_spans_up_down_load_edge_set_b()
1166 setup_spans_prologue_b()
1170 setup_spans_set_x4_alternate_no(none, up)
1171 subs height_minor_a, height_minor_a, #4
1174 add span_edge_data, span_edge_data, height_minor_a, lsl #3
1175 add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4
1176 add span_b_offset, span_b_offset, height_minor_a, lsl #2
1179 add temp, psx_gpu, #psx_gpu_uvrg_offset
1180 vld1.32 { uvrg }, [ temp ]
1183 setup_spans_load_b()
1185 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1187 subgt height_minor_b, height_minor_b, y_c
1188 addgt height_minor_b, height_minor_b, #1
1190 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1191 subs clip, temp, y_a
1194 sub height_minor_b, height_minor_b, clip
1196 setup_spans_clip(increment, no)
1199 cmp height_minor_b, #0
1202 orr temp, y_a, y_a, lsl #16
1203 add temp, temp, #(1 << 16)
1205 add y_a, y_a, #(2 << 16)
1206 vmov.u32 y_x4, temp, y_a
1208 setup_spans_adjust_edges_alternate_no(left, right)
1210 ldrh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1211 add temp, temp, height_minor_b
1212 strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1215 setup_spans_set_x4_alternate_no(none, down)
1216 subs height_minor_b, height_minor_b, #4
1220 setup_spans_epilogue()
1223 setup_spans_up_down_load_edge_set_b()
1224 setup_spans_prologue_b()
1228 #undef span_uvrg_offset
1229 #undef span_edge_data
1230 #undef span_b_offset
1235 #define num_spans r1
1236 #define span_uvrg_offset r2
1237 #define span_edge_data r3
1238 #define span_b_offset r4
1240 #define span_num_blocks r6
1244 #define dither_offset_ptr r10
1245 #define block_ptr_a r11
1247 #define num_blocks r14
1249 #define uvrg_dx_ptr r2
1250 #define texture_mask_ptr r3
1251 #define dither_shift r8
1252 #define dither_row r10
1257 #define block_ptr_b r10
1259 #define block_span_ptr r10
1260 #define right_mask r8
1279 #define b_whole_8 d14
1280 #define fb_mask_ptrs d15
1293 #define u_whole_low d16
1294 #define u_whole_high d17
1295 #define v_whole_low d18
1296 #define v_whole_high d19
1297 #define r_whole_low d20
1298 #define r_whole_high d21
1299 #define g_whole_low d22
1300 #define g_whole_high d23
1301 #define b_whole_low d24
1302 #define b_whole_high d25
1307 #define u_whole_8 d26
1308 #define v_whole_8 d27
1309 #define u_whole_8b d24
1310 #define r_whole_8 d24
1311 #define g_whole_8 d25
1313 #define uv_whole_8 q13
1314 #define uv_whole_8b q14
1316 #define dither_offsets q14
1317 #define texture_mask q15
1318 #define texture_mask_u d30
1319 #define texture_mask_v d31
1321 #define dither_offsets_short d28
1325 #define block_span q10
1330 #define draw_mask q1
1331 #define draw_mask_edge q13
1332 #define test_mask q0
1338 #define setup_blocks_texture_swizzled() \
1339 vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \
1340 vsli.u8 u_whole_8, v_whole_8, #4; \
1341 vsri.u8 v_whole_8, u_whole_8b, #4 \
1343 #define setup_blocks_texture_unswizzled() \
1346 #define setup_blocks_shaded_textured_builder(swizzling) \
1349 function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \
1350 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1351 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1353 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1354 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1356 cmp num_spans, #0; \
1359 stmdb sp!, { r4 - r11, r14 }; \
1360 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1362 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
1363 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1365 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1366 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1368 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1369 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1371 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
1372 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1374 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1377 vmov.u8 fb_mask_ptrs, #0; \
1379 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1380 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1382 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1383 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
1385 cmp span_num_blocks, #0; \
1388 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1389 add num_blocks, span_num_blocks, num_blocks; \
1391 cmp num_blocks, #MAX_BLOCKS; \
1395 ldr b, [ span_b_offset ]; \
1396 add fb_ptr, fb_ptr, y, lsl #11; \
1398 vdup.u32 v_left_x, left_x; \
1401 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1402 add fb_ptr, fb_ptr, left_x, lsl #1; \
1404 mla b, b_dx, left_x, b; \
1405 and dither_shift, left_x, #0x03; \
1407 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1408 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1410 mov dither_shift, dither_shift, lsl #3; \
1411 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1414 subs span_num_blocks, span_num_blocks, #1; \
1416 mov dither_row, dither_row, ror dither_shift; \
1417 mov b_dx4, b_dx, lsl #2; \
1419 vdup.u32 dither_offsets_short, dither_row; \
1420 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1422 vdup.u32 b_block, b; \
1423 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1425 vdup.u32 u_block, uv[0]; \
1426 mov b_dx8, b_dx, lsl #3; \
1428 vdup.u32 v_block, uv[1]; \
1429 vdup.u32 r_block, rg[0]; \
1430 vdup.u32 g_block, rg[1]; \
1432 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1434 vadd.u32 u_block, u_block, block_span; \
1435 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1437 vadd.u32 v_block, v_block, block_span; \
1438 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1440 vadd.u32 r_block, r_block, block_span; \
1441 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1443 vadd.u32 g_block, g_block, block_span; \
1444 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
1446 vadd.u32 b_block, b_block, block_span; \
1447 add block_ptr_b, block_ptr_a, #16; \
1449 vshrn.u32 u_whole_low, u_block, #16; \
1450 vshrn.u32 v_whole_low, v_block, #16; \
1451 vshrn.u32 r_whole_low, r_block, #16; \
1452 vshrn.u32 g_whole_low, g_block, #16; \
1454 vdup.u32 dx4, uv_dx4[0]; \
1455 vshrn.u32 b_whole_low, b_block, #16; \
1457 vaddhn.u32 u_whole_high, u_block, dx4; \
1458 vdup.u32 dx4, uv_dx4[1]; \
1460 vaddhn.u32 v_whole_high, v_block, dx4; \
1461 vdup.u32 dx4, rg_dx4[0]; \
1463 vaddhn.u32 r_whole_high, r_block, dx4; \
1464 vdup.u32 dx4, rg_dx4[1]; \
1466 vaddhn.u32 g_whole_high, g_block, dx4; \
1467 vdup.u32 dx4, b_dx4; \
1469 vaddhn.u32 b_whole_high, b_block, dx4; \
1470 vdup.u32 dx8, uv_dx8[0]; \
1472 vadd.u32 u_block, u_block, dx8; \
1473 vdup.u32 dx8, uv_dx8[1]; \
1475 vadd.u32 v_block, v_block, dx8; \
1476 vdup.u32 dx8, rg_dx8[0]; \
1478 vadd.u32 r_block, r_block, dx8; \
1479 vdup.u32 dx8, rg_dx8[1]; \
1481 vadd.u32 g_block, g_block, dx8; \
1482 vdup.u32 dx8, b_dx8; \
1484 vadd.u32 b_block, b_block, dx8; \
1485 vmovn.u16 u_whole_8, u_whole; \
1487 vmovn.u16 v_whole_8, v_whole; \
1489 vmovn.u16 b_whole_8, b_whole; \
1491 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1493 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1494 setup_blocks_texture_##swizzling(); \
1496 vmovn.u16 r_whole_8, r_whole; \
1500 vmovn.u16 g_whole_8, g_whole; \
1501 vshrn.u32 u_whole_low, u_block, #16; \
1503 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1504 vshrn.u32 v_whole_low, v_block, #16; \
1506 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1507 vshrn.u32 r_whole_low, r_block, #16; \
1509 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1510 vshrn.u32 g_whole_low, g_block, #16; \
1512 vdup.u32 dx4, uv_dx4[0]; \
1513 vshrn.u32 b_whole_low, b_block, #16; \
1515 vaddhn.u32 u_whole_high, u_block, dx4; \
1516 vdup.u32 dx4, uv_dx4[1]; \
1518 vaddhn.u32 v_whole_high, v_block, dx4; \
1519 vdup.u32 dx4, rg_dx4[0]; \
1521 vaddhn.u32 r_whole_high, r_block, dx4; \
1522 vdup.u32 dx4, rg_dx4[1]; \
1524 vaddhn.u32 g_whole_high, g_block, dx4; \
1525 vdup.u32 dx4, b_dx4; \
1527 vaddhn.u32 b_whole_high, b_block, dx4; \
1528 vdup.u32 dx8, uv_dx8[0]; \
1530 vadd.u32 u_block, u_block, dx8; \
1531 vdup.u32 dx8, uv_dx8[1]; \
1533 vadd.u32 v_block, v_block, dx8; \
1534 vdup.u32 dx8, rg_dx8[0]; \
1536 vadd.u32 r_block, r_block, dx8; \
1537 vdup.u32 dx8, rg_dx8[1]; \
1539 vadd.u32 g_block, g_block, dx8; \
1540 vdup.u32 dx8, b_dx8; \
1542 vadd.u32 b_block, b_block, dx8; \
1543 vmovn.u16 u_whole_8, u_whole; \
1545 add fb_ptr, fb_ptr, #16; \
1546 vmovn.u16 v_whole_8, v_whole; \
1548 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1549 vmovn.u16 b_whole_8, b_whole; \
1553 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1554 subs span_num_blocks, span_num_blocks, #1; \
1556 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1557 setup_blocks_texture_##swizzling(); \
1559 vmovn.u16 r_whole_8, r_whole; \
1563 vmovn.u16 g_whole_8, g_whole; \
1564 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1566 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1567 vdup.u8 draw_mask, right_mask; \
1569 vmov.u32 fb_mask_ptrs[0], right_mask; \
1570 vtst.u16 draw_mask, draw_mask, test_mask; \
1571 vzip.u8 u_whole_8, v_whole_8; \
1573 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1574 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1575 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1576 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1577 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1580 add span_uvrg_offset, span_uvrg_offset, #16; \
1581 add span_b_offset, span_b_offset, #4; \
1583 add span_edge_data, span_edge_data, #8; \
1584 subs num_spans, num_spans, #1; \
1586 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1589 ldmia sp!, { r4 - r11, pc }; \
1592 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1593 vpush { texture_mask }; \
1594 vpush { uvrg_dx4 }; \
1596 stmdb sp!, { r0 - r3, r12, r14 }; \
1597 bl flush_render_block_buffer; \
1598 ldmia sp!, { r0 - r3, r12, r14 }; \
1600 vpop { uvrg_dx4 }; \
1601 vpop { texture_mask }; \
1603 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1604 vmov.u8 fb_mask_ptrs, #0; \
1606 mov num_blocks, span_num_blocks; \
1607 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1611 setup_blocks_shaded_textured_builder(swizzled)
1612 setup_blocks_shaded_textured_builder(unswizzled)
1615 #define setup_blocks_unshaded_textured_builder(swizzling) \
1618 function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \
1619 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1620 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1622 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1623 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1625 cmp num_spans, #0; \
1628 stmdb sp!, { r4 - r11, r14 }; \
1629 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1631 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1633 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1634 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1636 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1637 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1639 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1641 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1644 vmov.u8 fb_mask_ptrs, #0; \
1646 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1647 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1649 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1650 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
1652 cmp span_num_blocks, #0; \
1655 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1656 add num_blocks, span_num_blocks, num_blocks; \
1658 cmp num_blocks, #MAX_BLOCKS; \
1662 add fb_ptr, fb_ptr, y, lsl #11; \
1664 vdup.u32 v_left_x, left_x; \
1667 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1668 add fb_ptr, fb_ptr, left_x, lsl #1; \
1670 and dither_shift, left_x, #0x03; \
1672 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1673 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1675 mov dither_shift, dither_shift, lsl #3; \
1676 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1679 subs span_num_blocks, span_num_blocks, #1; \
1681 mov dither_row, dither_row, ror dither_shift; \
1683 vdup.u32 dither_offsets_short, dither_row; \
1684 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1686 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1688 vdup.u32 u_block, uv[0]; \
1690 vdup.u32 v_block, uv[1]; \
1691 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1693 vadd.u32 u_block, u_block, block_span; \
1694 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1696 vadd.u32 v_block, v_block, block_span; \
1697 add block_ptr_b, block_ptr_a, #16; \
1699 vshrn.u32 u_whole_low, u_block, #16; \
1700 vshrn.u32 v_whole_low, v_block, #16; \
1702 vdup.u32 dx4, uv_dx4[0]; \
1704 vaddhn.u32 u_whole_high, u_block, dx4; \
1705 vdup.u32 dx4, uv_dx4[1]; \
1707 vaddhn.u32 v_whole_high, v_block, dx4; \
1708 vdup.u32 dx8, uv_dx8[0]; \
1710 vadd.u32 u_block, u_block, dx8; \
1711 vdup.u32 dx8, uv_dx8[1]; \
1713 vadd.u32 v_block, v_block, dx8; \
1714 vmovn.u16 u_whole_8, u_whole; \
1716 vmovn.u16 v_whole_8, v_whole; \
1719 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1721 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1722 setup_blocks_texture_##swizzling(); \
1727 vshrn.u32 u_whole_low, u_block, #16; \
1729 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1730 vshrn.u32 v_whole_low, v_block, #16; \
1732 add block_ptr_b, block_ptr_b, #32; \
1733 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1735 vdup.u32 dx4, uv_dx4[0]; \
1736 vaddhn.u32 u_whole_high, u_block, dx4; \
1737 vdup.u32 dx4, uv_dx4[1]; \
1739 vaddhn.u32 v_whole_high, v_block, dx4; \
1740 vdup.u32 dx8, uv_dx8[0]; \
1742 vadd.u32 u_block, u_block, dx8; \
1743 vdup.u32 dx8, uv_dx8[1]; \
1745 vadd.u32 v_block, v_block, dx8; \
1746 vmovn.u16 u_whole_8, u_whole; \
1748 add fb_ptr, fb_ptr, #16; \
1749 vmovn.u16 v_whole_8, v_whole; \
1751 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1754 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1755 subs span_num_blocks, span_num_blocks, #1; \
1757 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1758 setup_blocks_texture_##swizzling(); \
1763 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1765 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1766 vdup.u8 draw_mask, right_mask; \
1768 vmov.u32 fb_mask_ptrs[0], right_mask; \
1769 vtst.u16 draw_mask, draw_mask, test_mask; \
1770 vzip.u8 u_whole_8, v_whole_8; \
1772 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1773 add block_ptr_b, block_ptr_b, #32; \
1774 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1775 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1776 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1779 add span_uvrg_offset, span_uvrg_offset, #16; \
1780 add span_edge_data, span_edge_data, #8; \
1781 subs num_spans, num_spans, #1; \
1783 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1786 ldmia sp!, { r4 - r11, pc }; \
1789 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1790 vpush { texture_mask }; \
1791 vpush { uvrg_dx4 }; \
1793 stmdb sp!, { r0 - r3, r12, r14 }; \
1794 bl flush_render_block_buffer; \
1795 ldmia sp!, { r0 - r3, r12, r14 }; \
1797 vpop { uvrg_dx4 }; \
1798 vpop { texture_mask }; \
1800 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1801 vmov.u8 fb_mask_ptrs, #0; \
1803 mov num_blocks, span_num_blocks; \
1804 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1808 setup_blocks_unshaded_textured_builder(swizzled)
1809 setup_blocks_unshaded_textured_builder(unswizzled)
1814 function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect)
1815 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1816 veor.u32 draw_mask, draw_mask, draw_mask
1821 stmdb sp!, { r4 - r11, r14 }
1822 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1824 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1826 ubfx color_r, color, #3, #5
1827 ubfx color_g, color, #11, #5
1828 ubfx color_b, color, #19, #5
1830 orr color, color_r, color_b, lsl #10
1831 orr color, color, color_g, lsl #5
1833 vdup.u16 colors, color
1835 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1836 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1838 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1839 add block_ptr_a, block_ptr_a, num_blocks, lsl #6
1842 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1843 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1845 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
1847 cmp span_num_blocks, #0
1850 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1851 add num_blocks, span_num_blocks, num_blocks
1853 cmp num_blocks, #MAX_BLOCKS
1857 add fb_ptr, fb_ptr, y, lsl #11
1860 add fb_ptr, fb_ptr, left_x, lsl #1
1863 subs span_num_blocks, span_num_blocks, #1
1865 add block_ptr_b, block_ptr_a, #16
1868 vmov.u32 fb_mask_ptrs[1], fb_ptr
1872 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_32
1873 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1874 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1876 add fb_ptr, fb_ptr, #16
1877 add block_ptr_b, block_ptr_b, #32
1881 vmov.u32 fb_mask_ptrs[1], fb_ptr
1882 subs span_num_blocks, span_num_blocks, #1
1887 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1889 vdup.u8 draw_mask_edge, right_mask
1890 vtst.u16 draw_mask_edge, draw_mask_edge, test_mask
1892 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1893 vst1.u32 { draw_mask_edge }, [ block_ptr_a, :128 ], c_32
1894 add block_ptr_b, block_ptr_b, #32
1895 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1898 add span_edge_data, span_edge_data, #8
1899 subs num_spans, num_spans, #1
1901 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1904 ldmia sp!, { r4 - r11, pc }
1909 stmdb sp!, { r0 - r3, r12, r14 }
1910 bl flush_render_block_buffer
1911 ldmia sp!, { r0 - r3, r12, r14 }
1915 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1916 veor.u32 draw_mask, draw_mask, draw_mask
1918 mov num_blocks, span_num_blocks
1919 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1923 #define mask_msb_scalar r14
1925 #define msb_mask q15
1927 #define pixels_low d16
1929 #define msb_mask_low d30
1930 #define msb_mask_high d31
1935 function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct)
1936 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1941 stmdb sp!, { r4 - r11, r14 }
1943 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1945 ubfx color_r, color, #3, #5
1946 ubfx color_g, color, #11, #5
1948 ldrh mask_msb_scalar, [ psx_gpu, #psx_gpu_mask_msb_offset ]
1949 ubfx color_b, color, #19, #5
1951 orr color, color_r, color_b, lsl #10
1952 orr color, color, color_g, lsl #5
1953 orr color, color, mask_msb_scalar
1955 vdup.u16 colors, color
1957 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1960 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1961 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1963 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
1965 cmp span_num_blocks, #0
1968 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1970 add fb_ptr, fb_ptr, y, lsl #11
1971 subs span_num_blocks, span_num_blocks, #1
1973 add fb_ptr, fb_ptr, left_x, lsl #1
1977 vst1.u32 { colors }, [ fb_ptr ]!
1978 subs span_num_blocks, span_num_blocks, #1
1983 ldrb right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1984 eor right_mask, right_mask, #0xFF
1987 strh color, [ fb_ptr ], #2
1988 movs right_mask, right_mask, lsr #1
1992 add span_edge_data, span_edge_data, #8
1993 subs num_spans, num_spans, #1
1995 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1998 ldmia sp!, { r4 - r11, pc }
2005 #define rg_dx_ptr r2
2023 #undef dither_offsets
2043 #define r_whole_low d6
2044 #define r_whole_high d7
2045 #define g_whole_low d8
2046 #define g_whole_high d9
2047 #define b_whole_low d10
2048 #define b_whole_high d11
2050 #define gb_whole_8 q6
2052 #define g_whole_8 d12
2053 #define b_whole_8 d13
2055 #define r_whole_8 d14
2067 #define block_span q5
2075 #define d128_0x7 q13
2079 #define dither_offsets q14
2080 #define draw_mask q15
2082 #define dither_offsets_low d28
2085 #define test_mask q10
2088 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2089 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2090 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2092 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2093 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2094 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2096 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2098 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2101 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2104 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2105 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2106 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2108 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2110 cmp num_spans, #0; \
2113 stmdb sp!, { r4 - r11, r14 }; \
2114 vshl.u32 rg_dx4, rg_dx, #2; \
2116 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2117 vshl.u32 rg_dx8, rg_dx, #3; \
2119 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2121 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2122 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2124 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2125 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2127 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2128 vmov.u8 d64_1, #1; \
2130 vmov.u8 d128_4, #4; \
2131 vmov.u8 d64_128, #128; \
2133 vmov.u8 d128_0x7, #0x7; \
2136 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2137 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2139 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2140 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
2142 cmp span_num_blocks, #0; \
2145 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2146 add num_blocks, span_num_blocks, num_blocks; \
2148 cmp num_blocks, #MAX_BLOCKS; \
2152 ldr b, [ span_b_offset ]; \
2153 add fb_ptr, fb_ptr, y, lsl #11; \
2155 vdup.u32 v_left_x, left_x; \
2158 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2159 add fb_ptr, fb_ptr, left_x, lsl #1; \
2161 mla b, b_dx, left_x, b; \
2162 and dither_shift, left_x, #0x03; \
2164 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2165 vshr.u32 rg_dx, rg_dx4, #2; \
2167 mov dither_shift, dither_shift, lsl #3; \
2168 vmla.u32 rg, rg_dx, v_left_x; \
2171 subs span_num_blocks, span_num_blocks, #1; \
2173 mov dither_row, dither_row, ror dither_shift; \
2174 mov b_dx4, b_dx, lsl #2; \
2176 vdup.u32 dither_offsets, dither_row; \
2177 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2179 vdup.u32 b_block, b; \
2180 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2182 mov b_dx8, b_dx, lsl #3; \
2183 vdup.u32 r_block, rg[0]; \
2184 vdup.u32 g_block, rg[1]; \
2186 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2188 vadd.u32 r_block, r_block, block_span; \
2189 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2191 vadd.u32 g_block, g_block, block_span; \
2192 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2194 vadd.u32 b_block, b_block, block_span; \
2195 add block_ptr_b, block_ptr_a, #16; \
2197 vshrn.u32 r_whole_low, r_block, #16; \
2198 vshrn.u32 g_whole_low, g_block, #16; \
2199 vshrn.u32 b_whole_low, b_block, #16; \
2200 vdup.u32 dx4, rg_dx4[0]; \
2202 vaddhn.u32 r_whole_high, r_block, dx4; \
2203 vdup.u32 dx4, rg_dx4[1]; \
2205 vaddhn.u32 g_whole_high, g_block, dx4; \
2206 vdup.u32 dx4, b_dx4; \
2208 vaddhn.u32 b_whole_high, b_block, dx4; \
2209 vdup.u32 dx8, rg_dx8[0]; \
2211 vadd.u32 r_block, r_block, dx8; \
2212 vdup.u32 dx8, rg_dx8[1]; \
2214 vadd.u32 g_block, g_block, dx8; \
2215 vdup.u32 dx8, b_dx8; \
2217 vadd.u32 b_block, b_block, dx8; \
2219 vmovn.u16 r_whole_8, r_whole; \
2220 vmovn.u16 g_whole_8, g_whole; \
2221 vmovn.u16 b_whole_8, b_whole; \
2224 veor.u32 draw_mask, draw_mask, draw_mask; \
2227 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2228 vshrn.u32 r_whole_low, r_block, #16; \
2230 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2231 vshrn.u32 g_whole_low, g_block, #16; \
2233 vshrn.u32 b_whole_low, b_block, #16; \
2234 str fb_ptr, [ block_ptr_a, #44 ]; \
2236 vdup.u32 dx4, rg_dx4[0]; \
2237 vshr.u8 r_whole_8, r_whole_8, #3; \
2238 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2240 vaddhn.u32 r_whole_high, r_block, dx4; \
2241 vdup.u32 dx4, rg_dx4[1]; \
2243 vaddhn.u32 g_whole_high, g_block, dx4; \
2244 vdup.u32 dx4, b_dx4; \
2246 vaddhn.u32 b_whole_high, b_block, dx4; \
2247 vdup.u32 dx8, rg_dx8[0]; \
2249 vmull.u8 pixels, r_whole_8, d64_1; \
2250 vmlal.u8 pixels, g_whole_8, d64_4; \
2251 vmlal.u8 pixels, b_whole_8, d64_128; \
2253 vadd.u32 r_block, r_block, dx8; \
2254 vdup.u32 dx8, rg_dx8[1]; \
2256 vadd.u32 g_block, g_block, dx8; \
2257 vdup.u32 dx8, b_dx8; \
2259 vadd.u32 b_block, b_block, dx8; \
2260 add fb_ptr, fb_ptr, #16; \
2262 vmovn.u16 r_whole_8, r_whole; \
2263 vmovn.u16 g_whole_8, g_whole; \
2264 vmovn.u16 b_whole_8, b_whole; \
2266 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2267 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2271 subs span_num_blocks, span_num_blocks, #1; \
2275 str fb_ptr, [ block_ptr_a, #44 ]; \
2276 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2278 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2279 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2281 vshr.u8 r_whole_8, r_whole_8, #3; \
2282 vdup.u8 draw_mask, right_mask; \
2284 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2285 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
2287 vtst.u16 draw_mask, draw_mask, test_mask; \
2289 vmull.u8 pixels, r_whole_8, d64_1; \
2290 vmlal.u8 pixels, g_whole_8, d64_4; \
2291 vmlal.u8 pixels, b_whole_8, d64_128; \
2293 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2294 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2297 add span_uvrg_offset, span_uvrg_offset, #16; \
2298 add span_b_offset, span_b_offset, #4; \
2300 add span_edge_data, span_edge_data, #8; \
2301 subs num_spans, num_spans, #1; \
2303 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2306 ldmia sp!, { r4 - r11, pc }; \
2309 /* TODO: Load from psx_gpu instead of saving/restoring these */\
2312 stmdb sp!, { r0 - r3, r12, r14 }; \
2313 bl flush_render_block_buffer; \
2314 ldmia sp!, { r0 - r3, r12, r14 }; \
2318 vmov.u8 d64_1, #1; \
2319 vmov.u8 d128_4, #4; \
2320 vmov.u8 d64_128, #128; \
2321 vmov.u8 d128_0x7, #0x7; \
2323 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2325 mov num_blocks, span_num_blocks; \
2326 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2330 setup_blocks_shaded_untextured_indirect_builder(undithered)
2331 setup_blocks_shaded_untextured_indirect_builder(dithered)
2336 #define mask_msb_ptr r14
2338 #define draw_mask q0
2339 #define pixels_low d16
2343 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2346 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2347 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2348 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2350 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2352 cmp num_spans, #0; \
2355 stmdb sp!, { r4 - r11, r14 }; \
2356 vshl.u32 rg_dx4, rg_dx, #2; \
2358 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2359 vshl.u32 rg_dx8, rg_dx, #3; \
2361 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2362 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2364 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2365 vmov.u8 d64_1, #1; \
2367 vmov.u8 d128_4, #4; \
2368 vmov.u8 d64_128, #128; \
2370 vmov.u8 d128_0x7, #0x7; \
2371 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2372 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
2375 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2376 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2378 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2379 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
2381 cmp span_num_blocks, #0; \
2384 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2385 add fb_ptr, fb_ptr, y, lsl #11; \
2387 ldr b, [ span_b_offset ]; \
2388 vdup.u32 v_left_x, left_x; \
2391 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2392 add fb_ptr, fb_ptr, left_x, lsl #1; \
2394 mla b, b_dx, left_x, b; \
2395 and dither_shift, left_x, #0x03; \
2397 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2398 vshr.u32 rg_dx, rg_dx4, #2; \
2400 mov dither_shift, dither_shift, lsl #3; \
2401 vmla.u32 rg, rg_dx, v_left_x; \
2403 subs span_num_blocks, span_num_blocks, #1; \
2405 mov dither_row, dither_row, ror dither_shift; \
2406 mov b_dx4, b_dx, lsl #2; \
2408 vdup.u32 dither_offsets, dither_row; \
2409 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2411 vdup.u32 b_block, b; \
2412 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2414 mov b_dx8, b_dx, lsl #3; \
2415 vdup.u32 r_block, rg[0]; \
2416 vdup.u32 g_block, rg[1]; \
2418 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2420 vadd.u32 r_block, r_block, block_span; \
2421 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2423 vadd.u32 g_block, g_block, block_span; \
2424 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2426 vadd.u32 b_block, b_block, block_span; \
2427 add block_ptr_b, block_ptr_a, #16; \
2429 vshrn.u32 r_whole_low, r_block, #16; \
2430 vshrn.u32 g_whole_low, g_block, #16; \
2431 vshrn.u32 b_whole_low, b_block, #16; \
2432 vdup.u32 dx4, rg_dx4[0]; \
2434 vaddhn.u32 r_whole_high, r_block, dx4; \
2435 vdup.u32 dx4, rg_dx4[1]; \
2437 vaddhn.u32 g_whole_high, g_block, dx4; \
2438 vdup.u32 dx4, b_dx4; \
2440 vaddhn.u32 b_whole_high, b_block, dx4; \
2441 vdup.u32 dx8, rg_dx8[0]; \
2443 vadd.u32 r_block, r_block, dx8; \
2444 vdup.u32 dx8, rg_dx8[1]; \
2446 vadd.u32 g_block, g_block, dx8; \
2447 vdup.u32 dx8, b_dx8; \
2449 vadd.u32 b_block, b_block, dx8; \
2451 vmovn.u16 r_whole_8, r_whole; \
2452 vmovn.u16 g_whole_8, g_whole; \
2453 vmovn.u16 b_whole_8, b_whole; \
2458 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2459 vshrn.u32 r_whole_low, r_block, #16; \
2461 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2462 vshrn.u32 g_whole_low, g_block, #16; \
2464 vshrn.u32 b_whole_low, b_block, #16; \
2466 vdup.u32 dx4, rg_dx4[0]; \
2467 vshr.u8 r_whole_8, r_whole_8, #3; \
2468 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2470 vaddhn.u32 r_whole_high, r_block, dx4; \
2471 vdup.u32 dx4, rg_dx4[1]; \
2473 vmov pixels, msb_mask; \
2474 vaddhn.u32 g_whole_high, g_block, dx4; \
2475 vdup.u32 dx4, b_dx4; \
2477 vaddhn.u32 b_whole_high, b_block, dx4; \
2478 vdup.u32 dx8, rg_dx8[0]; \
2480 vmlal.u8 pixels, r_whole_8, d64_1; \
2481 vmlal.u8 pixels, g_whole_8, d64_4; \
2482 vmlal.u8 pixels, b_whole_8, d64_128; \
2484 vadd.u32 r_block, r_block, dx8; \
2485 vdup.u32 dx8, rg_dx8[1]; \
2487 vadd.u32 g_block, g_block, dx8; \
2488 vdup.u32 dx8, b_dx8; \
2490 vadd.u32 b_block, b_block, dx8; \
2492 vmovn.u16 r_whole_8, r_whole; \
2493 vmovn.u16 g_whole_8, g_whole; \
2494 vmovn.u16 b_whole_8, b_whole; \
2496 vst1.u32 { pixels }, [ fb_ptr ]!; \
2497 subs span_num_blocks, span_num_blocks, #1; \
2501 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2503 ldrb right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2504 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2506 vshr.u8 r_whole_8, r_whole_8, #3; \
2507 vmov pixels, msb_mask; \
2508 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2509 eor right_mask, right_mask, #0xFF; \
2511 vmlal.u8 pixels, r_whole_8, d64_1; \
2512 vmlal.u8 pixels, g_whole_8, d64_4; \
2513 vmlal.u8 pixels, b_whole_8, d64_128; \
2516 vst1.u16 { pixels_low[0] }, [ fb_ptr ]!; \
2517 vext.16 pixels, pixels, #1; \
2518 movs right_mask, right_mask, lsr #1; \
2522 add span_uvrg_offset, span_uvrg_offset, #16; \
2523 add span_b_offset, span_b_offset, #4; \
2525 add span_edge_data, span_edge_data, #8; \
2526 subs num_spans, num_spans, #1; \
2530 ldmia sp!, { r4 - r11, pc } \
2532 setup_blocks_shaded_untextured_direct_builder(undithered)
2533 setup_blocks_shaded_untextured_direct_builder(dithered)
2542 #define block_ptr r1
2543 #define num_blocks r2
2556 #define texture_ptr r11
2570 #define pixels_d r10
2574 #define clut_ptr r12
2575 #define current_texture_mask r5
2576 #define dirty_textures_mask r6
2580 #define clut_low_a d2
2581 #define clut_low_b d3
2582 #define clut_high_a d4
2583 #define clut_high_b d5
2588 #define texels_low d6
2589 #define texels_high d7
2593 function(texture_blocks_untextured)
2599 function(texture_blocks_4bpp)
2600 stmdb sp!, { r3 - r11, r14 }
2601 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2603 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2604 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2606 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2607 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]
2609 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2610 vuzp.u8 clut_a, clut_b
2612 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
2613 tst dirty_textures_mask, current_texture_mask
2619 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2621 uxtah uv_0, texture_ptr, uv_01
2622 uxtah uv_1, texture_ptr, uv_01, ror #16
2624 uxtah uv_2, texture_ptr, uv_23
2625 uxtah uv_3, texture_ptr, uv_23, ror #16
2627 uxtah uv_4, texture_ptr, uv_45
2628 ldrb pixel_0, [ uv_0 ]
2630 uxtah uv_5, texture_ptr, uv_45, ror #16
2631 ldrb pixel_1, [ uv_1 ]
2633 uxtah uv_6, texture_ptr, uv_67
2634 ldrb pixel_2, [ uv_2 ]
2636 uxtah uv_7, texture_ptr, uv_67, ror #16
2637 ldrb pixel_3, [ uv_3 ]
2639 ldrb pixel_4, [ uv_4 ]
2640 subs num_blocks, num_blocks, #1
2642 ldrb pixel_5, [ uv_5 ]
2643 orr pixels_a, pixel_0, pixel_1, lsl #8
2645 ldrb pixel_6, [ uv_6 ]
2646 orr pixels_b, pixel_4, pixel_5, lsl #8
2648 ldrb pixel_7, [ uv_7 ]
2649 orr pixels_a, pixels_a, pixel_2, lsl #16
2651 orr pixels_b, pixels_b, pixel_6, lsl #16
2652 orr pixels_a, pixels_a, pixel_3, lsl #24
2654 orr pixels_b, pixels_b, pixel_7, lsl #24
2655 vmov.u32 texels, pixels_a, pixels_b
2657 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2658 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2660 vst2.u8 { texels_low, texels_high }, [ block_ptr, :128 ], c_64
2663 ldmia sp!, { r3 - r11, pc }
2666 stmdb sp!, { r1 - r2 }
2667 bl update_texture_4bpp_cache
2670 ldmia sp!, { r1 - r2 }
2676 function(texture_blocks_8bpp)
2677 stmdb sp!, { r3 - r11, r14 }
2678 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2680 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2681 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2683 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2684 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2686 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]
2687 tst dirty_textures_mask, current_texture_mask
2693 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2695 uxtah uv_0, texture_ptr, uv_01
2696 uxtah uv_1, texture_ptr, uv_01, ror #16
2698 uxtah uv_2, texture_ptr, uv_23
2699 uxtah uv_3, texture_ptr, uv_23, ror #16
2701 uxtah uv_4, texture_ptr, uv_45
2702 ldrb pixel_0, [ uv_0 ]
2704 uxtah uv_5, texture_ptr, uv_45, ror #16
2705 ldrb pixel_1, [ uv_1 ]
2707 uxtah uv_6, texture_ptr, uv_67
2708 ldrb pixel_2, [ uv_2 ]
2710 uxtah uv_7, texture_ptr, uv_67, ror #16
2711 ldrb pixel_3, [ uv_3 ]
2713 ldrb pixel_4, [ uv_4 ]
2714 add pixel_0, pixel_0, pixel_0
2716 ldrb pixel_5, [ uv_5 ]
2717 add pixel_1, pixel_1, pixel_1
2719 ldrb pixel_6, [ uv_6 ]
2720 add pixel_2, pixel_2, pixel_2
2722 ldrb pixel_7, [ uv_7 ]
2723 add pixel_3, pixel_3, pixel_3
2725 ldrh pixel_0, [ clut_ptr, pixel_0 ]
2726 add pixel_4, pixel_4, pixel_4
2728 ldrh pixel_1, [ clut_ptr, pixel_1 ]
2729 add pixel_5, pixel_5, pixel_5
2731 ldrh pixel_2, [ clut_ptr, pixel_2 ]
2732 add pixel_6, pixel_6, pixel_6
2734 ldrh pixel_3, [ clut_ptr, pixel_3 ]
2735 add pixel_7, pixel_7, pixel_7
2737 ldrh pixel_4, [ clut_ptr, pixel_4 ]
2738 orr pixels_a, pixel_0, pixel_1, lsl #16
2740 ldrh pixel_5, [ clut_ptr, pixel_5 ]
2741 orr pixels_c, pixel_2, pixel_3, lsl #16
2743 ldrh pixel_6, [ clut_ptr, pixel_6 ]
2744 subs num_blocks, num_blocks, #1
2746 ldrh pixel_7, [ clut_ptr, pixel_7 ]
2747 orr pixels_b, pixel_4, pixel_5, lsl #16
2749 orr pixels_d, pixel_6, pixel_7, lsl #16
2750 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2752 add block_ptr, block_ptr, #64
2755 ldmia sp!, { r3 - r11, pc }
2758 stmdb sp!, { r1 - r2, r12 }
2760 bl update_texture_8bpp_cache
2762 ldmia sp!, { r1 - r2, r12 }
2792 #define block_ptr r1
2793 #define num_blocks r2
2837 #define texture_ptr r12
2842 function(texture_blocks_16bpp)
2843 stmdb sp!, { r3 - r11, r14 }
2844 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2846 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2847 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2850 ldrh uv_0, [ block_ptr ]
2851 subs num_blocks, num_blocks, #1
2853 ldrh uv_1, [ block_ptr, #2 ]
2855 and v_0, uv_0, #0xFF00
2856 and v_1, uv_1, #0xFF00
2858 and u_0, uv_0, #0xFF
2859 and u_1, uv_1, #0xFF
2861 add uv_0, u_0, v_0, lsl #2
2862 ldrh uv_2, [ block_ptr, #4 ]
2864 add uv_1, u_1, v_1, lsl #2
2865 ldrh uv_3, [ block_ptr, #6 ]
2867 add uv_0, uv_0, uv_0
2868 add uv_1, uv_1, uv_1
2870 and v_2, uv_2, #0xFF00
2871 and v_3, uv_3, #0xFF00
2873 and u_2, uv_2, #0xFF
2874 and u_3, uv_3, #0xFF
2876 add uv_2, u_2, v_2, lsl #2
2877 ldrh uv_4, [ block_ptr, #8 ]
2879 add uv_3, u_3, v_3, lsl #2
2880 ldrh uv_5, [ block_ptr, #10 ]
2882 add uv_2, uv_2, uv_2
2883 add uv_3, uv_3, uv_3
2885 and v_4, uv_4, #0xFF00
2886 and v_5, uv_5, #0xFF00
2888 and u_4, uv_4, #0xFF
2889 and u_5, uv_5, #0xFF
2891 add uv_4, u_4, v_4, lsl #2
2892 ldrh uv_6, [ block_ptr, #12 ]
2894 add uv_5, u_5, v_5, lsl #2
2895 ldrh uv_7, [ block_ptr, #14 ]
2897 add uv_4, uv_4, uv_4
2898 ldrh pixel_0, [ texture_ptr, uv_0 ]
2900 add uv_5, uv_5, uv_5
2901 ldrh pixel_1, [ texture_ptr, uv_1 ]
2903 and v_6, uv_6, #0xFF00
2904 ldrh pixel_2, [ texture_ptr, uv_2 ]
2906 and v_7, uv_7, #0xFF00
2907 ldrh pixel_3, [ texture_ptr, uv_3 ]
2909 and u_6, uv_6, #0xFF
2910 ldrh pixel_4, [ texture_ptr, uv_4 ]
2912 and u_7, uv_7, #0xFF
2913 ldrh pixel_5, [ texture_ptr, uv_5 ]
2915 add uv_6, u_6, v_6, lsl #2
2916 add uv_7, u_7, v_7, lsl #2
2918 add uv_6, uv_6, uv_6
2919 add uv_7, uv_7, uv_7
2921 orr pixels_a, pixel_0, pixel_1, lsl #16
2922 orr pixels_b, pixel_2, pixel_3, lsl #16
2924 ldrh pixel_6, [ texture_ptr, uv_6 ]
2925 orr pixels_c, pixel_4, pixel_5, lsl #16
2927 ldrh pixel_7, [ texture_ptr, uv_7 ]
2928 orr pixels_d, pixel_6, pixel_7, lsl #16
2930 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
2931 add block_ptr, block_ptr, #64
2935 ldmia sp!, { r3 - r11, pc }
2950 #undef msb_mask_high
2958 #define num_blocks r1
2959 #define color_ptr r2
2960 #define mask_msb_ptr r2
2962 #define block_ptr_load_a r0
2963 #define block_ptr_store r3
2964 #define block_ptr_load_b r12
2969 #define draw_mask_bits_scalar r5
2971 #define d128_0x07 q0
2972 #define d128_0x1F q1
2973 #define d128_0x8000 q2
2974 #define test_mask q3
2976 #define colors_rg q5
2977 #define colors_b_dm_bits q6
2978 #define texels_rg q7
2981 #define pixels_b q10
2983 #define zero_mask q4
2984 #define draw_mask q12
2985 #define msb_mask q13
2987 #define fb_pixels q8
2989 #define pixels_gb_low q9
2991 #define colors_r d10
2992 #define colors_g d11
2993 #define colors_b d12
2994 #define draw_mask_bits d13
2995 #define texels_r d14
2996 #define texels_g d15
2997 #define pixels_r_low d16
2998 #define pixels_g_low d18
2999 #define pixels_b_low d19
3000 #define msb_mask_low d26
3001 #define msb_mask_high d27
3006 #define texels_b d31
3008 #define shade_blocks_textured_modulated_prologue_indirect() \
3010 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3012 #define shade_blocks_textured_modulated_prologue_direct() \
3013 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3014 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ] \
3016 #define shade_blocks_textured_modulated_prologue_shaded() \
3018 #define shade_blocks_textured_modulated_prologue_unshaded() \
3019 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3020 vld1.u32 { colors_r[] }, [ color_ptr, :32 ]; \
3021 vdup.u8 colors_g, colors_r[1]; \
3022 vdup.u8 colors_b, colors_r[2]; \
3023 vdup.u8 colors_r, colors_r[0] \
3026 #define shade_blocks_textured_modulated_load_dithered(target) \
3027 vld1.u32 { target }, [ block_ptr_load_b, :128 ] \
3029 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3030 vld1.u32 { target }, [ block_ptr_load_b, :128 ], c_32 \
3032 #define shade_blocks_textured_modulated_load_undithered(target) \
3034 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3035 add block_ptr_load_b, block_ptr_load_b, #32 \
3037 #define shade_blocks_textured_modulate_dithered(channel) \
3038 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3040 #define shade_blocks_textured_modulate_undithered(channel) \
3041 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3044 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3045 vst1.u32 { draw_mask }, [ block_ptr_store, :128 ]! \
3047 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3048 ldr fb_ptr, [ block_ptr_load_b, #(offset - 64) ]; \
3049 vld1.u32 { fb_pixels }, [ fb_ptr ]; \
3050 vbit.u16 pixels, fb_pixels, draw_mask \
3052 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3053 vst1.u32 { pixels }, [ block_ptr_store, :128 ], c_48 \
3055 #define shade_blocks_textured_modulated_store_pixels_direct() \
3056 vst1.u32 { pixels }, [ fb_ptr ] \
3059 #define shade_blocks_textured_modulated_load_rg_shaded() \
3060 vld1.u32 { colors_r, colors_g }, [ block_ptr_load_b, :128 ], c_32 \
3062 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3063 add block_ptr_load_b, block_ptr_load_b, #32 \
3065 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3066 vld1.u32 { colors_b, draw_mask_bits }, [ block_ptr_load_a, :128 ], c_32 \
3068 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3069 ldr draw_mask_bits_scalar, [ block_ptr_load_a, #8 ]; \
3070 add block_ptr_load_a, block_ptr_load_a, #32 \
3072 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3073 vdup.u16 draw_mask, draw_mask_bits[0] \
3075 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3076 vdup.u16 draw_mask, draw_mask_bits_scalar \
3079 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3081 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3082 vorr.u16 pixels, pixels, msb_mask \
3085 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3088 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3089 stmdb sp!, { r4 - r5, lr }; \
3090 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3092 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
3094 shade_blocks_textured_modulated_prologue_##target(); \
3095 shade_blocks_textured_modulated_prologue_##shading(); \
3097 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3100 add block_ptr_load_b, block_ptr_load_a, #16; \
3101 vmov.u8 d64_1, #1; \
3102 vmov.u8 d64_4, #4; \
3103 vmov.u8 d64_128, #128; \
3105 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3106 vmov.u8 d128_0x07, #0x07; \
3108 shade_blocks_textured_modulated_load_rg_##shading(); \
3109 vmov.u8 d128_0x1F, #0x1F; \
3111 shade_blocks_textured_modulated_load_bdm_##shading(); \
3112 vmov.u16 d128_0x8000, #0x8000; \
3114 vmovn.u16 texels_r, texels; \
3115 vshrn.u16 texels_g, texels, #5; \
3117 vshrn.u16 texels_b, texels, #7; \
3118 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3120 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3121 vtst.u16 draw_mask, draw_mask, test_mask; \
3123 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3124 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3126 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3127 vshr.u8 texels_b, texels_b, #3; \
3129 shade_blocks_textured_modulate_##dithering(r); \
3130 shade_blocks_textured_modulate_##dithering(g); \
3131 shade_blocks_textured_modulate_##dithering(b); \
3133 vand.u16 pixels, texels, d128_0x8000; \
3134 vceq.u16 zero_mask, texels, #0; \
3136 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3137 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3138 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3140 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3141 vorr.u16 draw_mask, draw_mask, zero_mask; \
3142 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3143 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3145 subs num_blocks, num_blocks, #1; \
3151 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3152 shade_blocks_textured_modulated_load_rg_##shading(); \
3153 vshrn.u16 texels_g, texels, #5; \
3155 shade_blocks_textured_modulated_load_bdm_##shading(); \
3156 vshrn.u16 texels_b, texels, #7; \
3158 vmovn.u16 texels_r, texels; \
3159 vmlal.u8 pixels, pixels_r_low, d64_1; \
3161 vmlal.u8 pixels, pixels_g_low, d64_4; \
3162 vmlal.u8 pixels, pixels_b_low, d64_128; \
3163 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3165 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3166 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3168 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3169 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3171 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3172 vtst.u16 draw_mask, draw_mask, test_mask; \
3174 shade_blocks_textured_modulated_store_pixels_##target(); \
3175 vshr.u8 texels_b, texels_b, #3; \
3177 shade_blocks_textured_modulate_##dithering(r); \
3178 shade_blocks_textured_modulate_##dithering(g); \
3179 shade_blocks_textured_modulate_##dithering(b); \
3181 vand.u16 pixels, texels, d128_0x8000; \
3182 vceq.u16 zero_mask, texels, #0; \
3184 subs num_blocks, num_blocks, #1; \
3186 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3187 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3188 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3190 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3191 vorr.u16 draw_mask, draw_mask, zero_mask; \
3192 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3193 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3198 vmlal.u8 pixels, pixels_r_low, d64_1; \
3199 vmlal.u8 pixels, pixels_g_low, d64_4; \
3200 vmlal.u8 pixels, pixels_b_low, d64_128; \
3202 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3203 shade_blocks_textured_modulated_store_pixels_##target(); \
3205 ldmia sp!, { r4 - r5, pc } \
3208 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3209 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3210 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3211 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3213 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3214 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3215 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3216 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3234 #undef msb_mask_high
3237 #define num_blocks r1
3238 #define mask_msb_ptr r2
3239 #define color_ptr r3
3241 #define block_ptr_load r0
3242 #define draw_mask_store_ptr r3
3243 #define draw_mask_bits_ptr r12
3244 #define draw_mask_ptr r12
3245 #define pixel_store_ptr r14
3247 #define fb_ptr_cmp r4
3250 #define fb_ptr_next r14
3254 #define test_mask q0
3256 #define draw_mask q2
3257 #define zero_mask q3
3258 #define draw_mask_combined q4
3259 #define fb_pixels q5
3260 #define fb_pixels_next q6
3263 #define draw_mask_low d4
3264 #define draw_mask_high d5
3265 #define msb_mask_low d14
3266 #define msb_mask_high d15
3269 function(shade_blocks_textured_unmodulated_indirect)
3270 str r14, [ sp, #-4 ]
3271 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3273 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3274 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3276 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3277 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3280 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3282 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3283 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3284 [ draw_mask_bits_ptr, :16 ], c_64
3285 vceq.u16 zero_mask, pixels, #0
3287 vtst.u16 draw_mask, draw_mask, test_mask
3288 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3290 subs num_blocks, num_blocks, #1
3294 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3295 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3297 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3298 [ draw_mask_bits_ptr, :16 ], c_64
3299 vceq.u16 zero_mask, pixels, #0
3301 vtst.u16 draw_mask, draw_mask, test_mask
3302 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3304 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3305 subs num_blocks, num_blocks, #1
3310 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3311 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3318 function(shade_blocks_textured_unmodulated_direct)
3319 stmdb sp!, { r4, r14 }
3320 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3322 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3323 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3325 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3328 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3329 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3331 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3332 [ draw_mask_bits_ptr, :16 ], c_64
3333 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3335 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3336 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3337 vceq.u16 zero_mask, pixels, #0
3338 vtst.u16 draw_mask, draw_mask, test_mask
3340 subs num_blocks, num_blocks, #1
3344 mov fb_ptr, fb_ptr_next
3345 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3347 vorr.u16 pixels, pixels, msb_mask
3349 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3350 vmov fb_pixels, fb_pixels_next
3352 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3353 [ draw_mask_bits_ptr, :16 ], c_64
3354 vbif.u16 fb_pixels, pixels, draw_mask_combined
3356 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3358 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3359 add fb_ptr_cmp, fb_ptr_cmp, #14
3363 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3364 vceq.u16 zero_mask, pixels, #0
3366 vst1.u16 { fb_pixels }, [ fb_ptr ]
3367 vtst.u16 draw_mask, draw_mask, test_mask
3370 subs num_blocks, num_blocks, #1
3374 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3375 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3377 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3379 ldmia sp!, { r4, pc }
3382 vst1.u16 { fb_pixels }, [ fb_ptr ]
3383 vceq.u16 zero_mask, pixels, #0
3385 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3386 vtst.u16 draw_mask, draw_mask, test_mask
3391 function(shade_blocks_unshaded_untextured_indirect)
3396 function(shade_blocks_unshaded_untextured_direct)
3397 stmdb sp!, { r4, r14 }
3398 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3400 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3401 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3403 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3404 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3406 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3407 vld1.u16 { pixels }, [ color_ptr, :128 ]
3410 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3412 vorr.u16 pixels, pixels, msb_mask
3413 subs num_blocks, num_blocks, #1
3415 ldr fb_ptr_next, [ block_ptr_load ], #64
3417 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3421 vmov fb_pixels, fb_pixels_next
3422 mov fb_ptr, fb_ptr_next
3423 ldr fb_ptr_next, [ block_ptr_load ], #64
3425 vbif.u16 fb_pixels, pixels, draw_mask
3426 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3428 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3429 add fb_ptr_cmp, fb_ptr_cmp, #14
3433 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3434 vst1.u16 { fb_pixels }, [ fb_ptr ]
3437 subs num_blocks, num_blocks, #1
3441 vbif.u16 fb_pixels_next, pixels, draw_mask
3442 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3444 ldmia sp!, { r4, pc }
3447 vst1.u16 { fb_pixels }, [ fb_ptr ]
3448 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3452 #undef draw_mask_ptr
3459 #define num_blocks r1
3460 #define msb_mask_ptr r2
3461 #define pixel_ptr r3
3462 #define draw_mask_ptr r0
3465 #define fb_ptr_next r14
3466 #define fb_ptr_cmp r4
3474 #undef msb_mask_high
3475 #undef draw_mask_next
3478 #undef fb_pixels_next
3481 #define draw_mask q1
3483 #define fb_pixels q3
3484 #define blend_pixels q4
3485 #define pixels_no_msb q5
3486 #define blend_mask q6
3487 #define fb_pixels_no_msb q7
3488 #define d128_0x8000 q8
3489 #define d128_0x0421 q9
3490 #define fb_pixels_next q10
3491 #define blend_pixels_next q11
3492 #define pixels_next q12
3493 #define draw_mask_next q13
3494 #define write_mask q14
3496 #define pixels_rb q5
3497 #define pixels_mg q7
3499 #define d128_0x7C1F q8
3500 #define d128_0x03E0 q9
3501 #define fb_pixels_rb q10
3502 #define fb_pixels_g q11
3503 #define fb_pixels_masked q11
3504 #define d128_0x83E0 q15
3505 #define pixels_fourth q7
3506 #define d128_0x1C07 q12
3507 #define d128_0x00E0 q13
3508 #define d128_0x80E0 q13
3510 #define msb_mask_low d0
3511 #define msb_mask_high d1
3513 #define blend_blocks_average_set_blend_mask_textured(source) \
3514 vclt.s16 blend_mask, source, #0 \
3516 #define blend_blocks_average_set_stp_bit_textured() \
3517 vorr.u16 blend_pixels, #0x8000 \
3519 #define blend_blocks_average_combine_textured(source) \
3520 vbif.u16 blend_pixels, source, blend_mask \
3522 #define blend_blocks_average_set_blend_mask_untextured(source) \
3524 #define blend_blocks_average_set_stp_bit_untextured() \
3526 #define blend_blocks_average_combine_untextured(source) \
3528 #define blend_blocks_average_mask_set_on() \
3529 vclt.s16 write_mask, fb_pixels_next, #0 \
3531 #define blend_blocks_average_mask_copy_on() \
3532 vorr.u16 draw_mask, draw_mask_next, write_mask \
3534 #define blend_blocks_average_mask_copy_b_on() \
3535 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3537 #define blend_blocks_average_mask_set_off() \
3539 #define blend_blocks_average_mask_copy_off() \
3540 vmov draw_mask, draw_mask_next \
3542 #define blend_blocks_average_mask_copy_b_off() \
3544 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3547 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3548 stmdb sp!, { r4, r14 }; \
3549 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3550 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3552 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3553 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3555 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3558 vmov.u16 d128_0x8000, #0x8000; \
3559 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3560 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3562 vmov.u16 d128_0x0421, #0x0400; \
3563 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3565 vorr.u16 d128_0x0421, #0x0021; \
3566 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3568 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3569 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3570 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3571 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3572 blend_blocks_average_mask_set_##mask_evaluate(); \
3573 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3575 subs num_blocks, num_blocks, #1; \
3579 mov fb_ptr, fb_ptr_next; \
3580 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3582 vmov pixels, pixels_next; \
3583 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3585 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3587 blend_blocks_average_mask_copy_##mask_evaluate(); \
3588 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3590 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3591 blend_blocks_average_set_stp_bit_##texturing(); \
3592 vmov fb_pixels, fb_pixels_next; \
3593 blend_blocks_average_combine_##texturing(pixels); \
3595 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3596 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3597 cmp fb_ptr_cmp, #28; \
3600 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3601 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3603 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3604 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3606 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3607 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3609 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3610 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3611 blend_blocks_average_mask_set_##mask_evaluate(); \
3612 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3615 subs num_blocks, num_blocks, #1; \
3619 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3620 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3622 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3623 blend_blocks_average_set_stp_bit_##texturing(); \
3624 blend_blocks_average_combine_##texturing(pixels_next); \
3626 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3627 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3628 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3630 ldmia sp!, { r4, pc }; \
3633 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3634 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3635 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3637 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3638 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3639 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3640 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3641 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3642 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3646 blend_blocks_average_builder(textured, off)
3647 blend_blocks_average_builder(untextured, off)
3648 blend_blocks_average_builder(textured, on)
3649 blend_blocks_average_builder(untextured, on)
3652 #define blend_blocks_add_mask_set_on() \
3653 vclt.s16 write_mask, fb_pixels, #0 \
3655 #define blend_blocks_add_mask_copy_on() \
3656 vorr.u16 draw_mask, draw_mask, write_mask \
3658 #define blend_blocks_add_mask_set_off() \
3660 #define blend_blocks_add_mask_copy_off() \
3663 #define blend_blocks_add_textured_builder(mask_evaluate) \
3666 function(blend_blocks_textured_add_##mask_evaluate) \
3667 stmdb sp!, { r4, r14 }; \
3668 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3669 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3671 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3672 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3674 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3677 vmov.u16 d128_0x7C1F, #0x7C00; \
3678 vmov.u16 d128_0x03E0, #0x0300; \
3679 vmov.u16 d128_0x83E0, #0x8000; \
3680 vorr.u16 d128_0x03E0, #0x00E0; \
3681 vorr.u16 d128_0x7C1F, #0x001F; \
3682 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3684 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3685 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3686 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3687 vclt.s16 blend_mask, pixels, #0; \
3688 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3689 blend_blocks_add_mask_set_##mask_evaluate(); \
3690 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3692 blend_blocks_add_mask_copy_##mask_evaluate(); \
3693 vorr.u16 pixels, pixels, msb_mask; \
3694 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3695 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3696 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3697 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3698 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3699 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3700 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3701 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3703 subs num_blocks, num_blocks, #1; \
3707 mov fb_ptr, fb_ptr_next; \
3709 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3711 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3712 vclt.s16 blend_mask, pixels, #0; \
3714 vorr.u16 pixels, pixels, msb_mask; \
3715 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3716 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3718 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3719 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3721 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3722 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3723 cmp fb_ptr_cmp, #28; \
3726 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3727 blend_blocks_add_mask_set_##mask_evaluate(); \
3728 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3729 blend_blocks_add_mask_copy_##mask_evaluate(); \
3730 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3731 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3732 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3735 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3736 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3737 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3738 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3739 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3741 subs num_blocks, num_blocks, #1; \
3745 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3746 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3747 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3749 ldmia sp!, { r4, pc }; \
3752 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3753 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3755 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3756 blend_blocks_add_mask_set_##mask_evaluate(); \
3757 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3758 blend_blocks_add_mask_copy_##mask_evaluate(); \
3759 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3763 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3766 function(blend_blocks_untextured_add_##mask_evaluate) \
3767 stmdb sp!, { r4, r14 }; \
3768 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3769 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3771 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3772 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3774 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3777 vmov.u16 d128_0x7C1F, #0x7C00; \
3778 vmov.u16 d128_0x03E0, #0x0300; \
3779 vorr.u16 d128_0x7C1F, #0x001F; \
3780 vorr.u16 d128_0x03E0, #0x00E0; \
3782 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3783 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3784 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3785 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3786 blend_blocks_add_mask_set_##mask_evaluate(); \
3787 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3789 blend_blocks_add_mask_copy_##mask_evaluate(); \
3790 vand.u16 pixels_g, pixels, d128_0x03E0; \
3791 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3792 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3793 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3794 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3795 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3796 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3798 subs num_blocks, num_blocks, #1; \
3802 mov fb_ptr, fb_ptr_next; \
3804 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3806 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3808 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3809 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3810 vand.u16 pixels_g, pixels, d128_0x03E0; \
3812 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3813 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3815 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3816 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3817 cmp fb_ptr_cmp, #28; \
3820 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3821 blend_blocks_add_mask_set_##mask_evaluate(); \
3822 blend_blocks_add_mask_copy_##mask_evaluate(); \
3823 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3824 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3825 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3828 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3829 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3830 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3831 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3832 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3834 subs num_blocks, num_blocks, #1; \
3838 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3839 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3840 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3841 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3843 ldmia sp!, { r4, pc }; \
3846 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3847 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3849 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3850 blend_blocks_add_mask_set_##mask_evaluate(); \
3851 blend_blocks_add_mask_copy_##mask_evaluate(); \
3852 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3856 blend_blocks_add_textured_builder(off)
3857 blend_blocks_add_textured_builder(on)
3858 blend_blocks_add_untextured_builder(off)
3859 blend_blocks_add_untextured_builder(on)
3861 #define blend_blocks_subtract_set_blend_mask_textured() \
3862 vclt.s16 blend_mask, pixels_next, #0 \
3864 #define blend_blocks_subtract_combine_textured() \
3865 vbif.u16 blend_pixels, pixels, blend_mask \
3867 #define blend_blocks_subtract_set_stb_textured() \
3868 vorr.u16 blend_pixels, #0x8000 \
3870 #define blend_blocks_subtract_msb_mask_textured() \
3871 vorr.u16 pixels, pixels_next, msb_mask \
3873 #define blend_blocks_subtract_set_blend_mask_untextured() \
3875 #define blend_blocks_subtract_combine_untextured() \
3877 #define blend_blocks_subtract_set_stb_untextured() \
3878 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3880 #define blend_blocks_subtract_msb_mask_untextured() \
3883 #define blend_blocks_subtract_mask_set_on() \
3884 vclt.s16 write_mask, fb_pixels, #0 \
3886 #define blend_blocks_subtract_mask_copy_on() \
3887 vorr.u16 draw_mask, draw_mask_next, write_mask \
3889 #define blend_blocks_subtract_mask_set_off() \
3891 #define blend_blocks_subtract_mask_copy_off() \
3892 vmov draw_mask, draw_mask_next \
3895 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
3898 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
3899 stmdb sp!, { r4, r14 }; \
3900 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3901 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3903 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3904 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3906 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3909 vmov.u16 d128_0x7C1F, #0x7C00; \
3910 vmov.u16 d128_0x03E0, #0x0300; \
3911 vorr.u16 d128_0x7C1F, #0x001F; \
3912 vorr.u16 d128_0x03E0, #0x00E0; \
3914 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3915 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3916 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3917 blend_blocks_subtract_set_blend_mask_##texturing(); \
3918 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3919 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3920 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3922 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3923 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3924 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3925 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3926 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3928 subs num_blocks, num_blocks, #1; \
3932 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
3933 mov fb_ptr, fb_ptr_next; \
3934 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3936 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3937 blend_blocks_subtract_msb_mask_##texturing(); \
3939 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3940 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3941 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3942 blend_blocks_subtract_set_stb_##texturing(); \
3943 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3944 blend_blocks_subtract_combine_##texturing(); \
3945 blend_blocks_subtract_set_blend_mask_##texturing(); \
3946 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3948 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3949 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3950 cmp fb_ptr_cmp, #28; \
3953 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3954 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3955 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3956 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3957 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3958 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3959 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3962 subs num_blocks, num_blocks, #1; \
3966 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
3968 blend_blocks_subtract_msb_mask_##texturing(); \
3969 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3970 blend_blocks_subtract_set_stb_##texturing(); \
3971 blend_blocks_subtract_combine_##texturing(); \
3972 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3973 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3975 ldmia sp!, { r4, pc }; \
3978 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3979 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3980 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3981 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3982 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3983 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3984 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3988 blend_blocks_subtract_builder(textured, off)
3989 blend_blocks_subtract_builder(textured, on)
3990 blend_blocks_subtract_builder(untextured, off)
3991 blend_blocks_subtract_builder(untextured, on)
3994 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
3997 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
3998 stmdb sp!, { r4, r14 }; \
3999 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4000 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4002 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4003 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4005 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4008 vmov.u16 d128_0x7C1F, #0x7C00; \
4009 vmov.u16 d128_0x03E0, #0x0300; \
4010 vmov.u16 d128_0x83E0, #0x8300; \
4011 vmov.u16 d128_0x1C07, #0x1C00; \
4012 vmov.u16 d128_0x80E0, #0x8000; \
4013 vorr.u16 d128_0x7C1F, #0x001F; \
4014 vorr.u16 d128_0x03E0, #0x00E0; \
4015 vorr.u16 d128_0x83E0, #0x00E0; \
4016 vorr.u16 d128_0x1C07, #0x0007; \
4017 vorr.u16 d128_0x80E0, #0x00E0; \
4019 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4020 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4021 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4022 vclt.s16 blend_mask, pixels, #0; \
4023 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4024 blend_blocks_add_mask_set_##mask_evaluate(); \
4025 vshr.s16 pixels_fourth, pixels, #2; \
4027 blend_blocks_add_mask_copy_##mask_evaluate(); \
4028 vorr.u16 pixels, pixels, msb_mask; \
4029 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4030 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
4031 vand.u16 pixels_mg, pixels_fourth, d128_0x80E0; \
4032 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
4033 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
4034 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4035 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
4036 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4037 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
4039 subs num_blocks, num_blocks, #1; \
4043 mov fb_ptr, fb_ptr_next; \
4045 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4047 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4048 vclt.s16 blend_mask, pixels, #0; \
4050 vshr.s16 pixels_fourth, pixels, #2; \
4051 vorr.u16 pixels, pixels, msb_mask; \
4052 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4053 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4055 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4056 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4058 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4059 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4060 cmp fb_ptr_cmp, #28; \
4063 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4064 blend_blocks_add_mask_set_##mask_evaluate(); \
4065 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
4066 blend_blocks_add_mask_copy_##mask_evaluate(); \
4067 vand.u16 pixels_mg, pixels_fourth, d128_0x80E0; \
4068 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
4069 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4072 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
4073 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4074 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
4075 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4076 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
4078 subs num_blocks, num_blocks, #1; \
4082 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4083 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4084 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4086 ldmia sp!, { r4, pc }; \
4089 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4090 vand.u16 pixels_mg, pixels_fourth, d128_0x80E0; \
4092 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4093 blend_blocks_add_mask_set_##mask_evaluate(); \
4094 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
4095 blend_blocks_add_mask_copy_##mask_evaluate(); \
4096 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
4100 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4103 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4104 stmdb sp!, { r4, r14 }; \
4105 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4106 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4108 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4109 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4111 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4114 vmov.u16 d128_0x7C1F, #0x7C00; \
4115 vmov.u16 d128_0x03E0, #0x0300; \
4116 vmov.u16 d128_0x83E0, #0x8300; \
4117 vmov.u16 d128_0x1C07, #0x1C00; \
4118 vmov.u16 d128_0x00E0, #0x00E0; \
4119 vorr.u16 d128_0x7C1F, #0x001F; \
4120 vorr.u16 d128_0x03E0, #0x00E0; \
4121 vorr.u16 d128_0x83E0, #0x00E0; \
4122 vorr.u16 d128_0x1C07, #0x0007; \
4124 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4125 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4126 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4127 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4128 blend_blocks_add_mask_set_##mask_evaluate(); \
4129 vshr.s16 pixels_fourth, pixels, #2; \
4130 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4132 blend_blocks_add_mask_copy_##mask_evaluate(); \
4133 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4134 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4135 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4136 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4137 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4138 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4139 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4141 subs num_blocks, num_blocks, #1; \
4145 mov fb_ptr, fb_ptr_next; \
4147 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4149 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4151 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4152 vshr.s16 pixels_fourth, pixels, #2; \
4153 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4154 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4156 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4157 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4159 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4160 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4161 cmp fb_ptr_cmp, #28; \
4164 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4165 blend_blocks_add_mask_set_##mask_evaluate(); \
4166 blend_blocks_add_mask_copy_##mask_evaluate(); \
4167 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4168 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4169 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4172 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4173 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4174 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4175 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4176 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4178 subs num_blocks, num_blocks, #1; \
4182 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4183 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4184 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4185 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4187 ldmia sp!, { r4, pc }; \
4190 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4191 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4193 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4194 blend_blocks_add_mask_set_##mask_evaluate(); \
4195 blend_blocks_add_mask_copy_##mask_evaluate(); \
4196 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4200 blend_blocks_add_fourth_textured_builder(off)
4201 blend_blocks_add_fourth_textured_builder(on)
4202 blend_blocks_add_fourth_untextured_builder(off)
4203 blend_blocks_add_fourth_untextured_builder(on)
4205 // TODO: Optimize this more. Need a scene that actually uses it for
4210 function(blend_blocks_textured_unblended_on)
4211 stmdb sp!, { r4, r14 }
4212 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4213 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
4215 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4216 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
4218 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4221 ldr fb_ptr, [ pixel_ptr, #28 ]
4222 vld1.u16 { fb_pixels }, [ fb_ptr ]
4223 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4224 vclt.s16 write_mask, fb_pixels, #0
4225 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4227 subs num_blocks, num_blocks, #1
4231 vorr.u16 draw_mask, draw_mask, write_mask
4232 vbif.u16 fb_pixels, pixels, draw_mask
4233 vst1.u16 { fb_pixels }, [ fb_ptr ]
4235 ldr fb_ptr, [ pixel_ptr, #28 ]
4236 vld1.u16 { fb_pixels }, [ fb_ptr ]
4237 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4238 vclt.s16 write_mask, fb_pixels, #0
4239 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4241 subs num_blocks, num_blocks, #1
4245 vorr.u16 draw_mask, draw_mask, write_mask
4246 vbif.u16 fb_pixels, pixels, draw_mask
4247 vst1.u16 { fb_pixels }, [ fb_ptr ]
4249 ldmia sp!, { r4, pc }
4252 function(blend_blocks_textured_unblended_off)
4262 vld1.u32 { u_whole_8, v_whole_8 }, [ r1, :128 ], r3
4282 #define parameter_width_offset 0
4283 #define parameter_height_offset 4
4289 #define left_unaligned r14
4290 #define right_unaligned r4
4292 #define num_unaligned r2
4293 #define num_width r6
4301 function(render_block_fill_body)
4302 ldr vram_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
4303 ldr height, [ sp, #parameter_height_offset ]
4305 add vram_ptr, vram_ptr, y, lsl #11
4306 ldr width, [ sp, #parameter_width_offset ]
4308 add vram_ptr, vram_ptr, x, lsl #1
4309 stmdb sp!, { r4 - r6, r14 }
4311 ubfx color_r, color, #3, #5
4312 ubfx color_g, color, #11, #5
4314 ubfx color_b, color, #19, #5
4315 orr color, color_r, color_g, lsl #5
4317 orr color, color, color_b, lsl #10
4318 add left_unaligned, x, #0x7
4320 bic left_unaligned, left_unaligned, #0x7
4321 vdup.u16 colors, color
4323 sub left_unaligned, left_unaligned, x
4326 sub pitch, pitch, width, lsl #1
4327 sub width, width, left_unaligned
4329 and right_unaligned, width, #0x7
4330 bic width, width, #0x7
4333 mov num_width, width, lsr #3
4335 movs num_unaligned, left_unaligned
4339 strh color, [ vram_ptr ], #2
4341 subs num_unaligned, num_unaligned, #1
4345 vst1.u32 { colors }, [ vram_ptr, :128 ]!
4346 subs num_width, num_width, #1
4349 movs num_unaligned, right_unaligned
4353 strh color, [ vram_ptr ], #2
4355 subs num_unaligned, num_unaligned, #1
4359 add vram_ptr, vram_ptr, pitch
4360 subs height, height, #1
4363 ldmia sp!, { r4 - r6, pc }
4374 #undef dirty_textures_mask
4376 #undef current_texture_mask
4387 #define offset_u_right r10
4388 #define width_rounded r11
4389 #define height_rounded r12
4391 #define texture_offset_base r1
4392 #define tile_width r2
4393 #define tile_height r3
4394 #define num_blocks r4
4396 #define sub_tile_height r6
4398 #define texture_mask r8
4399 #define column_data r9
4400 #define texture_offset r10
4401 #define tiles_remaining r11
4402 #define fb_ptr_advance_column r12
4403 #define texture_block_ptr r14
4405 #define texture_page_ptr r3
4406 #define left_block_mask r4
4407 #define right_block_mask r5
4408 #define texture_mask_rev r10
4409 #define control_mask r11
4411 #define dirty_textures_mask r4
4413 #define current_texture_mask r6
4427 #define draw_masks_fb_ptrs q1
4429 #define draw_mask_fb_ptr_left d2
4430 #define draw_mask_fb_ptr_right d3
4432 #define clut_low_a d4
4433 #define clut_low_b d5
4434 #define clut_high_a d6
4435 #define clut_high_b d7
4437 #define block_masks d8
4438 #define block_masks_shifted d9
4443 #define texels_low d10
4444 #define texels_high d11
4447 setup_sprite_flush_blocks_single:
4450 stmdb sp!, { r0 - r3, r12, r14 }
4451 bl flush_render_block_buffer
4452 ldmia sp!, { r0 - r3, r12, r14 }
4456 add block, psx_gpu, #psx_gpu_blocks_offset
4458 mov num_blocks, sub_tile_height
4462 setup_sprite_flush_blocks_double:
4465 stmdb sp!, { r0 - r3, r12, r14 }
4466 bl flush_render_block_buffer
4467 ldmia sp!, { r0 - r3, r12, r14 }
4471 add block, psx_gpu, #psx_gpu_blocks_offset
4473 mov num_blocks, sub_tile_height, lsl #1
4477 setup_sprite_update_texture_4bpp_cache:
4478 stmdb sp!, { r0 - r3, r14 }
4479 bl update_texture_4bpp_cache
4480 ldmia sp!, { r0 - r3, pc }
4483 setup_sprite_update_texture_8bpp_cache:
4484 stmdb sp!, { r0 - r3, r14 }
4485 bl update_texture_8bpp_cache
4486 ldmia sp!, { r0 - r3, pc }
4489 #define setup_sprite_tiled_initialize_4bpp() \
4490 ldr dirty_textures_mask, \
4491 [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]; \
4492 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4494 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4495 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4497 tst current_texture_mask, dirty_textures_mask; \
4498 vuzp.u8 clut_a, clut_b; \
4500 blne setup_sprite_update_texture_4bpp_cache \
4502 #define setup_sprite_tiled_initialize_8bpp() \
4503 ldr dirty_textures_mask, \
4504 [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]; \
4505 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4507 tst current_texture_mask, dirty_textures_mask; \
4508 blne setup_sprite_update_texture_8bpp_cache \
4511 #define setup_sprite_tile_setup_block_no(side, offset, texture_mode) \
4513 #define setup_sprite_block_count_single() \
4516 #define setup_sprite_block_count_double() \
4517 sub_tile_height, lsl #1 \
4519 #define setup_sprite_tile_add_blocks(type) \
4520 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4521 cmp num_blocks, #MAX_BLOCKS; \
4523 blgt setup_sprite_flush_blocks_##type \
4526 #define setup_sprite_tile_full_4bpp(edge) \
4527 setup_sprite_tile_add_blocks(double); \
4530 and texture_block_ptr, texture_offset, texture_mask; \
4531 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4534 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4535 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4537 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4538 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4540 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4541 add texture_block_ptr, texture_offset, #8; \
4543 and texture_block_ptr, texture_block_ptr, texture_mask; \
4544 add block, block, #40; \
4546 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4547 add fb_ptr, fb_ptr, #16; \
4549 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4550 add block, block, #24; \
4552 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4553 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4556 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4557 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4559 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4560 add block, block, #40; \
4562 add texture_offset, texture_offset, #0x10; \
4563 add fb_ptr, fb_ptr, #(2048 - 16); \
4565 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4566 add block, block, #24; \
4568 subs sub_tile_height, sub_tile_height, #1; \
4571 add texture_offset, texture_offset, #0xF00; \
4572 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4575 #define setup_sprite_tile_half_4bpp(edge) \
4576 setup_sprite_tile_add_blocks(single); \
4579 and texture_block_ptr, texture_offset, texture_mask; \
4580 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4583 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4584 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4586 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4587 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4589 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4590 add block, block, #40; \
4592 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4593 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4595 add block, block, #24; \
4596 add texture_offset, texture_offset, #0x10; \
4598 add fb_ptr, fb_ptr, #2048; \
4599 subs sub_tile_height, sub_tile_height, #1; \
4603 add texture_offset, texture_offset, #0xF00; \
4604 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4607 #define setup_sprite_tile_full_8bpp(edge) \
4608 setup_sprite_tile_add_blocks(double); \
4609 add block, block, #16; \
4612 and texture_block_ptr, texture_offset, texture_mask; \
4613 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4616 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4617 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4619 add texture_block_ptr, texture_offset, #8; \
4620 vst1.u32 { texels }, [ block, :64 ]; \
4622 and texture_block_ptr, texture_block_ptr, texture_mask; \
4623 add block, block, #24; \
4625 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4627 add fb_ptr, fb_ptr, #16; \
4628 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4630 add block, block, #40; \
4631 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4634 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4635 vst1.u32 { texels }, [ block, :64 ]; \
4636 add block, block, #24; \
4638 add texture_offset, texture_offset, #0x10; \
4639 add fb_ptr, fb_ptr, #(2048 - 16); \
4641 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4642 add block, block, #40; \
4644 subs sub_tile_height, sub_tile_height, #1; \
4647 sub block, block, #16; \
4648 add texture_offset, texture_offset, #0xF00; \
4649 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4652 #define setup_sprite_tile_half_8bpp(edge) \
4653 setup_sprite_tile_add_blocks(single); \
4654 add block, block, #16; \
4657 and texture_block_ptr, texture_offset, texture_mask; \
4658 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4661 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4662 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4664 vst1.u32 { texels }, [ block, :64 ]; \
4665 add block, block, #24; \
4667 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4668 add block, block, #40; \
4670 add texture_offset, texture_offset, #0x10; \
4671 add fb_ptr, fb_ptr, #2048; \
4673 subs sub_tile_height, sub_tile_height, #1; \
4676 sub block, block, #16; \
4677 add texture_offset, texture_offset, #0xF00; \
4678 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4681 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4682 add texture_offset, texture_offset_base, #8; \
4683 add fb_ptr, fb_ptr, #16 \
4685 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4686 mov texture_offset, texture_offset_base \
4688 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4689 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4691 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4692 mov texture_offset, texture_offset_base \
4694 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4695 sub fb_ptr, fb_ptr, #16 \
4697 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4699 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4700 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4702 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4705 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode) \
4706 mov sub_tile_height, column_data; \
4707 setup_sprite_tile_column_edge_pre_adjust_##edge_mode(edge); \
4708 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4709 setup_sprite_tile_column_edge_post_adjust_##edge_mode(edge) \
4711 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode) \
4712 and sub_tile_height, column_data, #0xFF; \
4713 mov tiles_remaining, column_data, lsr #16; \
4714 setup_sprite_tile_column_edge_pre_adjust_##edge_mode(edge); \
4715 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4717 subs tiles_remaining, tiles_remaining, #1; \
4721 mov sub_tile_height, #16; \
4722 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4723 subs tiles_remaining, tiles_remaining, #1; \
4727 uxtb sub_tile_height, column_data, ror #8; \
4728 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4729 setup_sprite_tile_column_edge_post_adjust_##edge_mode(edge) \
4732 #define setup_sprite_column_data_single() \
4733 mov column_data, height; \
4734 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] \
4736 #define setup_sprite_column_data_multi() \
4737 and height_rounded, height_rounded, #0xF; \
4738 rsb column_data, offset_v, #16; \
4740 add height_rounded, height_rounded, #1; \
4741 sub tile_height, tile_height, #1; \
4743 orr column_data, column_data, tile_height, lsl #16; \
4744 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]; \
4746 orr column_data, column_data, height_rounded, lsl #8 \
4748 #define setup_sprite_tile_column_width_single(texture_mode, multi_height, \
4750 setup_sprite_##texture_mode##_single_##multi_height##_##edge_mode##_##edge: \
4751 setup_sprite_column_data_##multi_height(); \
4752 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4753 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4754 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4755 vdup.u8 draw_mask_fb_ptr_right, block_masks[1]; \
4757 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, \
4759 ldmia sp!, { r4 - r11, pc } \
4761 #define setup_sprite_tiled_advance_column() \
4762 add texture_offset_base, texture_offset_base, #0x100; \
4763 tst texture_offset_base, #0xF00; \
4764 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4766 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4768 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode: \
4769 setup_sprite_column_data_##multi_height(); \
4770 mov fb_ptr_advance_column, #32; \
4772 sub fb_ptr_advance_column, height, lsl #11; \
4773 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4775 vdup.u8 draw_mask_fb_ptr_right, block_masks[1]; \
4776 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm); \
4778 subs tile_width, tile_width, #2; \
4779 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4781 vmov.u8 draw_masks_fb_ptrs, #0; \
4785 setup_sprite_tiled_advance_column(); \
4786 setup_sprite_tile_column_height_##multi_height(full, none, tm); \
4787 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4788 subs tile_width, tile_width, #1; \
4792 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4793 vdup.u8 draw_mask_fb_ptr_right, block_masks[5]; \
4795 setup_sprite_tiled_advance_column(); \
4796 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm); \
4797 ldmia sp!, { r4 - r11, pc } \
4805 // [ sp + 4 ]: width
4806 // [ sp + 8 ]: height
4807 // [ sp + 12 ]: color (unused)
4809 #define setup_sprite_tiled_builder(texture_mode) \
4811 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full); \
4812 setup_sprite_tile_column_width_single(texture_mode, multi, full, none); \
4813 setup_sprite_tile_column_width_multi(texture_mode, single, full, full); \
4814 setup_sprite_tile_column_width_single(texture_mode, single, full, none); \
4815 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full); \
4816 setup_sprite_tile_column_width_single(texture_mode, multi, half, right); \
4817 setup_sprite_tile_column_width_multi(texture_mode, single, half, full); \
4818 setup_sprite_tile_column_width_single(texture_mode, single, half, right); \
4819 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half); \
4820 setup_sprite_tile_column_width_single(texture_mode, multi, half, left); \
4821 setup_sprite_tile_column_width_multi(texture_mode, single, full, half); \
4822 setup_sprite_tile_column_width_single(texture_mode, single, half, left); \
4823 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half); \
4824 setup_sprite_tile_column_width_multi(texture_mode, single, half, half); \
4828 function(setup_sprite_##texture_mode) \
4829 stmdb sp!, { r4 - r11, r14 }; \
4830 setup_sprite_tiled_initialize_##texture_mode(); \
4832 ldr v, [ sp, #36 ]; \
4833 and offset_u, u, #0xF; \
4835 ldr width, [ sp, #40 ]; \
4836 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
4838 ldr height, [ sp, #44 ]; \
4839 add fb_ptr, fb_ptr, y, lsl #11; \
4841 add fb_ptr, fb_ptr, x, lsl #1; \
4842 and offset_v, v, #0xF; \
4844 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4845 add width_rounded, offset_u, width; \
4847 add height_rounded, offset_v, height; \
4848 add width_rounded, width_rounded, #15; \
4850 add height_rounded, height_rounded, #15; \
4851 mov tile_width, width_rounded, lsr #4; \
4853 /* texture_offset_base = VH-VL-00-00 */\
4854 mov texture_offset_base, v, lsl #8; \
4855 and offset_u_right, width_rounded, #0xF; \
4857 /* texture_offset_base = VH-UH-UL-00 */\
4858 bfi texture_offset_base, u, #4, #8; \
4859 movw right_block_mask, #0xFFFE; \
4861 /* texture_offset_base = VH-UH-VL-00 */\
4862 bfi texture_offset_base, v, #4, #4; \
4863 movw left_block_mask, #0xFFFF; \
4865 mov tile_height, height_rounded, lsr #4; \
4866 mvn left_block_mask, left_block_mask, lsl offset_u; \
4868 /* texture_mask = HH-HL-WH-WL */\
4869 ldrh texture_mask, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]; \
4870 mov right_block_mask, right_block_mask, lsl offset_u_right; \
4872 /* texture_mask_rev = WH-WL-HH-HL */\
4873 rev16 texture_mask_rev, texture_mask; \
4874 vmov block_masks, left_block_mask, right_block_mask; \
4876 /* texture_mask = HH-HL-HL-WL */\
4877 bfi texture_mask, texture_mask_rev, #4, #4; \
4878 /* texture_mask_rev = 00-00-00-WH */\
4879 mov texture_mask_rev, texture_mask_rev, lsr #12; \
4881 /* texture_mask = HH-WH-HL-WL */\
4882 bfi texture_mask, texture_mask_rev, #8, #4; \
4883 and left_block_mask, left_block_mask, #0xFF; \
4885 mov control_mask, #0; \
4886 cmp left_block_mask, #0xFF; \
4888 uxtb right_block_mask, right_block_mask, ror #8; \
4889 orreq control_mask, control_mask, #0x4; \
4891 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4892 cmp right_block_mask, #0xFF; \
4894 orreq control_mask, control_mask, #0x8; \
4895 cmp tile_width, #1; \
4897 add block, psx_gpu, #psx_gpu_blocks_offset; \
4898 orreq control_mask, control_mask, #0x1; \
4900 cmp tile_height, #1; \
4901 add block, block, num_blocks, lsl #6; \
4903 orreq control_mask, control_mask, #0x2; \
4904 ldr pc, [ pc, control_mask, lsl #2 ]; \
4907 .word setup_sprite_##texture_mode##_multi_multi_full_full; \
4908 .word setup_sprite_##texture_mode##_single_multi_full_none; \
4909 .word setup_sprite_##texture_mode##_multi_single_full_full; \
4910 .word setup_sprite_##texture_mode##_single_single_full_none; \
4911 .word setup_sprite_##texture_mode##_multi_multi_half_full; \
4912 .word setup_sprite_##texture_mode##_single_multi_half_right; \
4913 .word setup_sprite_##texture_mode##_multi_single_half_full; \
4914 .word setup_sprite_##texture_mode##_single_single_half_right; \
4915 .word setup_sprite_##texture_mode##_multi_multi_full_half; \
4916 .word setup_sprite_##texture_mode##_single_multi_half_left; \
4917 .word setup_sprite_##texture_mode##_multi_single_full_half; \
4918 .word setup_sprite_##texture_mode##_single_single_half_left; \
4919 .word setup_sprite_##texture_mode##_multi_multi_half_half; \
4921 .word setup_sprite_##texture_mode##_multi_single_half_half \
4924 setup_sprite_tiled_builder(4bpp);
4925 setup_sprite_tiled_builder(8bpp);
4933 #define block_ptr r0
4934 #define num_blocks r1
4936 #define texel_shift_mask r3
4937 #define block_pixels_a r4
4938 #define block_pixels_b r5
4947 #define texels_01 r6
4948 #define texels_23 r7
4949 #define texels_45 r8
4950 #define texels_67 r9
4952 function(texture_sprite_blocks_8bpp)
4953 stmdb sp!, { r4 - r11, r14 }
4954 movw texel_shift_mask, #(0xFF << 1)
4956 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
4957 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
4959 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
4960 ldr block_pixels_a, [ block_ptr, #16 ]
4963 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
4964 ldr block_pixels_b, [ block_ptr, #20 ]
4966 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
4967 ldrh texel_0, [ clut_ptr, texel_0 ]
4969 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
4970 ldrh texel_1, [ clut_ptr, texel_1 ]
4972 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
4973 ldr block_pixels_a, [ block_ptr, #(64 + 16) ]
4975 ldrh texel_2, [ clut_ptr, texel_2 ]
4976 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
4978 ldrh texel_3, [ clut_ptr, texel_3 ]
4979 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
4981 ldrh texel_4, [ clut_ptr, texel_4 ]
4982 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
4984 ldrh texel_5, [ clut_ptr, texel_5 ]
4985 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
4987 ldrh texel_6, [ clut_ptr, texel_6 ]
4988 orr texels_01, texel_0, texel_1, lsl #16
4990 ldrh texel_7, [ clut_ptr, texel_7 ]
4991 orr texels_23, texel_2, texel_3, lsl #16
4993 orr texels_45, texel_4, texel_5, lsl #16
4994 str texels_01, [ block_ptr, #0 ]
4996 orr texels_67, texel_6, texel_7, lsl #16
4997 str texels_23, [ block_ptr, #4 ]
4999 subs num_blocks, num_blocks, #1
5000 str texels_45, [ block_ptr, #8 ]
5002 str texels_67, [ block_ptr, #12 ]
5003 add block_ptr, block_ptr, #64
5007 ldmia sp!, { r4 - r11, pc }
5010 #undef width_rounded
5013 #undef texture_offset
5022 #define left_offset r8
5023 #define width_rounded r9
5024 #define right_width r10
5025 #define block_width r11
5027 #define texture_offset_base r1
5028 #define texture_mask r2
5029 #define texture_page_ptr r3
5030 #define num_blocks r4
5033 #define texture_offset r8
5034 #define blocks_remaining r9
5035 #define fb_ptr_pitch r12
5036 #define texture_block_ptr r14
5038 #define texture_mask_width r2
5039 #define texture_mask_height r3
5040 #define left_mask_bits r4
5041 #define right_mask_bits r5
5045 #undef block_masks_shifted
5048 #define block_masks d0
5049 #define block_masks_shifted d1
5050 #define draw_mask_fb_ptr d2
5054 setup_sprites_16bpp_flush_single:
5057 stmdb sp!, { r0 - r3, r12, r14 }
5058 bl flush_render_block_buffer
5059 ldmia sp!, { r0 - r3, r12, r14 }
5063 add block, psx_gpu, #psx_gpu_blocks_offset
5068 setup_sprites_16bpp_flush_row:
5071 stmdb sp!, { r0 - r3, r12, r14 }
5072 bl flush_render_block_buffer
5073 ldmia sp!, { r0 - r3, r12, r14 }
5077 add block, psx_gpu, #psx_gpu_blocks_offset
5078 mov num_blocks, block_width
5082 function(setup_sprite_16bpp)
5083 stmdb sp!, { r4 - r11, r14 }
5084 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5087 add fb_ptr, fb_ptr, y, lsl #11
5089 ldr width, [ sp, #40 ]
5090 add fb_ptr, fb_ptr, x, lsl #1
5092 ldr height, [ sp, #44 ]
5093 and left_offset, u, #0x7
5095 add texture_offset_base, u, u
5096 add width_rounded, width, #7
5098 add texture_offset_base, v, lsl #11
5099 mov left_mask_bits, #0xFF
5101 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5102 add width_rounded, width_rounded, left_offset
5104 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5105 sub fb_ptr, fb_ptr, left_offset, lsl #1
5107 add texture_mask, texture_mask_width, texture_mask_width
5108 mov right_mask_bits, #0xFE
5110 and right_width, width_rounded, #0x7
5111 mvn left_mask_bits, left_mask_bits, lsl left_offset
5113 add texture_mask, texture_mask_height, lsl #11
5114 mov block_width, width_rounded, lsr #3
5116 mov right_mask_bits, right_mask_bits, lsl right_width
5117 movw fb_ptr_pitch, #(2048 + 16)
5119 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5120 vmov block_masks, left_mask_bits, right_mask_bits
5122 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5123 add block, psx_gpu, #psx_gpu_blocks_offset
5125 bic texture_offset_base, texture_offset_base, #0xF
5128 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5129 add block, block, num_blocks, lsl #6
5133 vext.32 block_masks_shifted, block_masks, block_masks, #1
5134 vorr.u32 block_masks, block_masks, block_masks_shifted
5135 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5138 add num_blocks, num_blocks, #1
5139 cmp num_blocks, #MAX_BLOCKS
5140 blgt setup_sprites_16bpp_flush_single
5142 and texture_block_ptr, texture_offset_base, texture_mask
5143 subs height, height, #1
5145 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5146 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5148 vst1.u32 { texels }, [ block, :128 ]
5149 add block, block, #40
5151 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5154 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5156 add block, block, #24
5157 add texture_offset_base, texture_offset_base, #2048
5158 add fb_ptr, fb_ptr, #2048
5159 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5162 ldmia sp!, { r4 - r11, pc }
5165 add num_blocks, num_blocks, block_width
5166 mov texture_offset, texture_offset_base
5168 cmp num_blocks, #MAX_BLOCKS
5169 blgt setup_sprites_16bpp_flush_row
5171 add texture_offset_base, texture_offset_base, #2048
5172 and texture_block_ptr, texture_offset, texture_mask
5174 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5175 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5177 vst1.u32 { texels }, [ block, :128 ]
5178 add block, block, #40
5180 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5181 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5184 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5185 subs blocks_remaining, block_width, #2
5187 add texture_offset, texture_offset, #16
5188 add fb_ptr, fb_ptr, #16
5190 vmov.u8 draw_mask_fb_ptr, #0
5192 add block, block, #24
5196 and texture_block_ptr, texture_offset, texture_mask
5197 subs blocks_remaining, blocks_remaining, #1
5199 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5200 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5202 vst1.u32 { texels }, [ block, :128 ]
5203 add block, block, #40
5205 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5208 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5210 add texture_offset, texture_offset, #16
5211 add fb_ptr, fb_ptr, #16
5213 add block, block, #24
5217 and texture_block_ptr, texture_offset, texture_mask
5218 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5220 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5221 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5223 vst1.u32 { texels }, [ block, :128 ]
5224 add block, block, #40
5226 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5227 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5229 add block, block, #24
5230 subs height, height, #1
5232 add fb_ptr, fb_ptr, fb_ptr_pitch
5233 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5237 ldmia sp!, { r4 - r11, pc }
5240 #undef texture_page_ptr
5242 #undef dirty_textures_mask
5243 #undef current_texture_mask
5246 #define current_texture_page r1
5247 #define texture_page_ptr r2
5248 #define vram_ptr_a r3
5249 #define current_texture_page_x r12
5250 #define current_texture_page_y r4
5251 #define dirty_textures_mask r5
5255 #define current_texture_mask r9
5257 #define vram_ptr_b r11
5259 #define texel_block_a d0
5260 #define texel_block_b d1
5261 #define texel_block_expanded_a q1
5262 #define texel_block_expanded_b q2
5263 #define texel_block_expanded_ab q2
5264 #define texel_block_expanded_c q3
5265 #define texel_block_expanded_d q4
5266 #define texel_block_expanded_cd q3
5268 function(update_texture_4bpp_cache)
5269 stmdb sp!, { r4 - r11, r14 }
5272 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5274 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5275 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5277 and current_texture_page_x, current_texture_page, #0xF
5278 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
5280 mov current_texture_page_y, current_texture_page, lsr #4
5281 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5283 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5286 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5287 bic dirty_textures_mask, current_texture_mask
5290 str dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5295 add vram_ptr_b, vram_ptr_a, #2048
5298 vld1.u32 { texel_block_a }, [ vram_ptr_a, :64 ], c_4096
5299 vld1.u32 { texel_block_b }, [ vram_ptr_b, :64 ], c_4096
5301 vmovl.u8 texel_block_expanded_a, texel_block_a
5302 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5303 vmovl.u8 texel_block_expanded_c, texel_block_b
5304 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5306 vbic.u16 texel_block_expanded_a, #0x00F0
5307 vbic.u16 texel_block_expanded_b, #0x00F0
5308 vbic.u16 texel_block_expanded_c, #0x00F0
5309 vbic.u16 texel_block_expanded_d, #0x00F0
5311 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
5312 texel_block_expanded_b
5313 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
5314 texel_block_expanded_d
5316 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
5317 [ texture_page_ptr, :256 ]!
5319 subs sub_y, sub_y, #1
5323 add vram_ptr_a, vram_ptr_a, #8
5324 add vram_ptr_b, vram_ptr_b, #8
5326 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5327 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5329 subs tile_x, tile_x, #1
5333 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5334 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5336 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5337 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5339 subs tile_y, tile_y, #1
5343 ldmia sp!, { r4 - r11, pc }
5346 #undef current_texture_page
5349 #define texture_page r1
5350 #define texture_page_ptr r2
5351 #define vram_ptr_a r3
5352 #define texture_page_x r12
5353 #define texture_page_y r4
5354 #define current_texture_page r5
5359 #define vram_ptr_b r11
5371 function(update_texture_8bpp_cache_slice)
5372 stmdb sp!, { r4 - r11, r14 }
5375 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5376 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5378 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5381 and texture_page_x, texture_page, #0xF
5382 mov texture_page_y, texture_page, lsr #4
5384 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
5387 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
5388 eor current_texture_page, current_texture_page, texture_page
5390 ands current_texture_page, current_texture_page, #0x1
5393 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
5396 add vram_ptr_b, vram_ptr_a, #2048
5399 vld1.u32 { texels_a }, [ vram_ptr_a, :128 ], c_4096
5400 vld1.u32 { texels_b }, [ vram_ptr_b, :128 ], c_4096
5401 vld1.u32 { texels_c }, [ vram_ptr_a, :128 ], c_4096
5402 vld1.u32 { texels_d }, [ vram_ptr_b, :128 ], c_4096
5404 vst1.u32 { texels_a, texels_b }, [ texture_page_ptr, :256 ]!
5405 vst1.u32 { texels_c, texels_d }, [ texture_page_ptr, :256 ]!
5407 subs sub_y, sub_y, #1
5412 add vram_ptr_a, vram_ptr_a, #16
5413 add vram_ptr_b, vram_ptr_b, #16
5415 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5416 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5418 subs tile_x, tile_x, #1
5423 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5424 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5426 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5427 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5429 subs tile_y, tile_y, #1
5430 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
5435 ldmia sp!, { r4 - r11, pc }