2 * Copyright (C) 2011 Gilead Kutnick "Exophase" <exophase@gmail.com>
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public License as
6 * published by the Free Software Foundation; either version 2 of
7 * the License, or (at your option) any later version.
9 * This program is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * General Public License for more details.
17 #define MAX_BLOCKS_PER_ROW 128
19 #define psx_gpu_test_mask_offset 0
20 #define psx_gpu_uvrg_offset 16
21 #define psx_gpu_uvrg_dx_offset 32
22 #define psx_gpu_uvrg_dy_offset 48
23 #define psx_gpu_u_block_span_offset 64
24 #define psx_gpu_v_block_span_offset 80
25 #define psx_gpu_r_block_span_offset 96
26 #define psx_gpu_g_block_span_offset 112
27 #define psx_gpu_b_block_span_offset 128
29 #define psx_gpu_b_dx_offset 132
31 #define psx_gpu_b_offset 144
32 #define psx_gpu_b_dy_offset 148
33 #define psx_gpu_triangle_area_offset 152
34 #define psx_gpu_texture_window_settings_offset 156
35 #define psx_gpu_current_texture_mask_offset 160
36 #define psx_gpu_viewport_mask_offset 164
37 #define psx_gpu_dirty_textures_4bpp_mask_offset 168
38 #define psx_gpu_dirty_textures_8bpp_mask_offset 172
39 #define psx_gpu_dirty_textures_8bpp_alternate_mask_offset 176
40 #define psx_gpu_triangle_color_offset 180
41 #define psx_gpu_dither_table_offset 184
42 #define psx_gpu_render_block_handler_offset 200
43 #define psx_gpu_texture_page_ptr_offset 204
44 #define psx_gpu_texture_page_base_offset 208
45 #define psx_gpu_clut_ptr_offset 212
46 #define psx_gpu_vram_ptr_offset 216
48 #define psx_gpu_render_state_base_offset 220
49 #define psx_gpu_render_state_offset 222
50 #define psx_gpu_num_spans_offset 224
51 #define psx_gpu_num_blocks_offset 226
52 #define psx_gpu_offset_x_offset 228
53 #define psx_gpu_offset_y_offset 230
54 #define psx_gpu_clut_settings_offset 232
55 #define psx_gpu_texture_settings_offset 234
56 #define psx_gpu_viewport_start_x_offset 236
57 #define psx_gpu_viewport_start_y_offset 238
58 #define psx_gpu_viewport_end_x_offset 240
59 #define psx_gpu_viewport_end_y_offset 242
60 #define psx_gpu_mask_msb_offset 244
62 #define psx_gpu_triangle_winding_offset 246
63 #define psx_gpu_display_area_draw_enable_offset 247
64 #define psx_gpu_current_texture_page_offset 248
65 #define psx_gpu_last_8bpp_texture_page_offset 249
66 #define psx_gpu_texture_mask_width_offset 250
67 #define psx_gpu_texture_mask_height_offset 251
68 #define psx_gpu_texture_window_x_offset 252
69 #define psx_gpu_texture_window_y_offset 253
70 #define psx_gpu_primitive_type_offset 254
72 #define psx_gpu_reserved_a_offset 255
74 #define psx_gpu_blocks_offset 0x0100
75 #define psx_gpu_span_uvrg_offset_offset 0x2100
76 #define psx_gpu_span_edge_data_offset 0x4100
77 #define psx_gpu_span_b_offset_offset 0x5100
79 #define edge_data_left_x_offset 0
80 #define edge_data_num_blocks_offset 2
81 #define edge_data_right_mask_offset 4
82 #define edge_data_y_offset 6
126 #define gw_by_l gw_bx_l
146 #define uvrg_xxxx0 q3
151 #define uvrg_xxxx1 q4
156 #define uvrg_xxxx2 q5
161 #define yyyy_uvrg0 q6
166 #define yyyy_uvrg1 q7
171 #define yyyy_uvrg2 q8
198 #define uvrg_base q11
244 #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
1958 orr color, color, lsl #16
1962 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1963 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1965 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
1967 cmp span_num_blocks, #0
1970 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1972 add fb_ptr, fb_ptr, y, lsl #11
1973 subs span_num_blocks, span_num_blocks, #1
1975 add fb_ptr, fb_ptr, left_x, lsl #1
1979 vst1.u32 { colors }, [ fb_ptr ]!
1980 subs span_num_blocks, span_num_blocks, #1
1985 ldrb right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1987 cmp right_mask, #0x0
1990 tst right_mask, #0xF
1991 streq color, [ fb_ptr ], #4
1992 moveq right_mask, right_mask, lsr #4
1993 streq color, [ fb_ptr ], #4
1995 tst right_mask, #0x3
1996 streq color, [ fb_ptr ], #4
1997 moveq right_mask, right_mask, lsr #2
1999 tst right_mask, #0x1
2000 streqh color, [ fb_ptr ]
2003 add span_edge_data, span_edge_data, #8
2004 subs num_spans, num_spans, #1
2007 ldmia sp!, { r4 - r11, pc }
2010 vst1.u32 { colors }, [ fb_ptr ]
2017 #define rg_dx_ptr r2
2035 #undef dither_offsets
2055 #define r_whole_low d6
2056 #define r_whole_high d7
2057 #define g_whole_low d8
2058 #define g_whole_high d9
2059 #define b_whole_low d10
2060 #define b_whole_high d11
2062 #define gb_whole_8 q6
2064 #define g_whole_8 d12
2065 #define b_whole_8 d13
2067 #define r_whole_8 d14
2079 #define block_span q5
2087 #define d128_0x7 q13
2091 #define dither_offsets q14
2092 #define draw_mask q15
2094 #define dither_offsets_low d28
2097 #define test_mask q10
2100 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2101 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2102 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2104 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2105 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2106 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2108 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2110 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2113 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2116 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2117 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2118 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2120 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2122 cmp num_spans, #0; \
2125 stmdb sp!, { r4 - r11, r14 }; \
2126 vshl.u32 rg_dx4, rg_dx, #2; \
2128 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2129 vshl.u32 rg_dx8, rg_dx, #3; \
2131 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2133 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2134 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2136 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2137 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2139 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2140 vmov.u8 d64_1, #1; \
2142 vmov.u8 d128_4, #4; \
2143 vmov.u8 d64_128, #128; \
2145 vmov.u8 d128_0x7, #0x7; \
2148 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2149 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2151 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2152 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
2154 cmp span_num_blocks, #0; \
2157 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2158 add num_blocks, span_num_blocks, num_blocks; \
2160 cmp num_blocks, #MAX_BLOCKS; \
2164 ldr b, [ span_b_offset ]; \
2165 add fb_ptr, fb_ptr, y, lsl #11; \
2167 vdup.u32 v_left_x, left_x; \
2170 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2171 add fb_ptr, fb_ptr, left_x, lsl #1; \
2173 mla b, b_dx, left_x, b; \
2174 and dither_shift, left_x, #0x03; \
2176 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2177 vshr.u32 rg_dx, rg_dx4, #2; \
2179 mov dither_shift, dither_shift, lsl #3; \
2180 vmla.u32 rg, rg_dx, v_left_x; \
2183 subs span_num_blocks, span_num_blocks, #1; \
2185 mov dither_row, dither_row, ror dither_shift; \
2186 mov b_dx4, b_dx, lsl #2; \
2188 vdup.u32 dither_offsets, dither_row; \
2189 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2191 vdup.u32 b_block, b; \
2192 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2194 mov b_dx8, b_dx, lsl #3; \
2195 vdup.u32 r_block, rg[0]; \
2196 vdup.u32 g_block, rg[1]; \
2198 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2200 vadd.u32 r_block, r_block, block_span; \
2201 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2203 vadd.u32 g_block, g_block, block_span; \
2204 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2206 vadd.u32 b_block, b_block, block_span; \
2207 add block_ptr_b, block_ptr_a, #16; \
2209 vshrn.u32 r_whole_low, r_block, #16; \
2210 vshrn.u32 g_whole_low, g_block, #16; \
2211 vshrn.u32 b_whole_low, b_block, #16; \
2212 vdup.u32 dx4, rg_dx4[0]; \
2214 vaddhn.u32 r_whole_high, r_block, dx4; \
2215 vdup.u32 dx4, rg_dx4[1]; \
2217 vaddhn.u32 g_whole_high, g_block, dx4; \
2218 vdup.u32 dx4, b_dx4; \
2220 vaddhn.u32 b_whole_high, b_block, dx4; \
2221 vdup.u32 dx8, rg_dx8[0]; \
2223 vadd.u32 r_block, r_block, dx8; \
2224 vdup.u32 dx8, rg_dx8[1]; \
2226 vadd.u32 g_block, g_block, dx8; \
2227 vdup.u32 dx8, b_dx8; \
2229 vadd.u32 b_block, b_block, dx8; \
2231 vmovn.u16 r_whole_8, r_whole; \
2232 vmovn.u16 g_whole_8, g_whole; \
2233 vmovn.u16 b_whole_8, b_whole; \
2236 veor.u32 draw_mask, draw_mask, draw_mask; \
2239 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2240 vshrn.u32 r_whole_low, r_block, #16; \
2242 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2243 vshrn.u32 g_whole_low, g_block, #16; \
2245 vshrn.u32 b_whole_low, b_block, #16; \
2246 str fb_ptr, [ block_ptr_a, #44 ]; \
2248 vdup.u32 dx4, rg_dx4[0]; \
2249 vshr.u8 r_whole_8, r_whole_8, #3; \
2250 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2252 vaddhn.u32 r_whole_high, r_block, dx4; \
2253 vdup.u32 dx4, rg_dx4[1]; \
2255 vaddhn.u32 g_whole_high, g_block, dx4; \
2256 vdup.u32 dx4, b_dx4; \
2258 vaddhn.u32 b_whole_high, b_block, dx4; \
2259 vdup.u32 dx8, rg_dx8[0]; \
2261 vmull.u8 pixels, r_whole_8, d64_1; \
2262 vmlal.u8 pixels, g_whole_8, d64_4; \
2263 vmlal.u8 pixels, b_whole_8, d64_128; \
2265 vadd.u32 r_block, r_block, dx8; \
2266 vdup.u32 dx8, rg_dx8[1]; \
2268 vadd.u32 g_block, g_block, dx8; \
2269 vdup.u32 dx8, b_dx8; \
2271 vadd.u32 b_block, b_block, dx8; \
2272 add fb_ptr, fb_ptr, #16; \
2274 vmovn.u16 r_whole_8, r_whole; \
2275 vmovn.u16 g_whole_8, g_whole; \
2276 vmovn.u16 b_whole_8, b_whole; \
2278 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2279 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2283 subs span_num_blocks, span_num_blocks, #1; \
2287 str fb_ptr, [ block_ptr_a, #44 ]; \
2288 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2290 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2291 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2293 vshr.u8 r_whole_8, r_whole_8, #3; \
2294 vdup.u8 draw_mask, right_mask; \
2296 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2297 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
2299 vtst.u16 draw_mask, draw_mask, test_mask; \
2301 vmull.u8 pixels, r_whole_8, d64_1; \
2302 vmlal.u8 pixels, g_whole_8, d64_4; \
2303 vmlal.u8 pixels, b_whole_8, d64_128; \
2305 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2306 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2309 add span_uvrg_offset, span_uvrg_offset, #16; \
2310 add span_b_offset, span_b_offset, #4; \
2312 add span_edge_data, span_edge_data, #8; \
2313 subs num_spans, num_spans, #1; \
2315 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2318 ldmia sp!, { r4 - r11, pc }; \
2321 /* TODO: Load from psx_gpu instead of saving/restoring these */\
2324 stmdb sp!, { r0 - r3, r12, r14 }; \
2325 bl flush_render_block_buffer; \
2326 ldmia sp!, { r0 - r3, r12, r14 }; \
2330 vmov.u8 d64_1, #1; \
2331 vmov.u8 d128_4, #4; \
2332 vmov.u8 d64_128, #128; \
2333 vmov.u8 d128_0x7, #0x7; \
2335 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2337 mov num_blocks, span_num_blocks; \
2338 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2342 setup_blocks_shaded_untextured_indirect_builder(undithered)
2343 setup_blocks_shaded_untextured_indirect_builder(dithered)
2348 #define mask_msb_ptr r14
2350 #define draw_mask q0
2351 #define pixels_low d16
2352 #define pixels_high d17
2356 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2359 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2360 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2361 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2363 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2365 cmp num_spans, #0; \
2368 stmdb sp!, { r4 - r11, r14 }; \
2369 vshl.u32 rg_dx4, rg_dx, #2; \
2371 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2372 vshl.u32 rg_dx8, rg_dx, #3; \
2374 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2375 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2377 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2378 vmov.u8 d64_1, #1; \
2380 vmov.u8 d128_4, #4; \
2381 vmov.u8 d64_128, #128; \
2383 vmov.u8 d128_0x7, #0x7; \
2384 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2385 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
2388 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2389 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2391 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2392 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
2394 cmp span_num_blocks, #0; \
2397 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2398 add fb_ptr, fb_ptr, y, lsl #11; \
2400 ldr b, [ span_b_offset ]; \
2401 vdup.u32 v_left_x, left_x; \
2404 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2405 add fb_ptr, fb_ptr, left_x, lsl #1; \
2407 mla b, b_dx, left_x, b; \
2408 and dither_shift, left_x, #0x03; \
2410 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2411 vshr.u32 rg_dx, rg_dx4, #2; \
2413 mov dither_shift, dither_shift, lsl #3; \
2414 vmla.u32 rg, rg_dx, v_left_x; \
2416 subs span_num_blocks, span_num_blocks, #1; \
2418 mov dither_row, dither_row, ror dither_shift; \
2419 mov b_dx4, b_dx, lsl #2; \
2421 vdup.u32 dither_offsets, dither_row; \
2422 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2424 vdup.u32 b_block, b; \
2425 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2427 mov b_dx8, b_dx, lsl #3; \
2428 vdup.u32 r_block, rg[0]; \
2429 vdup.u32 g_block, rg[1]; \
2431 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2433 vadd.u32 r_block, r_block, block_span; \
2434 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2436 vadd.u32 g_block, g_block, block_span; \
2437 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2439 vadd.u32 b_block, b_block, block_span; \
2440 add block_ptr_b, block_ptr_a, #16; \
2442 vshrn.u32 r_whole_low, r_block, #16; \
2443 vshrn.u32 g_whole_low, g_block, #16; \
2444 vshrn.u32 b_whole_low, b_block, #16; \
2445 vdup.u32 dx4, rg_dx4[0]; \
2447 vaddhn.u32 r_whole_high, r_block, dx4; \
2448 vdup.u32 dx4, rg_dx4[1]; \
2450 vaddhn.u32 g_whole_high, g_block, dx4; \
2451 vdup.u32 dx4, b_dx4; \
2453 vaddhn.u32 b_whole_high, b_block, dx4; \
2454 vdup.u32 dx8, rg_dx8[0]; \
2456 vadd.u32 r_block, r_block, dx8; \
2457 vdup.u32 dx8, rg_dx8[1]; \
2459 vadd.u32 g_block, g_block, dx8; \
2460 vdup.u32 dx8, b_dx8; \
2462 vadd.u32 b_block, b_block, dx8; \
2464 vmovn.u16 r_whole_8, r_whole; \
2465 vmovn.u16 g_whole_8, g_whole; \
2466 vmovn.u16 b_whole_8, b_whole; \
2471 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2472 vshrn.u32 r_whole_low, r_block, #16; \
2474 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2475 vshrn.u32 g_whole_low, g_block, #16; \
2477 vshrn.u32 b_whole_low, b_block, #16; \
2479 vdup.u32 dx4, rg_dx4[0]; \
2480 vshr.u8 r_whole_8, r_whole_8, #3; \
2481 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2483 vaddhn.u32 r_whole_high, r_block, dx4; \
2484 vdup.u32 dx4, rg_dx4[1]; \
2486 vmov pixels, msb_mask; \
2487 vaddhn.u32 g_whole_high, g_block, dx4; \
2488 vdup.u32 dx4, b_dx4; \
2490 vaddhn.u32 b_whole_high, b_block, dx4; \
2491 vdup.u32 dx8, rg_dx8[0]; \
2493 vmlal.u8 pixels, r_whole_8, d64_1; \
2494 vmlal.u8 pixels, g_whole_8, d64_4; \
2495 vmlal.u8 pixels, b_whole_8, d64_128; \
2497 vadd.u32 r_block, r_block, dx8; \
2498 vdup.u32 dx8, rg_dx8[1]; \
2500 vadd.u32 g_block, g_block, dx8; \
2501 vdup.u32 dx8, b_dx8; \
2503 vadd.u32 b_block, b_block, dx8; \
2505 vmovn.u16 r_whole_8, r_whole; \
2506 vmovn.u16 g_whole_8, g_whole; \
2507 vmovn.u16 b_whole_8, b_whole; \
2509 vst1.u32 { pixels }, [ fb_ptr ]!; \
2510 subs span_num_blocks, span_num_blocks, #1; \
2514 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2516 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2517 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2519 vshr.u8 r_whole_8, r_whole_8, #3; \
2520 rbit right_mask, right_mask; \
2521 vmov pixels, msb_mask; \
2522 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2523 clz right_mask, right_mask; \
2525 vmlal.u8 pixels, r_whole_8, d64_1; \
2526 vmlal.u8 pixels, g_whole_8, d64_4; \
2527 vmlal.u8 pixels, b_whole_8, d64_128; \
2529 ldr pc, [ pc, right_mask, lsl #2 ]; \
2542 vst1.u16 { pixels_low[0] }, [ fb_ptr ]; \
2546 vst1.u32 { pixels_low[0] }, [ fb_ptr ]; \
2550 vst1.u32 { pixels_low[0] }, [ fb_ptr ]!; \
2551 vst1.u16 { pixels_low[2] }, [ fb_ptr ]; \
2555 vst1.u32 { pixels_low }, [ fb_ptr ]; \
2559 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2560 vst1.u16 { pixels_high[0] }, [ fb_ptr ]; \
2564 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2565 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2569 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2570 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2571 vst1.u16 { pixels_high[2] }, [ fb_ptr ]; \
2575 vst1.u32 { pixels }, [ fb_ptr ]; \
2579 add span_uvrg_offset, span_uvrg_offset, #16; \
2580 add span_b_offset, span_b_offset, #4; \
2582 add span_edge_data, span_edge_data, #8; \
2583 subs num_spans, num_spans, #1; \
2587 ldmia sp!, { r4 - r11, pc } \
2589 setup_blocks_shaded_untextured_direct_builder(undithered)
2590 setup_blocks_shaded_untextured_direct_builder(dithered)
2599 #define block_ptr r1
2600 #define num_blocks r2
2613 #define texture_ptr r11
2627 #define pixels_d r10
2631 #define clut_ptr r12
2632 #define current_texture_mask r5
2633 #define dirty_textures_mask r6
2637 #define clut_low_a d2
2638 #define clut_low_b d3
2639 #define clut_high_a d4
2640 #define clut_high_b d5
2645 #define texels_low d6
2646 #define texels_high d7
2650 function(texture_blocks_untextured)
2656 function(texture_blocks_4bpp)
2657 stmdb sp!, { r3 - r11, r14 }
2658 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2660 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2661 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2663 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2664 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]
2666 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2667 vuzp.u8 clut_a, clut_b
2669 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
2670 tst dirty_textures_mask, current_texture_mask
2676 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2678 uxtah uv_0, texture_ptr, uv_01
2679 uxtah uv_1, texture_ptr, uv_01, ror #16
2681 uxtah uv_2, texture_ptr, uv_23
2682 uxtah uv_3, texture_ptr, uv_23, ror #16
2684 uxtah uv_4, texture_ptr, uv_45
2685 ldrb pixel_0, [ uv_0 ]
2687 uxtah uv_5, texture_ptr, uv_45, ror #16
2688 ldrb pixel_1, [ uv_1 ]
2690 uxtah uv_6, texture_ptr, uv_67
2691 ldrb pixel_2, [ uv_2 ]
2693 uxtah uv_7, texture_ptr, uv_67, ror #16
2694 ldrb pixel_3, [ uv_3 ]
2696 ldrb pixel_4, [ uv_4 ]
2697 subs num_blocks, num_blocks, #1
2699 ldrb pixel_5, [ uv_5 ]
2700 orr pixels_a, pixel_0, pixel_1, lsl #8
2702 ldrb pixel_6, [ uv_6 ]
2703 orr pixels_b, pixel_4, pixel_5, lsl #8
2705 ldrb pixel_7, [ uv_7 ]
2706 orr pixels_a, pixels_a, pixel_2, lsl #16
2708 orr pixels_b, pixels_b, pixel_6, lsl #16
2709 orr pixels_a, pixels_a, pixel_3, lsl #24
2711 orr pixels_b, pixels_b, pixel_7, lsl #24
2712 vmov.u32 texels, pixels_a, pixels_b
2714 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2715 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2717 vst2.u8 { texels_low, texels_high }, [ block_ptr, :128 ], c_64
2720 ldmia sp!, { r3 - r11, pc }
2723 stmdb sp!, { r1 - r2 }
2724 bl update_texture_4bpp_cache
2727 ldmia sp!, { r1 - r2 }
2733 function(texture_blocks_8bpp)
2734 stmdb sp!, { r3 - r11, r14 }
2735 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2737 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2738 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2740 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2741 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2743 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]
2744 tst dirty_textures_mask, current_texture_mask
2750 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2752 uxtah uv_0, texture_ptr, uv_01
2753 uxtah uv_1, texture_ptr, uv_01, ror #16
2755 uxtah uv_2, texture_ptr, uv_23
2756 uxtah uv_3, texture_ptr, uv_23, ror #16
2758 uxtah uv_4, texture_ptr, uv_45
2759 ldrb pixel_0, [ uv_0 ]
2761 uxtah uv_5, texture_ptr, uv_45, ror #16
2762 ldrb pixel_1, [ uv_1 ]
2764 uxtah uv_6, texture_ptr, uv_67
2765 ldrb pixel_2, [ uv_2 ]
2767 uxtah uv_7, texture_ptr, uv_67, ror #16
2768 ldrb pixel_3, [ uv_3 ]
2770 ldrb pixel_4, [ uv_4 ]
2771 add pixel_0, pixel_0, pixel_0
2773 ldrb pixel_5, [ uv_5 ]
2774 add pixel_1, pixel_1, pixel_1
2776 ldrb pixel_6, [ uv_6 ]
2777 add pixel_2, pixel_2, pixel_2
2779 ldrb pixel_7, [ uv_7 ]
2780 add pixel_3, pixel_3, pixel_3
2782 ldrh pixel_0, [ clut_ptr, pixel_0 ]
2783 add pixel_4, pixel_4, pixel_4
2785 ldrh pixel_1, [ clut_ptr, pixel_1 ]
2786 add pixel_5, pixel_5, pixel_5
2788 ldrh pixel_2, [ clut_ptr, pixel_2 ]
2789 add pixel_6, pixel_6, pixel_6
2791 ldrh pixel_3, [ clut_ptr, pixel_3 ]
2792 add pixel_7, pixel_7, pixel_7
2794 ldrh pixel_4, [ clut_ptr, pixel_4 ]
2795 orr pixels_a, pixel_0, pixel_1, lsl #16
2797 ldrh pixel_5, [ clut_ptr, pixel_5 ]
2798 orr pixels_c, pixel_2, pixel_3, lsl #16
2800 ldrh pixel_6, [ clut_ptr, pixel_6 ]
2801 subs num_blocks, num_blocks, #1
2803 ldrh pixel_7, [ clut_ptr, pixel_7 ]
2804 orr pixels_b, pixel_4, pixel_5, lsl #16
2806 orr pixels_d, pixel_6, pixel_7, lsl #16
2807 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2809 add block_ptr, block_ptr, #64
2812 ldmia sp!, { r3 - r11, pc }
2815 stmdb sp!, { r1 - r2, r12 }
2817 bl update_texture_8bpp_cache
2819 ldmia sp!, { r1 - r2, r12 }
2849 #define block_ptr r1
2850 #define num_blocks r2
2894 #define texture_ptr r12
2899 function(texture_blocks_16bpp)
2900 stmdb sp!, { r3 - r11, r14 }
2901 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2903 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2904 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2907 ldrh uv_0, [ block_ptr ]
2908 subs num_blocks, num_blocks, #1
2910 ldrh uv_1, [ block_ptr, #2 ]
2912 and v_0, uv_0, #0xFF00
2913 and v_1, uv_1, #0xFF00
2915 and u_0, uv_0, #0xFF
2916 and u_1, uv_1, #0xFF
2918 add uv_0, u_0, v_0, lsl #2
2919 ldrh uv_2, [ block_ptr, #4 ]
2921 add uv_1, u_1, v_1, lsl #2
2922 ldrh uv_3, [ block_ptr, #6 ]
2924 add uv_0, uv_0, uv_0
2925 add uv_1, uv_1, uv_1
2927 and v_2, uv_2, #0xFF00
2928 and v_3, uv_3, #0xFF00
2930 and u_2, uv_2, #0xFF
2931 and u_3, uv_3, #0xFF
2933 add uv_2, u_2, v_2, lsl #2
2934 ldrh uv_4, [ block_ptr, #8 ]
2936 add uv_3, u_3, v_3, lsl #2
2937 ldrh uv_5, [ block_ptr, #10 ]
2939 add uv_2, uv_2, uv_2
2940 add uv_3, uv_3, uv_3
2942 and v_4, uv_4, #0xFF00
2943 and v_5, uv_5, #0xFF00
2945 and u_4, uv_4, #0xFF
2946 and u_5, uv_5, #0xFF
2948 add uv_4, u_4, v_4, lsl #2
2949 ldrh uv_6, [ block_ptr, #12 ]
2951 add uv_5, u_5, v_5, lsl #2
2952 ldrh uv_7, [ block_ptr, #14 ]
2954 add uv_4, uv_4, uv_4
2955 ldrh pixel_0, [ texture_ptr, uv_0 ]
2957 add uv_5, uv_5, uv_5
2958 ldrh pixel_1, [ texture_ptr, uv_1 ]
2960 and v_6, uv_6, #0xFF00
2961 ldrh pixel_2, [ texture_ptr, uv_2 ]
2963 and v_7, uv_7, #0xFF00
2964 ldrh pixel_3, [ texture_ptr, uv_3 ]
2966 and u_6, uv_6, #0xFF
2967 ldrh pixel_4, [ texture_ptr, uv_4 ]
2969 and u_7, uv_7, #0xFF
2970 ldrh pixel_5, [ texture_ptr, uv_5 ]
2972 add uv_6, u_6, v_6, lsl #2
2973 add uv_7, u_7, v_7, lsl #2
2975 add uv_6, uv_6, uv_6
2976 add uv_7, uv_7, uv_7
2978 orr pixels_a, pixel_0, pixel_1, lsl #16
2979 orr pixels_b, pixel_2, pixel_3, lsl #16
2981 ldrh pixel_6, [ texture_ptr, uv_6 ]
2982 orr pixels_c, pixel_4, pixel_5, lsl #16
2984 ldrh pixel_7, [ texture_ptr, uv_7 ]
2985 orr pixels_d, pixel_6, pixel_7, lsl #16
2987 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
2988 add block_ptr, block_ptr, #64
2992 ldmia sp!, { r3 - r11, pc }
3007 #undef msb_mask_high
3015 #define num_blocks r1
3016 #define color_ptr r2
3017 #define colors_scalar r2
3018 #define colors_scalar_compare r3
3019 #define mask_msb_ptr r2
3021 #define block_ptr_load_a r0
3022 #define block_ptr_store r3
3023 #define block_ptr_load_b r12
3028 #define draw_mask_bits_scalar r5
3030 #define d128_0x07 q0
3031 #define d128_0x1F q1
3032 #define d128_0x8000 q2
3033 #define test_mask q3
3035 #define colors_rg q5
3036 #define colors_b_dm_bits q6
3037 #define texels_rg q7
3040 #define pixels_b q10
3042 #define zero_mask q4
3043 #define draw_mask q12
3044 #define msb_mask q13
3046 #define fb_pixels q8
3048 #define pixels_gb_low q9
3050 #define colors_r d10
3051 #define colors_g d11
3052 #define colors_b d12
3053 #define draw_mask_bits d13
3054 #define texels_r d14
3055 #define texels_g d15
3056 #define pixels_r_low d16
3057 #define pixels_g_low d18
3058 #define pixels_b_low d19
3059 #define msb_mask_low d26
3060 #define msb_mask_high d27
3065 #define texels_b d31
3067 #define shade_blocks_textured_modulated_prologue_indirect() \
3069 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3071 #define shade_blocks_textured_modulated_prologue_direct() \
3072 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3073 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ] \
3076 #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \
3078 #define shade_blocks_textured_false_modulation_check_undithered(target) \
3079 ldr colors_scalar, [ psx_gpu, #psx_gpu_triangle_color_offset ]; \
3080 movw colors_scalar_compare, #0x8080; \
3082 movt colors_scalar_compare, #0x80; \
3083 cmp colors_scalar, colors_scalar_compare; \
3084 beq shade_blocks_textured_unmodulated_##target \
3086 #define shade_blocks_textured_false_modulation_check_dithered(target) \
3088 #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \
3089 shade_blocks_textured_false_modulation_check_##dithering(target); \
3090 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3091 vld1.u32 { colors_r[] }, [ color_ptr, :32 ]; \
3092 vdup.u8 colors_g, colors_r[1]; \
3093 vdup.u8 colors_b, colors_r[2]; \
3094 vdup.u8 colors_r, colors_r[0] \
3097 #define shade_blocks_textured_modulated_load_dithered(target) \
3098 vld1.u32 { target }, [ block_ptr_load_b, :128 ] \
3100 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3101 vld1.u32 { target }, [ block_ptr_load_b, :128 ], c_32 \
3103 #define shade_blocks_textured_modulated_load_undithered(target) \
3105 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3106 add block_ptr_load_b, block_ptr_load_b, #32 \
3108 #define shade_blocks_textured_modulate_dithered(channel) \
3109 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3111 #define shade_blocks_textured_modulate_undithered(channel) \
3112 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3115 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3116 vst1.u32 { draw_mask }, [ block_ptr_store, :128 ]! \
3118 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3119 ldr fb_ptr, [ block_ptr_load_b, #(offset - 64) ]; \
3120 vld1.u32 { fb_pixels }, [ fb_ptr ]; \
3121 vbit.u16 pixels, fb_pixels, draw_mask \
3123 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3124 vst1.u32 { pixels }, [ block_ptr_store, :128 ], c_48 \
3126 #define shade_blocks_textured_modulated_store_pixels_direct() \
3127 vst1.u32 { pixels }, [ fb_ptr ] \
3130 #define shade_blocks_textured_modulated_load_rg_shaded() \
3131 vld1.u32 { colors_r, colors_g }, [ block_ptr_load_b, :128 ], c_32 \
3133 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3134 add block_ptr_load_b, block_ptr_load_b, #32 \
3136 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3137 vld1.u32 { colors_b, draw_mask_bits }, [ block_ptr_load_a, :128 ], c_32 \
3139 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3140 ldr draw_mask_bits_scalar, [ block_ptr_load_a, #8 ]; \
3141 add block_ptr_load_a, block_ptr_load_a, #32 \
3143 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3144 vdup.u16 draw_mask, draw_mask_bits[0] \
3146 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3147 vdup.u16 draw_mask, draw_mask_bits_scalar \
3150 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3152 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3153 vorr.u16 pixels, pixels, msb_mask \
3156 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3159 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3160 shade_blocks_textured_modulated_prologue_##shading(dithering, target); \
3161 stmdb sp!, { r4 - r5, lr }; \
3162 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3164 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
3166 shade_blocks_textured_modulated_prologue_##target(); \
3168 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3171 add block_ptr_load_b, block_ptr_load_a, #16; \
3172 vmov.u8 d64_1, #1; \
3173 vmov.u8 d64_4, #4; \
3174 vmov.u8 d64_128, #128; \
3176 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3177 vmov.u8 d128_0x07, #0x07; \
3179 shade_blocks_textured_modulated_load_rg_##shading(); \
3180 vmov.u8 d128_0x1F, #0x1F; \
3182 shade_blocks_textured_modulated_load_bdm_##shading(); \
3183 vmov.u16 d128_0x8000, #0x8000; \
3185 vmovn.u16 texels_r, texels; \
3186 vshrn.u16 texels_g, texels, #5; \
3188 vshrn.u16 texels_b, texels, #7; \
3189 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3191 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3192 vtst.u16 draw_mask, draw_mask, test_mask; \
3194 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3195 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3197 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3198 vshr.u8 texels_b, texels_b, #3; \
3200 shade_blocks_textured_modulate_##dithering(r); \
3201 shade_blocks_textured_modulate_##dithering(g); \
3202 shade_blocks_textured_modulate_##dithering(b); \
3204 vand.u16 pixels, texels, d128_0x8000; \
3205 vceq.u16 zero_mask, texels, #0; \
3207 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3208 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3209 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3211 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3212 vorr.u16 draw_mask, draw_mask, zero_mask; \
3213 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3214 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3216 subs num_blocks, num_blocks, #1; \
3222 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3223 shade_blocks_textured_modulated_load_rg_##shading(); \
3224 vshrn.u16 texels_g, texels, #5; \
3226 shade_blocks_textured_modulated_load_bdm_##shading(); \
3227 vshrn.u16 texels_b, texels, #7; \
3229 vmovn.u16 texels_r, texels; \
3230 vmlal.u8 pixels, pixels_r_low, d64_1; \
3232 vmlal.u8 pixels, pixels_g_low, d64_4; \
3233 vmlal.u8 pixels, pixels_b_low, d64_128; \
3234 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3236 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3237 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3239 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3240 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3242 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3243 vtst.u16 draw_mask, draw_mask, test_mask; \
3245 shade_blocks_textured_modulated_store_pixels_##target(); \
3246 vshr.u8 texels_b, texels_b, #3; \
3248 shade_blocks_textured_modulate_##dithering(r); \
3249 shade_blocks_textured_modulate_##dithering(g); \
3250 shade_blocks_textured_modulate_##dithering(b); \
3252 vand.u16 pixels, texels, d128_0x8000; \
3253 vceq.u16 zero_mask, texels, #0; \
3255 subs num_blocks, num_blocks, #1; \
3257 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3258 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3259 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3261 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3262 vorr.u16 draw_mask, draw_mask, zero_mask; \
3263 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3264 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3269 vmlal.u8 pixels, pixels_r_low, d64_1; \
3270 vmlal.u8 pixels, pixels_g_low, d64_4; \
3271 vmlal.u8 pixels, pixels_b_low, d64_128; \
3273 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3274 shade_blocks_textured_modulated_store_pixels_##target(); \
3276 ldmia sp!, { r4 - r5, pc } \
3279 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3280 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3281 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3282 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3284 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3285 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3286 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3287 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3305 #undef msb_mask_high
3308 #define num_blocks r1
3309 #define mask_msb_ptr r2
3310 #define color_ptr r3
3312 #define block_ptr_load r0
3313 #define draw_mask_store_ptr r3
3314 #define draw_mask_bits_ptr r12
3315 #define draw_mask_ptr r12
3316 #define pixel_store_ptr r14
3318 #define fb_ptr_cmp r4
3321 #define fb_ptr_next r14
3325 #define test_mask q0
3327 #define draw_mask q2
3328 #define zero_mask q3
3329 #define draw_mask_combined q4
3330 #define fb_pixels q5
3331 #define fb_pixels_next q6
3334 #define draw_mask_low d4
3335 #define draw_mask_high d5
3336 #define msb_mask_low d14
3337 #define msb_mask_high d15
3340 function(shade_blocks_textured_unmodulated_indirect)
3341 str r14, [ sp, #-4 ]
3342 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3344 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3345 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3347 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3348 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3351 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3353 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3354 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3355 [ draw_mask_bits_ptr, :16 ], c_64
3356 vceq.u16 zero_mask, pixels, #0
3358 vtst.u16 draw_mask, draw_mask, test_mask
3359 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3361 subs num_blocks, num_blocks, #1
3365 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3366 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3368 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3369 [ draw_mask_bits_ptr, :16 ], c_64
3370 vceq.u16 zero_mask, pixels, #0
3372 vtst.u16 draw_mask, draw_mask, test_mask
3373 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3375 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3376 subs num_blocks, num_blocks, #1
3381 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3382 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3389 function(shade_blocks_textured_unmodulated_direct)
3390 stmdb sp!, { r4, r14 }
3391 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3393 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3394 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3396 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3399 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3400 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3402 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3403 [ draw_mask_bits_ptr, :16 ], c_64
3404 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3406 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3407 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3408 vceq.u16 zero_mask, pixels, #0
3409 vtst.u16 draw_mask, draw_mask, test_mask
3411 subs num_blocks, num_blocks, #1
3415 mov fb_ptr, fb_ptr_next
3416 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3418 vorr.u16 pixels, pixels, msb_mask
3420 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3421 vmov fb_pixels, fb_pixels_next
3423 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3424 [ draw_mask_bits_ptr, :16 ], c_64
3425 vbif.u16 fb_pixels, pixels, draw_mask_combined
3427 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3429 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3430 add fb_ptr_cmp, fb_ptr_cmp, #14
3434 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3435 vceq.u16 zero_mask, pixels, #0
3437 vst1.u16 { fb_pixels }, [ fb_ptr ]
3438 vtst.u16 draw_mask, draw_mask, test_mask
3441 subs num_blocks, num_blocks, #1
3445 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3446 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3448 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3450 ldmia sp!, { r4, pc }
3453 vst1.u16 { fb_pixels }, [ fb_ptr ]
3454 vceq.u16 zero_mask, pixels, #0
3456 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3457 vtst.u16 draw_mask, draw_mask, test_mask
3462 function(shade_blocks_unshaded_untextured_indirect)
3467 function(shade_blocks_unshaded_untextured_direct)
3468 stmdb sp!, { r4, r14 }
3469 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3471 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3472 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3474 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3475 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3477 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3478 vld1.u16 { pixels }, [ color_ptr, :128 ]
3481 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3483 vorr.u16 pixels, pixels, msb_mask
3484 subs num_blocks, num_blocks, #1
3486 ldr fb_ptr_next, [ block_ptr_load ], #64
3488 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3492 vmov fb_pixels, fb_pixels_next
3493 mov fb_ptr, fb_ptr_next
3494 ldr fb_ptr_next, [ block_ptr_load ], #64
3496 vbif.u16 fb_pixels, pixels, draw_mask
3497 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3499 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3500 add fb_ptr_cmp, fb_ptr_cmp, #14
3504 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3505 vst1.u16 { fb_pixels }, [ fb_ptr ]
3508 subs num_blocks, num_blocks, #1
3512 vbif.u16 fb_pixels_next, pixels, draw_mask
3513 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3515 ldmia sp!, { r4, pc }
3518 vst1.u16 { fb_pixels }, [ fb_ptr ]
3519 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3523 #undef draw_mask_ptr
3530 #define num_blocks r1
3531 #define msb_mask_ptr r2
3532 #define pixel_ptr r3
3533 #define draw_mask_ptr r0
3536 #define fb_ptr_next r14
3537 #define fb_ptr_cmp r4
3545 #undef msb_mask_high
3546 #undef draw_mask_next
3549 #undef fb_pixels_next
3552 #define draw_mask q1
3554 #define fb_pixels q3
3555 #define blend_pixels q4
3556 #define pixels_no_msb q5
3557 #define blend_mask q6
3558 #define fb_pixels_no_msb q7
3559 #define d128_0x8000 q8
3560 #define d128_0x0421 q9
3561 #define fb_pixels_next q10
3562 #define blend_pixels_next q11
3563 #define pixels_next q12
3564 #define draw_mask_next q13
3565 #define write_mask q14
3567 #define pixels_rb q5
3568 #define pixels_mg q7
3570 #define d128_0x7C1F q8
3571 #define d128_0x03E0 q9
3572 #define fb_pixels_rb q10
3573 #define fb_pixels_g q11
3574 #define fb_pixels_masked q11
3575 #define d128_0x83E0 q15
3576 #define pixels_fourth q7
3577 #define d128_0x1C07 q12
3578 #define d128_0x00E0 q13
3579 #define d128_0x80E0 q13
3581 #define msb_mask_low d0
3582 #define msb_mask_high d1
3584 #define blend_blocks_average_set_blend_mask_textured(source) \
3585 vclt.s16 blend_mask, source, #0 \
3587 #define blend_blocks_average_set_stp_bit_textured() \
3588 vorr.u16 blend_pixels, #0x8000 \
3590 #define blend_blocks_average_combine_textured(source) \
3591 vbif.u16 blend_pixels, source, blend_mask \
3593 #define blend_blocks_average_set_blend_mask_untextured(source) \
3595 #define blend_blocks_average_set_stp_bit_untextured() \
3597 #define blend_blocks_average_combine_untextured(source) \
3599 #define blend_blocks_average_mask_set_on() \
3600 vclt.s16 write_mask, fb_pixels_next, #0 \
3602 #define blend_blocks_average_mask_copy_on() \
3603 vorr.u16 draw_mask, draw_mask_next, write_mask \
3605 #define blend_blocks_average_mask_copy_b_on() \
3606 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3608 #define blend_blocks_average_mask_set_off() \
3610 #define blend_blocks_average_mask_copy_off() \
3611 vmov draw_mask, draw_mask_next \
3613 #define blend_blocks_average_mask_copy_b_off() \
3615 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3618 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3619 stmdb sp!, { r4, r14 }; \
3620 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3621 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3623 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3624 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3626 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3629 vmov.u16 d128_0x8000, #0x8000; \
3630 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3631 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3633 vmov.u16 d128_0x0421, #0x0400; \
3634 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3636 vorr.u16 d128_0x0421, #0x0021; \
3637 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3639 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3640 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3641 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3642 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3643 blend_blocks_average_mask_set_##mask_evaluate(); \
3644 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3646 subs num_blocks, num_blocks, #1; \
3650 mov fb_ptr, fb_ptr_next; \
3651 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3653 vmov pixels, pixels_next; \
3654 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3656 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3658 blend_blocks_average_mask_copy_##mask_evaluate(); \
3659 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3661 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3662 blend_blocks_average_set_stp_bit_##texturing(); \
3663 vmov fb_pixels, fb_pixels_next; \
3664 blend_blocks_average_combine_##texturing(pixels); \
3666 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3667 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3668 cmp fb_ptr_cmp, #28; \
3671 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3672 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3674 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3675 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3677 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3678 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3680 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3681 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3682 blend_blocks_average_mask_set_##mask_evaluate(); \
3683 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3686 subs num_blocks, num_blocks, #1; \
3690 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3691 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3693 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3694 blend_blocks_average_set_stp_bit_##texturing(); \
3695 blend_blocks_average_combine_##texturing(pixels_next); \
3697 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3698 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3699 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3701 ldmia sp!, { r4, pc }; \
3704 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3705 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3706 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3708 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3709 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3710 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3711 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3712 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3713 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3717 blend_blocks_average_builder(textured, off)
3718 blend_blocks_average_builder(untextured, off)
3719 blend_blocks_average_builder(textured, on)
3720 blend_blocks_average_builder(untextured, on)
3723 #define blend_blocks_add_mask_set_on() \
3724 vclt.s16 write_mask, fb_pixels, #0 \
3726 #define blend_blocks_add_mask_copy_on() \
3727 vorr.u16 draw_mask, draw_mask, write_mask \
3729 #define blend_blocks_add_mask_set_off() \
3731 #define blend_blocks_add_mask_copy_off() \
3734 #define blend_blocks_add_textured_builder(mask_evaluate) \
3737 function(blend_blocks_textured_add_##mask_evaluate) \
3738 stmdb sp!, { r4, r14 }; \
3739 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3740 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3742 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3743 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3745 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3748 vmov.u16 d128_0x7C1F, #0x7C00; \
3749 vmov.u16 d128_0x03E0, #0x0300; \
3750 vmov.u16 d128_0x83E0, #0x8000; \
3751 vorr.u16 d128_0x03E0, #0x00E0; \
3752 vorr.u16 d128_0x7C1F, #0x001F; \
3753 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3755 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3756 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3757 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3758 vclt.s16 blend_mask, pixels, #0; \
3759 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3760 blend_blocks_add_mask_set_##mask_evaluate(); \
3761 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3763 blend_blocks_add_mask_copy_##mask_evaluate(); \
3764 vorr.u16 pixels, pixels, msb_mask; \
3765 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3766 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3767 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3768 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3769 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3770 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3771 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3772 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3774 subs num_blocks, num_blocks, #1; \
3778 mov fb_ptr, fb_ptr_next; \
3780 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3782 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3783 vclt.s16 blend_mask, pixels, #0; \
3785 vorr.u16 pixels, pixels, msb_mask; \
3786 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3787 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3789 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3790 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3792 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3793 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3794 cmp fb_ptr_cmp, #28; \
3797 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3798 blend_blocks_add_mask_set_##mask_evaluate(); \
3799 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3800 blend_blocks_add_mask_copy_##mask_evaluate(); \
3801 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3802 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3803 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3806 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3807 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3808 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3809 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3810 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3812 subs num_blocks, num_blocks, #1; \
3816 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3817 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3818 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3820 ldmia sp!, { r4, pc }; \
3823 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3824 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3826 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3827 blend_blocks_add_mask_set_##mask_evaluate(); \
3828 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3829 blend_blocks_add_mask_copy_##mask_evaluate(); \
3830 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3834 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3837 function(blend_blocks_untextured_add_##mask_evaluate) \
3838 stmdb sp!, { r4, r14 }; \
3839 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3840 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3842 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3843 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3845 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3848 vmov.u16 d128_0x7C1F, #0x7C00; \
3849 vmov.u16 d128_0x03E0, #0x0300; \
3850 vorr.u16 d128_0x7C1F, #0x001F; \
3851 vorr.u16 d128_0x03E0, #0x00E0; \
3853 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3854 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3855 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3856 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3857 blend_blocks_add_mask_set_##mask_evaluate(); \
3858 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3860 blend_blocks_add_mask_copy_##mask_evaluate(); \
3861 vand.u16 pixels_g, pixels, d128_0x03E0; \
3862 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3863 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3864 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3865 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3866 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3867 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3869 subs num_blocks, num_blocks, #1; \
3873 mov fb_ptr, fb_ptr_next; \
3875 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3877 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3879 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3880 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3881 vand.u16 pixels_g, pixels, d128_0x03E0; \
3883 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3884 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3886 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3887 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3888 cmp fb_ptr_cmp, #28; \
3891 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3892 blend_blocks_add_mask_set_##mask_evaluate(); \
3893 blend_blocks_add_mask_copy_##mask_evaluate(); \
3894 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3895 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3896 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3899 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3900 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3901 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3902 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3903 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3905 subs num_blocks, num_blocks, #1; \
3909 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3910 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3911 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3912 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3914 ldmia sp!, { r4, pc }; \
3917 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3918 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3920 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3921 blend_blocks_add_mask_set_##mask_evaluate(); \
3922 blend_blocks_add_mask_copy_##mask_evaluate(); \
3923 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3927 blend_blocks_add_textured_builder(off)
3928 blend_blocks_add_textured_builder(on)
3929 blend_blocks_add_untextured_builder(off)
3930 blend_blocks_add_untextured_builder(on)
3932 #define blend_blocks_subtract_set_blend_mask_textured() \
3933 vclt.s16 blend_mask, pixels_next, #0 \
3935 #define blend_blocks_subtract_combine_textured() \
3936 vbif.u16 blend_pixels, pixels, blend_mask \
3938 #define blend_blocks_subtract_set_stb_textured() \
3939 vorr.u16 blend_pixels, #0x8000 \
3941 #define blend_blocks_subtract_msb_mask_textured() \
3942 vorr.u16 pixels, pixels_next, msb_mask \
3944 #define blend_blocks_subtract_set_blend_mask_untextured() \
3946 #define blend_blocks_subtract_combine_untextured() \
3948 #define blend_blocks_subtract_set_stb_untextured() \
3949 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3951 #define blend_blocks_subtract_msb_mask_untextured() \
3954 #define blend_blocks_subtract_mask_set_on() \
3955 vclt.s16 write_mask, fb_pixels, #0 \
3957 #define blend_blocks_subtract_mask_copy_on() \
3958 vorr.u16 draw_mask, draw_mask_next, write_mask \
3960 #define blend_blocks_subtract_mask_set_off() \
3962 #define blend_blocks_subtract_mask_copy_off() \
3963 vmov draw_mask, draw_mask_next \
3966 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
3969 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
3970 stmdb sp!, { r4, r14 }; \
3971 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3972 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3974 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3975 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3977 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3980 vmov.u16 d128_0x7C1F, #0x7C00; \
3981 vmov.u16 d128_0x03E0, #0x0300; \
3982 vorr.u16 d128_0x7C1F, #0x001F; \
3983 vorr.u16 d128_0x03E0, #0x00E0; \
3985 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3986 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3987 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3988 blend_blocks_subtract_set_blend_mask_##texturing(); \
3989 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3990 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3991 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3993 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3994 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3995 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3996 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3997 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3999 subs num_blocks, num_blocks, #1; \
4003 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4004 mov fb_ptr, fb_ptr_next; \
4005 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4007 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
4008 blend_blocks_subtract_msb_mask_##texturing(); \
4010 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
4011 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4012 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4013 blend_blocks_subtract_set_stb_##texturing(); \
4014 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4015 blend_blocks_subtract_combine_##texturing(); \
4016 blend_blocks_subtract_set_blend_mask_##texturing(); \
4017 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4019 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4020 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4021 cmp fb_ptr_cmp, #28; \
4024 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4025 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4026 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4027 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4028 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4029 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4030 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4033 subs num_blocks, num_blocks, #1; \
4037 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4039 blend_blocks_subtract_msb_mask_##texturing(); \
4040 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4041 blend_blocks_subtract_set_stb_##texturing(); \
4042 blend_blocks_subtract_combine_##texturing(); \
4043 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4044 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4046 ldmia sp!, { r4, pc }; \
4049 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4050 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4051 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4052 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4053 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4054 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4055 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4059 blend_blocks_subtract_builder(textured, off)
4060 blend_blocks_subtract_builder(textured, on)
4061 blend_blocks_subtract_builder(untextured, off)
4062 blend_blocks_subtract_builder(untextured, on)
4065 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4068 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4069 stmdb sp!, { r4, r14 }; \
4070 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4071 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4073 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4074 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4076 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4079 vmov.u16 d128_0x7C1F, #0x7C00; \
4080 vmov.u16 d128_0x03E0, #0x0300; \
4081 vmov.u16 d128_0x83E0, #0x8300; \
4082 vmov.u16 d128_0x1C07, #0x1C00; \
4083 vmov.u16 d128_0x80E0, #0x8000; \
4084 vorr.u16 d128_0x7C1F, #0x001F; \
4085 vorr.u16 d128_0x03E0, #0x00E0; \
4086 vorr.u16 d128_0x83E0, #0x00E0; \
4087 vorr.u16 d128_0x1C07, #0x0007; \
4088 vorr.u16 d128_0x80E0, #0x00E0; \
4090 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4091 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4092 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4093 vclt.s16 blend_mask, pixels, #0; \
4094 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4095 blend_blocks_add_mask_set_##mask_evaluate(); \
4096 vshr.s16 pixels_fourth, pixels, #2; \
4098 blend_blocks_add_mask_copy_##mask_evaluate(); \
4099 vorr.u16 pixels, pixels, msb_mask; \
4100 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4101 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
4102 vand.u16 pixels_mg, pixels_fourth, d128_0x80E0; \
4103 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
4104 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
4105 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4106 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
4107 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4108 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
4110 subs num_blocks, num_blocks, #1; \
4114 mov fb_ptr, fb_ptr_next; \
4116 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4118 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4119 vclt.s16 blend_mask, pixels, #0; \
4121 vshr.s16 pixels_fourth, pixels, #2; \
4122 vorr.u16 pixels, pixels, msb_mask; \
4123 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4124 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4126 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4127 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4129 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4130 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4131 cmp fb_ptr_cmp, #28; \
4134 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4135 blend_blocks_add_mask_set_##mask_evaluate(); \
4136 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
4137 blend_blocks_add_mask_copy_##mask_evaluate(); \
4138 vand.u16 pixels_mg, pixels_fourth, d128_0x80E0; \
4139 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
4140 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4143 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
4144 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4145 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
4146 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4147 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
4149 subs num_blocks, num_blocks, #1; \
4153 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4154 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4155 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4157 ldmia sp!, { r4, pc }; \
4160 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4161 vand.u16 pixels_mg, pixels_fourth, d128_0x80E0; \
4163 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4164 blend_blocks_add_mask_set_##mask_evaluate(); \
4165 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
4166 blend_blocks_add_mask_copy_##mask_evaluate(); \
4167 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
4171 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4174 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4175 stmdb sp!, { r4, r14 }; \
4176 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4177 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4179 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4180 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4182 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4185 vmov.u16 d128_0x7C1F, #0x7C00; \
4186 vmov.u16 d128_0x03E0, #0x0300; \
4187 vmov.u16 d128_0x83E0, #0x8300; \
4188 vmov.u16 d128_0x1C07, #0x1C00; \
4189 vmov.u16 d128_0x00E0, #0x00E0; \
4190 vorr.u16 d128_0x7C1F, #0x001F; \
4191 vorr.u16 d128_0x03E0, #0x00E0; \
4192 vorr.u16 d128_0x83E0, #0x00E0; \
4193 vorr.u16 d128_0x1C07, #0x0007; \
4195 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4196 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4197 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4198 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4199 blend_blocks_add_mask_set_##mask_evaluate(); \
4200 vshr.s16 pixels_fourth, pixels, #2; \
4201 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4203 blend_blocks_add_mask_copy_##mask_evaluate(); \
4204 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4205 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4206 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4207 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4208 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4209 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4210 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4212 subs num_blocks, num_blocks, #1; \
4216 mov fb_ptr, fb_ptr_next; \
4218 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4220 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4222 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4223 vshr.s16 pixels_fourth, pixels, #2; \
4224 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4225 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4227 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4228 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4230 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4231 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4232 cmp fb_ptr_cmp, #28; \
4235 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4236 blend_blocks_add_mask_set_##mask_evaluate(); \
4237 blend_blocks_add_mask_copy_##mask_evaluate(); \
4238 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4239 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4240 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4243 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4244 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4245 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4246 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4247 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4249 subs num_blocks, num_blocks, #1; \
4253 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4254 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4255 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4256 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4258 ldmia sp!, { r4, pc }; \
4261 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4262 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4264 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4265 blend_blocks_add_mask_set_##mask_evaluate(); \
4266 blend_blocks_add_mask_copy_##mask_evaluate(); \
4267 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4271 blend_blocks_add_fourth_textured_builder(off)
4272 blend_blocks_add_fourth_textured_builder(on)
4273 blend_blocks_add_fourth_untextured_builder(off)
4274 blend_blocks_add_fourth_untextured_builder(on)
4276 // TODO: Optimize this more. Need a scene that actually uses it for
4281 function(blend_blocks_textured_unblended_on)
4282 stmdb sp!, { r4, r14 }
4283 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4284 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
4286 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4287 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
4289 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4292 ldr fb_ptr, [ pixel_ptr, #28 ]
4293 vld1.u16 { fb_pixels }, [ fb_ptr ]
4294 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4295 vclt.s16 write_mask, fb_pixels, #0
4296 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4298 subs num_blocks, num_blocks, #1
4302 vorr.u16 draw_mask, draw_mask, write_mask
4303 vbif.u16 fb_pixels, pixels, draw_mask
4304 vst1.u16 { fb_pixels }, [ fb_ptr ]
4306 ldr fb_ptr, [ pixel_ptr, #28 ]
4307 vld1.u16 { fb_pixels }, [ fb_ptr ]
4308 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4309 vclt.s16 write_mask, fb_pixels, #0
4310 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4312 subs num_blocks, num_blocks, #1
4316 vorr.u16 draw_mask, draw_mask, write_mask
4317 vbif.u16 fb_pixels, pixels, draw_mask
4318 vst1.u16 { fb_pixels }, [ fb_ptr ]
4320 ldmia sp!, { r4, pc }
4323 function(blend_blocks_textured_unblended_off)
4333 vld1.u32 { u_whole_8, v_whole_8 }, [ r1, :128 ], r3
4353 #define parameter_width_offset 0
4354 #define parameter_height_offset 4
4360 #define left_unaligned r14
4361 #define right_unaligned r4
4363 #define num_unaligned r2
4364 #define num_width r6
4374 function(render_block_fill_body)
4375 ldr vram_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
4376 ldr height, [ sp, #parameter_height_offset ]
4378 add vram_ptr, vram_ptr, y, lsl #11
4379 ldr width, [ sp, #parameter_width_offset ]
4381 add vram_ptr, vram_ptr, x, lsl #1
4382 stmdb sp!, { r4 - r6, r14 }
4384 ubfx color_r, color, #3, #5
4385 ubfx color_g, color, #11, #5
4387 ubfx color_b, color, #19, #5
4388 orr color, color_r, color_g, lsl #5
4390 orr color, color, color_b, lsl #10
4391 vdup.u16 colors_a, color
4393 vmov colors_b, colors_a
4395 sub pitch, pitch, width, lsl #1
4398 mov num_width, width, lsr #4
4401 vst1.u32 { colors_a, colors_b }, [ vram_ptr, :128 ]!
4403 subs num_width, num_width, #1
4406 add vram_ptr, vram_ptr, pitch
4407 subs height, height, #1
4411 ldmia sp!, { r4 - r6, pc }
4422 #undef dirty_textures_mask
4424 #undef current_texture_mask
4435 #define offset_u_right r10
4436 #define width_rounded r11
4437 #define height_rounded r12
4439 #define texture_offset_base r1
4440 #define tile_width r2
4441 #define tile_height r3
4442 #define num_blocks r4
4444 #define sub_tile_height r6
4446 #define texture_mask r8
4447 #define column_data r9
4448 #define texture_offset r10
4449 #define tiles_remaining r11
4450 #define fb_ptr_advance_column r12
4451 #define texture_block_ptr r14
4453 #define texture_page_ptr r3
4454 #define left_block_mask r4
4455 #define right_block_mask r5
4456 #define texture_mask_rev r10
4457 #define control_mask r11
4459 #define dirty_textures_mask r4
4461 #define current_texture_mask r6
4475 #define draw_masks_fb_ptrs q1
4477 #define draw_mask_fb_ptr_left d2
4478 #define draw_mask_fb_ptr_right d3
4480 #define clut_low_a d4
4481 #define clut_low_b d5
4482 #define clut_high_a d6
4483 #define clut_high_b d7
4485 #define block_masks d8
4486 #define block_masks_shifted d9
4491 #define texels_low d10
4492 #define texels_high d11
4495 setup_sprite_flush_blocks_single:
4498 stmdb sp!, { r0 - r3, r12, r14 }
4499 bl flush_render_block_buffer
4500 ldmia sp!, { r0 - r3, r12, r14 }
4504 add block, psx_gpu, #psx_gpu_blocks_offset
4506 mov num_blocks, sub_tile_height
4510 setup_sprite_flush_blocks_double:
4513 stmdb sp!, { r0 - r3, r12, r14 }
4514 bl flush_render_block_buffer
4515 ldmia sp!, { r0 - r3, r12, r14 }
4519 add block, psx_gpu, #psx_gpu_blocks_offset
4521 mov num_blocks, sub_tile_height, lsl #1
4525 setup_sprite_update_texture_4bpp_cache:
4526 stmdb sp!, { r0 - r3, r14 }
4527 bl update_texture_4bpp_cache
4528 ldmia sp!, { r0 - r3, pc }
4531 setup_sprite_update_texture_8bpp_cache:
4532 stmdb sp!, { r0 - r3, r14 }
4533 bl update_texture_8bpp_cache
4534 ldmia sp!, { r0 - r3, pc }
4537 #define setup_sprite_tiled_initialize_4bpp() \
4538 ldr dirty_textures_mask, \
4539 [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]; \
4540 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4542 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4543 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4545 tst current_texture_mask, dirty_textures_mask; \
4546 vuzp.u8 clut_a, clut_b; \
4548 blne setup_sprite_update_texture_4bpp_cache \
4550 #define setup_sprite_tiled_initialize_8bpp() \
4551 ldr dirty_textures_mask, \
4552 [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]; \
4553 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4555 tst current_texture_mask, dirty_textures_mask; \
4556 blne setup_sprite_update_texture_8bpp_cache \
4559 #define setup_sprite_tile_setup_block_no(side, offset, texture_mode) \
4561 #define setup_sprite_block_count_single() \
4564 #define setup_sprite_block_count_double() \
4565 sub_tile_height, lsl #1 \
4567 #define setup_sprite_tile_add_blocks(type) \
4568 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4569 cmp num_blocks, #MAX_BLOCKS; \
4571 blgt setup_sprite_flush_blocks_##type \
4574 #define setup_sprite_tile_full_4bpp(edge) \
4575 setup_sprite_tile_add_blocks(double); \
4578 and texture_block_ptr, texture_offset, texture_mask; \
4579 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4582 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4583 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4585 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4586 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4588 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4589 add texture_block_ptr, texture_offset, #8; \
4591 and texture_block_ptr, texture_block_ptr, texture_mask; \
4592 add block, block, #40; \
4594 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4595 add fb_ptr, fb_ptr, #16; \
4597 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4598 add block, block, #24; \
4600 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4601 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4604 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4605 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4607 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4608 add block, block, #40; \
4610 add texture_offset, texture_offset, #0x10; \
4611 add fb_ptr, fb_ptr, #(2048 - 16); \
4613 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4614 add block, block, #24; \
4616 subs sub_tile_height, sub_tile_height, #1; \
4619 add texture_offset, texture_offset, #0xF00; \
4620 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4623 #define setup_sprite_tile_half_4bpp(edge) \
4624 setup_sprite_tile_add_blocks(single); \
4627 and texture_block_ptr, texture_offset, texture_mask; \
4628 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4631 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4632 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4634 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4635 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4637 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4638 add block, block, #40; \
4640 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4641 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4643 add block, block, #24; \
4644 add texture_offset, texture_offset, #0x10; \
4646 add fb_ptr, fb_ptr, #2048; \
4647 subs sub_tile_height, sub_tile_height, #1; \
4651 add texture_offset, texture_offset, #0xF00; \
4652 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4655 #define setup_sprite_tile_full_8bpp(edge) \
4656 setup_sprite_tile_add_blocks(double); \
4657 add block, block, #16; \
4660 and texture_block_ptr, texture_offset, texture_mask; \
4661 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4664 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4665 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4667 add texture_block_ptr, texture_offset, #8; \
4668 vst1.u32 { texels }, [ block, :64 ]; \
4670 and texture_block_ptr, texture_block_ptr, texture_mask; \
4671 add block, block, #24; \
4673 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4675 add fb_ptr, fb_ptr, #16; \
4676 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4678 add block, block, #40; \
4679 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4682 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4683 vst1.u32 { texels }, [ block, :64 ]; \
4684 add block, block, #24; \
4686 add texture_offset, texture_offset, #0x10; \
4687 add fb_ptr, fb_ptr, #(2048 - 16); \
4689 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4690 add block, block, #40; \
4692 subs sub_tile_height, sub_tile_height, #1; \
4695 sub block, block, #16; \
4696 add texture_offset, texture_offset, #0xF00; \
4697 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4700 #define setup_sprite_tile_half_8bpp(edge) \
4701 setup_sprite_tile_add_blocks(single); \
4702 add block, block, #16; \
4705 and texture_block_ptr, texture_offset, texture_mask; \
4706 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4709 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4710 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4712 vst1.u32 { texels }, [ block, :64 ]; \
4713 add block, block, #24; \
4715 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4716 add block, block, #40; \
4718 add texture_offset, texture_offset, #0x10; \
4719 add fb_ptr, fb_ptr, #2048; \
4721 subs sub_tile_height, sub_tile_height, #1; \
4724 sub block, block, #16; \
4725 add texture_offset, texture_offset, #0xF00; \
4726 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4729 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4730 add texture_offset, texture_offset_base, #8; \
4731 add fb_ptr, fb_ptr, #16 \
4733 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4734 mov texture_offset, texture_offset_base \
4736 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4737 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4739 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4740 mov texture_offset, texture_offset_base \
4742 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4743 sub fb_ptr, fb_ptr, #16 \
4745 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4747 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4748 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4750 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4753 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode) \
4754 mov sub_tile_height, column_data; \
4755 setup_sprite_tile_column_edge_pre_adjust_##edge_mode(edge); \
4756 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4757 setup_sprite_tile_column_edge_post_adjust_##edge_mode(edge) \
4759 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode) \
4760 and sub_tile_height, column_data, #0xFF; \
4761 mov tiles_remaining, column_data, lsr #16; \
4762 setup_sprite_tile_column_edge_pre_adjust_##edge_mode(edge); \
4763 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4765 subs tiles_remaining, tiles_remaining, #1; \
4769 mov sub_tile_height, #16; \
4770 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4771 subs tiles_remaining, tiles_remaining, #1; \
4775 uxtb sub_tile_height, column_data, ror #8; \
4776 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4777 setup_sprite_tile_column_edge_post_adjust_##edge_mode(edge) \
4780 #define setup_sprite_column_data_single() \
4781 mov column_data, height; \
4782 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] \
4784 #define setup_sprite_column_data_multi() \
4785 and height_rounded, height_rounded, #0xF; \
4786 rsb column_data, offset_v, #16; \
4788 add height_rounded, height_rounded, #1; \
4789 sub tile_height, tile_height, #1; \
4791 orr column_data, column_data, tile_height, lsl #16; \
4792 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]; \
4794 orr column_data, column_data, height_rounded, lsl #8 \
4796 #define setup_sprite_tile_column_width_single(texture_mode, multi_height, \
4798 setup_sprite_##texture_mode##_single_##multi_height##_##edge_mode##_##edge: \
4799 setup_sprite_column_data_##multi_height(); \
4800 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4801 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4802 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4803 vdup.u8 draw_mask_fb_ptr_right, block_masks[1]; \
4805 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, \
4807 ldmia sp!, { r4 - r11, pc } \
4809 #define setup_sprite_tiled_advance_column() \
4810 add texture_offset_base, texture_offset_base, #0x100; \
4811 tst texture_offset_base, #0xF00; \
4812 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4814 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4816 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode: \
4817 setup_sprite_column_data_##multi_height(); \
4818 mov fb_ptr_advance_column, #32; \
4820 sub fb_ptr_advance_column, height, lsl #11; \
4821 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4823 vdup.u8 draw_mask_fb_ptr_right, block_masks[1]; \
4824 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm); \
4826 subs tile_width, tile_width, #2; \
4827 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4829 vmov.u8 draw_masks_fb_ptrs, #0; \
4833 setup_sprite_tiled_advance_column(); \
4834 setup_sprite_tile_column_height_##multi_height(full, none, tm); \
4835 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4836 subs tile_width, tile_width, #1; \
4840 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4841 vdup.u8 draw_mask_fb_ptr_right, block_masks[5]; \
4843 setup_sprite_tiled_advance_column(); \
4844 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm); \
4845 ldmia sp!, { r4 - r11, pc } \
4853 // [ sp + 4 ]: width
4854 // [ sp + 8 ]: height
4855 // [ sp + 12 ]: color (unused)
4857 #define setup_sprite_tiled_builder(texture_mode) \
4859 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full); \
4860 setup_sprite_tile_column_width_single(texture_mode, multi, full, none); \
4861 setup_sprite_tile_column_width_multi(texture_mode, single, full, full); \
4862 setup_sprite_tile_column_width_single(texture_mode, single, full, none); \
4863 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full); \
4864 setup_sprite_tile_column_width_single(texture_mode, multi, half, right); \
4865 setup_sprite_tile_column_width_multi(texture_mode, single, half, full); \
4866 setup_sprite_tile_column_width_single(texture_mode, single, half, right); \
4867 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half); \
4868 setup_sprite_tile_column_width_single(texture_mode, multi, half, left); \
4869 setup_sprite_tile_column_width_multi(texture_mode, single, full, half); \
4870 setup_sprite_tile_column_width_single(texture_mode, single, half, left); \
4871 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half); \
4872 setup_sprite_tile_column_width_multi(texture_mode, single, half, half); \
4876 function(setup_sprite_##texture_mode) \
4877 stmdb sp!, { r4 - r11, r14 }; \
4878 setup_sprite_tiled_initialize_##texture_mode(); \
4880 ldr v, [ sp, #36 ]; \
4881 and offset_u, u, #0xF; \
4883 ldr width, [ sp, #40 ]; \
4884 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
4886 ldr height, [ sp, #44 ]; \
4887 add fb_ptr, fb_ptr, y, lsl #11; \
4889 add fb_ptr, fb_ptr, x, lsl #1; \
4890 and offset_v, v, #0xF; \
4892 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4893 add width_rounded, offset_u, width; \
4895 add height_rounded, offset_v, height; \
4896 add width_rounded, width_rounded, #15; \
4898 add height_rounded, height_rounded, #15; \
4899 mov tile_width, width_rounded, lsr #4; \
4901 /* texture_offset_base = VH-VL-00-00 */\
4902 mov texture_offset_base, v, lsl #8; \
4903 and offset_u_right, width_rounded, #0xF; \
4905 /* texture_offset_base = VH-UH-UL-00 */\
4906 bfi texture_offset_base, u, #4, #8; \
4907 movw right_block_mask, #0xFFFE; \
4909 /* texture_offset_base = VH-UH-VL-00 */\
4910 bfi texture_offset_base, v, #4, #4; \
4911 movw left_block_mask, #0xFFFF; \
4913 mov tile_height, height_rounded, lsr #4; \
4914 mvn left_block_mask, left_block_mask, lsl offset_u; \
4916 /* texture_mask = HH-HL-WH-WL */\
4917 ldrh texture_mask, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]; \
4918 mov right_block_mask, right_block_mask, lsl offset_u_right; \
4920 /* texture_mask_rev = WH-WL-HH-HL */\
4921 rev16 texture_mask_rev, texture_mask; \
4922 vmov block_masks, left_block_mask, right_block_mask; \
4924 /* texture_mask = HH-HL-HL-WL */\
4925 bfi texture_mask, texture_mask_rev, #4, #4; \
4926 /* texture_mask_rev = 00-00-00-WH */\
4927 mov texture_mask_rev, texture_mask_rev, lsr #12; \
4929 /* texture_mask = HH-WH-HL-WL */\
4930 bfi texture_mask, texture_mask_rev, #8, #4; \
4931 and left_block_mask, left_block_mask, #0xFF; \
4933 mov control_mask, #0; \
4934 cmp left_block_mask, #0xFF; \
4936 uxtb right_block_mask, right_block_mask, ror #8; \
4937 orreq control_mask, control_mask, #0x4; \
4939 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4940 cmp right_block_mask, #0xFF; \
4942 orreq control_mask, control_mask, #0x8; \
4943 cmp tile_width, #1; \
4945 add block, psx_gpu, #psx_gpu_blocks_offset; \
4946 orreq control_mask, control_mask, #0x1; \
4948 cmp tile_height, #1; \
4949 add block, block, num_blocks, lsl #6; \
4951 orreq control_mask, control_mask, #0x2; \
4952 ldr pc, [ pc, control_mask, lsl #2 ]; \
4955 .word setup_sprite_##texture_mode##_multi_multi_full_full; \
4956 .word setup_sprite_##texture_mode##_single_multi_full_none; \
4957 .word setup_sprite_##texture_mode##_multi_single_full_full; \
4958 .word setup_sprite_##texture_mode##_single_single_full_none; \
4959 .word setup_sprite_##texture_mode##_multi_multi_half_full; \
4960 .word setup_sprite_##texture_mode##_single_multi_half_right; \
4961 .word setup_sprite_##texture_mode##_multi_single_half_full; \
4962 .word setup_sprite_##texture_mode##_single_single_half_right; \
4963 .word setup_sprite_##texture_mode##_multi_multi_full_half; \
4964 .word setup_sprite_##texture_mode##_single_multi_half_left; \
4965 .word setup_sprite_##texture_mode##_multi_single_full_half; \
4966 .word setup_sprite_##texture_mode##_single_single_half_left; \
4967 .word setup_sprite_##texture_mode##_multi_multi_half_half; \
4969 .word setup_sprite_##texture_mode##_multi_single_half_half \
4972 setup_sprite_tiled_builder(4bpp);
4973 setup_sprite_tiled_builder(8bpp);
4981 #define block_ptr r0
4982 #define num_blocks r1
4984 #define texel_shift_mask r3
4985 #define block_pixels_a r4
4986 #define block_pixels_b r5
4995 #define texels_01 r6
4996 #define texels_23 r7
4997 #define texels_45 r8
4998 #define texels_67 r9
5000 function(texture_sprite_blocks_8bpp)
5001 stmdb sp!, { r4 - r11, r14 }
5002 movw texel_shift_mask, #(0xFF << 1)
5004 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5005 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
5007 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
5008 ldr block_pixels_a, [ block_ptr, #16 ]
5011 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
5012 ldr block_pixels_b, [ block_ptr, #20 ]
5014 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
5015 ldrh texel_0, [ clut_ptr, texel_0 ]
5017 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
5018 ldrh texel_1, [ clut_ptr, texel_1 ]
5020 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
5021 ldr block_pixels_a, [ block_ptr, #(64 + 16) ]
5023 ldrh texel_2, [ clut_ptr, texel_2 ]
5024 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
5026 ldrh texel_3, [ clut_ptr, texel_3 ]
5027 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
5029 ldrh texel_4, [ clut_ptr, texel_4 ]
5030 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
5032 ldrh texel_5, [ clut_ptr, texel_5 ]
5033 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
5035 ldrh texel_6, [ clut_ptr, texel_6 ]
5036 orr texels_01, texel_0, texel_1, lsl #16
5038 ldrh texel_7, [ clut_ptr, texel_7 ]
5039 orr texels_23, texel_2, texel_3, lsl #16
5041 orr texels_45, texel_4, texel_5, lsl #16
5042 str texels_01, [ block_ptr, #0 ]
5044 orr texels_67, texel_6, texel_7, lsl #16
5045 str texels_23, [ block_ptr, #4 ]
5047 subs num_blocks, num_blocks, #1
5048 str texels_45, [ block_ptr, #8 ]
5050 str texels_67, [ block_ptr, #12 ]
5051 add block_ptr, block_ptr, #64
5055 ldmia sp!, { r4 - r11, pc }
5058 #undef width_rounded
5061 #undef texture_offset
5070 #define left_offset r8
5071 #define width_rounded r9
5072 #define right_width r10
5073 #define block_width r11
5075 #define texture_offset_base r1
5076 #define texture_mask r2
5077 #define texture_page_ptr r3
5078 #define num_blocks r4
5081 #define texture_offset r8
5082 #define blocks_remaining r9
5083 #define fb_ptr_pitch r12
5084 #define texture_block_ptr r14
5086 #define texture_mask_width r2
5087 #define texture_mask_height r3
5088 #define left_mask_bits r4
5089 #define right_mask_bits r5
5093 #undef block_masks_shifted
5096 #define block_masks d0
5097 #define block_masks_shifted d1
5098 #define draw_mask_fb_ptr d2
5102 setup_sprites_16bpp_flush_single:
5105 stmdb sp!, { r0 - r3, r12, r14 }
5106 bl flush_render_block_buffer
5107 ldmia sp!, { r0 - r3, r12, r14 }
5111 add block, psx_gpu, #psx_gpu_blocks_offset
5116 setup_sprites_16bpp_flush_row:
5119 stmdb sp!, { r0 - r3, r12, r14 }
5120 bl flush_render_block_buffer
5121 ldmia sp!, { r0 - r3, r12, r14 }
5125 add block, psx_gpu, #psx_gpu_blocks_offset
5126 mov num_blocks, block_width
5130 function(setup_sprite_16bpp)
5131 stmdb sp!, { r4 - r11, r14 }
5132 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5135 add fb_ptr, fb_ptr, y, lsl #11
5137 ldr width, [ sp, #40 ]
5138 add fb_ptr, fb_ptr, x, lsl #1
5140 ldr height, [ sp, #44 ]
5141 and left_offset, u, #0x7
5143 add texture_offset_base, u, u
5144 add width_rounded, width, #7
5146 add texture_offset_base, v, lsl #11
5147 mov left_mask_bits, #0xFF
5149 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5150 add width_rounded, width_rounded, left_offset
5152 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5153 sub fb_ptr, fb_ptr, left_offset, lsl #1
5155 add texture_mask, texture_mask_width, texture_mask_width
5156 mov right_mask_bits, #0xFE
5158 and right_width, width_rounded, #0x7
5159 mvn left_mask_bits, left_mask_bits, lsl left_offset
5161 add texture_mask, texture_mask_height, lsl #11
5162 mov block_width, width_rounded, lsr #3
5164 mov right_mask_bits, right_mask_bits, lsl right_width
5165 movw fb_ptr_pitch, #(2048 + 16)
5167 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5168 vmov block_masks, left_mask_bits, right_mask_bits
5170 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5171 add block, psx_gpu, #psx_gpu_blocks_offset
5173 bic texture_offset_base, texture_offset_base, #0xF
5176 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5177 add block, block, num_blocks, lsl #6
5181 vext.32 block_masks_shifted, block_masks, block_masks, #1
5182 vorr.u32 block_masks, block_masks, block_masks_shifted
5183 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5186 add num_blocks, num_blocks, #1
5187 cmp num_blocks, #MAX_BLOCKS
5188 blgt setup_sprites_16bpp_flush_single
5190 and texture_block_ptr, texture_offset_base, texture_mask
5191 subs height, height, #1
5193 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5194 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5196 vst1.u32 { texels }, [ block, :128 ]
5197 add block, block, #40
5199 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5202 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5204 add block, block, #24
5205 add texture_offset_base, texture_offset_base, #2048
5206 add fb_ptr, fb_ptr, #2048
5207 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5210 ldmia sp!, { r4 - r11, pc }
5213 add num_blocks, num_blocks, block_width
5214 mov texture_offset, texture_offset_base
5216 cmp num_blocks, #MAX_BLOCKS
5217 blgt setup_sprites_16bpp_flush_row
5219 add texture_offset_base, texture_offset_base, #2048
5220 and texture_block_ptr, texture_offset, texture_mask
5222 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5223 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5225 vst1.u32 { texels }, [ block, :128 ]
5226 add block, block, #40
5228 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5229 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5232 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5233 subs blocks_remaining, block_width, #2
5235 add texture_offset, texture_offset, #16
5236 add fb_ptr, fb_ptr, #16
5238 vmov.u8 draw_mask_fb_ptr, #0
5240 add block, block, #24
5244 and texture_block_ptr, texture_offset, texture_mask
5245 subs blocks_remaining, blocks_remaining, #1
5247 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5248 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5250 vst1.u32 { texels }, [ block, :128 ]
5251 add block, block, #40
5253 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5256 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5258 add texture_offset, texture_offset, #16
5259 add fb_ptr, fb_ptr, #16
5261 add block, block, #24
5265 and texture_block_ptr, texture_offset, texture_mask
5266 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5268 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5269 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5271 vst1.u32 { texels }, [ block, :128 ]
5272 add block, block, #40
5274 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5275 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5277 add block, block, #24
5278 subs height, height, #1
5280 add fb_ptr, fb_ptr, fb_ptr_pitch
5281 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5285 ldmia sp!, { r4 - r11, pc }
5288 #undef texture_page_ptr
5290 #undef dirty_textures_mask
5291 #undef current_texture_mask
5294 #define current_texture_page r1
5295 #define texture_page_ptr r2
5296 #define vram_ptr_a r3
5297 #define current_texture_page_x r12
5298 #define current_texture_page_y r4
5299 #define dirty_textures_mask r5
5303 #define current_texture_mask r9
5305 #define vram_ptr_b r11
5307 #define texel_block_a d0
5308 #define texel_block_b d1
5309 #define texel_block_expanded_a q1
5310 #define texel_block_expanded_b q2
5311 #define texel_block_expanded_ab q2
5312 #define texel_block_expanded_c q3
5313 #define texel_block_expanded_d q4
5314 #define texel_block_expanded_cd q3
5316 function(update_texture_4bpp_cache)
5317 stmdb sp!, { r4 - r11, r14 }
5320 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5322 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5323 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5325 and current_texture_page_x, current_texture_page, #0xF
5326 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
5328 mov current_texture_page_y, current_texture_page, lsr #4
5329 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5331 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5334 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5335 bic dirty_textures_mask, current_texture_mask
5338 str dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5343 add vram_ptr_b, vram_ptr_a, #2048
5346 vld1.u32 { texel_block_a }, [ vram_ptr_a, :64 ], c_4096
5347 vld1.u32 { texel_block_b }, [ vram_ptr_b, :64 ], c_4096
5349 vmovl.u8 texel_block_expanded_a, texel_block_a
5350 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5351 vmovl.u8 texel_block_expanded_c, texel_block_b
5352 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5354 vbic.u16 texel_block_expanded_a, #0x00F0
5355 vbic.u16 texel_block_expanded_b, #0x00F0
5356 vbic.u16 texel_block_expanded_c, #0x00F0
5357 vbic.u16 texel_block_expanded_d, #0x00F0
5359 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
5360 texel_block_expanded_b
5361 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
5362 texel_block_expanded_d
5364 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
5365 [ texture_page_ptr, :256 ]!
5367 subs sub_y, sub_y, #1
5371 add vram_ptr_a, vram_ptr_a, #8
5372 add vram_ptr_b, vram_ptr_b, #8
5374 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5375 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5377 subs tile_x, tile_x, #1
5381 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5382 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5384 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5385 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5387 subs tile_y, tile_y, #1
5391 ldmia sp!, { r4 - r11, pc }
5394 #undef current_texture_page
5397 #define texture_page r1
5398 #define texture_page_ptr r2
5399 #define vram_ptr_a r3
5400 #define texture_page_x r12
5401 #define texture_page_y r4
5402 #define current_texture_page r5
5407 #define vram_ptr_b r11
5419 function(update_texture_8bpp_cache_slice)
5420 stmdb sp!, { r4 - r11, r14 }
5423 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5424 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5426 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5429 and texture_page_x, texture_page, #0xF
5430 mov texture_page_y, texture_page, lsr #4
5432 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
5435 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
5436 eor current_texture_page, current_texture_page, texture_page
5438 ands current_texture_page, current_texture_page, #0x1
5441 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
5444 add vram_ptr_b, vram_ptr_a, #2048
5447 vld1.u32 { texels_a }, [ vram_ptr_a, :128 ], c_4096
5448 vld1.u32 { texels_b }, [ vram_ptr_b, :128 ], c_4096
5449 vld1.u32 { texels_c }, [ vram_ptr_a, :128 ], c_4096
5450 vld1.u32 { texels_d }, [ vram_ptr_b, :128 ], c_4096
5452 vst1.u32 { texels_a, texels_b }, [ texture_page_ptr, :256 ]!
5453 vst1.u32 { texels_c, texels_d }, [ texture_page_ptr, :256 ]!
5455 subs sub_y, sub_y, #1
5460 add vram_ptr_a, vram_ptr_a, #16
5461 add vram_ptr_b, vram_ptr_b, #16
5463 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5464 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5466 subs tile_x, tile_x, #1
5471 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5472 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5474 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5475 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5477 subs tile_y, tile_y, #1
5478 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
5483 ldmia sp!, { r4 - r11, pc }