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
2006 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2009 ldmia sp!, { r4 - r11, pc }
2012 vst1.u32 { colors }, [ fb_ptr ]
2019 #define rg_dx_ptr r2
2037 #undef dither_offsets
2057 #define r_whole_low d6
2058 #define r_whole_high d7
2059 #define g_whole_low d8
2060 #define g_whole_high d9
2061 #define b_whole_low d10
2062 #define b_whole_high d11
2064 #define gb_whole_8 q6
2066 #define g_whole_8 d12
2067 #define b_whole_8 d13
2069 #define r_whole_8 d14
2081 #define block_span q5
2089 #define d128_0x7 q13
2093 #define dither_offsets q14
2094 #define draw_mask q15
2096 #define dither_offsets_low d28
2099 #define test_mask q10
2102 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2103 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2104 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2106 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2107 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2108 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2110 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2112 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2115 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2118 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2119 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2120 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2122 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2124 cmp num_spans, #0; \
2127 stmdb sp!, { r4 - r11, r14 }; \
2128 vshl.u32 rg_dx4, rg_dx, #2; \
2130 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2131 vshl.u32 rg_dx8, rg_dx, #3; \
2133 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2135 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2136 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2138 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2139 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2141 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2142 vmov.u8 d64_1, #1; \
2144 vmov.u8 d128_4, #4; \
2145 vmov.u8 d64_128, #128; \
2147 vmov.u8 d128_0x7, #0x7; \
2150 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2151 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2153 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2154 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
2156 cmp span_num_blocks, #0; \
2159 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2160 add num_blocks, span_num_blocks, num_blocks; \
2162 cmp num_blocks, #MAX_BLOCKS; \
2166 ldr b, [ span_b_offset ]; \
2167 add fb_ptr, fb_ptr, y, lsl #11; \
2169 vdup.u32 v_left_x, left_x; \
2172 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2173 add fb_ptr, fb_ptr, left_x, lsl #1; \
2175 mla b, b_dx, left_x, b; \
2176 and dither_shift, left_x, #0x03; \
2178 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2179 vshr.u32 rg_dx, rg_dx4, #2; \
2181 mov dither_shift, dither_shift, lsl #3; \
2182 vmla.u32 rg, rg_dx, v_left_x; \
2185 subs span_num_blocks, span_num_blocks, #1; \
2187 mov dither_row, dither_row, ror dither_shift; \
2188 mov b_dx4, b_dx, lsl #2; \
2190 vdup.u32 dither_offsets, dither_row; \
2191 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2193 vdup.u32 b_block, b; \
2194 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2196 mov b_dx8, b_dx, lsl #3; \
2197 vdup.u32 r_block, rg[0]; \
2198 vdup.u32 g_block, rg[1]; \
2200 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2202 vadd.u32 r_block, r_block, block_span; \
2203 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2205 vadd.u32 g_block, g_block, block_span; \
2206 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2208 vadd.u32 b_block, b_block, block_span; \
2209 add block_ptr_b, block_ptr_a, #16; \
2211 vshrn.u32 r_whole_low, r_block, #16; \
2212 vshrn.u32 g_whole_low, g_block, #16; \
2213 vshrn.u32 b_whole_low, b_block, #16; \
2214 vdup.u32 dx4, rg_dx4[0]; \
2216 vaddhn.u32 r_whole_high, r_block, dx4; \
2217 vdup.u32 dx4, rg_dx4[1]; \
2219 vaddhn.u32 g_whole_high, g_block, dx4; \
2220 vdup.u32 dx4, b_dx4; \
2222 vaddhn.u32 b_whole_high, b_block, dx4; \
2223 vdup.u32 dx8, rg_dx8[0]; \
2225 vadd.u32 r_block, r_block, dx8; \
2226 vdup.u32 dx8, rg_dx8[1]; \
2228 vadd.u32 g_block, g_block, dx8; \
2229 vdup.u32 dx8, b_dx8; \
2231 vadd.u32 b_block, b_block, dx8; \
2233 vmovn.u16 r_whole_8, r_whole; \
2234 vmovn.u16 g_whole_8, g_whole; \
2235 vmovn.u16 b_whole_8, b_whole; \
2238 veor.u32 draw_mask, draw_mask, draw_mask; \
2241 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2242 vshrn.u32 r_whole_low, r_block, #16; \
2244 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2245 vshrn.u32 g_whole_low, g_block, #16; \
2247 vshrn.u32 b_whole_low, b_block, #16; \
2248 str fb_ptr, [ block_ptr_a, #44 ]; \
2250 vdup.u32 dx4, rg_dx4[0]; \
2251 vshr.u8 r_whole_8, r_whole_8, #3; \
2252 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2254 vaddhn.u32 r_whole_high, r_block, dx4; \
2255 vdup.u32 dx4, rg_dx4[1]; \
2257 vaddhn.u32 g_whole_high, g_block, dx4; \
2258 vdup.u32 dx4, b_dx4; \
2260 vaddhn.u32 b_whole_high, b_block, dx4; \
2261 vdup.u32 dx8, rg_dx8[0]; \
2263 vmull.u8 pixels, r_whole_8, d64_1; \
2264 vmlal.u8 pixels, g_whole_8, d64_4; \
2265 vmlal.u8 pixels, b_whole_8, d64_128; \
2267 vadd.u32 r_block, r_block, dx8; \
2268 vdup.u32 dx8, rg_dx8[1]; \
2270 vadd.u32 g_block, g_block, dx8; \
2271 vdup.u32 dx8, b_dx8; \
2273 vadd.u32 b_block, b_block, dx8; \
2274 add fb_ptr, fb_ptr, #16; \
2276 vmovn.u16 r_whole_8, r_whole; \
2277 vmovn.u16 g_whole_8, g_whole; \
2278 vmovn.u16 b_whole_8, b_whole; \
2280 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2281 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2285 subs span_num_blocks, span_num_blocks, #1; \
2289 str fb_ptr, [ block_ptr_a, #44 ]; \
2290 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2292 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2293 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2295 vshr.u8 r_whole_8, r_whole_8, #3; \
2296 vdup.u8 draw_mask, right_mask; \
2298 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2299 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
2301 vtst.u16 draw_mask, draw_mask, test_mask; \
2303 vmull.u8 pixels, r_whole_8, d64_1; \
2304 vmlal.u8 pixels, g_whole_8, d64_4; \
2305 vmlal.u8 pixels, b_whole_8, d64_128; \
2307 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2308 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2311 add span_uvrg_offset, span_uvrg_offset, #16; \
2312 add span_b_offset, span_b_offset, #4; \
2314 add span_edge_data, span_edge_data, #8; \
2315 subs num_spans, num_spans, #1; \
2317 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2320 ldmia sp!, { r4 - r11, pc }; \
2323 /* TODO: Load from psx_gpu instead of saving/restoring these */\
2326 stmdb sp!, { r0 - r3, r12, r14 }; \
2327 bl flush_render_block_buffer; \
2328 ldmia sp!, { r0 - r3, r12, r14 }; \
2332 vmov.u8 d64_1, #1; \
2333 vmov.u8 d128_4, #4; \
2334 vmov.u8 d64_128, #128; \
2335 vmov.u8 d128_0x7, #0x7; \
2337 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2339 mov num_blocks, span_num_blocks; \
2340 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2344 setup_blocks_shaded_untextured_indirect_builder(undithered)
2345 setup_blocks_shaded_untextured_indirect_builder(dithered)
2350 #define mask_msb_ptr r14
2352 #define draw_mask q0
2353 #define pixels_low d16
2354 #define pixels_high d17
2358 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2361 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2362 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2363 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2365 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2367 cmp num_spans, #0; \
2370 stmdb sp!, { r4 - r11, r14 }; \
2371 vshl.u32 rg_dx4, rg_dx, #2; \
2373 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2374 vshl.u32 rg_dx8, rg_dx, #3; \
2376 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2377 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2379 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2380 vmov.u8 d64_1, #1; \
2382 vmov.u8 d128_4, #4; \
2383 vmov.u8 d64_128, #128; \
2385 vmov.u8 d128_0x7, #0x7; \
2386 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2387 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
2390 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2391 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2393 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2394 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
2396 cmp span_num_blocks, #0; \
2399 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2400 add fb_ptr, fb_ptr, y, lsl #11; \
2402 ldr b, [ span_b_offset ]; \
2403 vdup.u32 v_left_x, left_x; \
2406 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2407 add fb_ptr, fb_ptr, left_x, lsl #1; \
2409 mla b, b_dx, left_x, b; \
2410 and dither_shift, left_x, #0x03; \
2412 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2413 vshr.u32 rg_dx, rg_dx4, #2; \
2415 mov dither_shift, dither_shift, lsl #3; \
2416 vmla.u32 rg, rg_dx, v_left_x; \
2418 subs span_num_blocks, span_num_blocks, #1; \
2420 mov dither_row, dither_row, ror dither_shift; \
2421 mov b_dx4, b_dx, lsl #2; \
2423 vdup.u32 dither_offsets, dither_row; \
2424 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2426 vdup.u32 b_block, b; \
2427 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2429 mov b_dx8, b_dx, lsl #3; \
2430 vdup.u32 r_block, rg[0]; \
2431 vdup.u32 g_block, rg[1]; \
2433 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2435 vadd.u32 r_block, r_block, block_span; \
2436 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2438 vadd.u32 g_block, g_block, block_span; \
2439 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2441 vadd.u32 b_block, b_block, block_span; \
2442 add block_ptr_b, block_ptr_a, #16; \
2444 vshrn.u32 r_whole_low, r_block, #16; \
2445 vshrn.u32 g_whole_low, g_block, #16; \
2446 vshrn.u32 b_whole_low, b_block, #16; \
2447 vdup.u32 dx4, rg_dx4[0]; \
2449 vaddhn.u32 r_whole_high, r_block, dx4; \
2450 vdup.u32 dx4, rg_dx4[1]; \
2452 vaddhn.u32 g_whole_high, g_block, dx4; \
2453 vdup.u32 dx4, b_dx4; \
2455 vaddhn.u32 b_whole_high, b_block, dx4; \
2456 vdup.u32 dx8, rg_dx8[0]; \
2458 vadd.u32 r_block, r_block, dx8; \
2459 vdup.u32 dx8, rg_dx8[1]; \
2461 vadd.u32 g_block, g_block, dx8; \
2462 vdup.u32 dx8, b_dx8; \
2464 vadd.u32 b_block, b_block, dx8; \
2466 vmovn.u16 r_whole_8, r_whole; \
2467 vmovn.u16 g_whole_8, g_whole; \
2468 vmovn.u16 b_whole_8, b_whole; \
2473 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2474 vshrn.u32 r_whole_low, r_block, #16; \
2476 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2477 vshrn.u32 g_whole_low, g_block, #16; \
2479 vshrn.u32 b_whole_low, b_block, #16; \
2481 vdup.u32 dx4, rg_dx4[0]; \
2482 vshr.u8 r_whole_8, r_whole_8, #3; \
2483 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2485 vaddhn.u32 r_whole_high, r_block, dx4; \
2486 vdup.u32 dx4, rg_dx4[1]; \
2488 vmov pixels, msb_mask; \
2489 vaddhn.u32 g_whole_high, g_block, dx4; \
2490 vdup.u32 dx4, b_dx4; \
2492 vaddhn.u32 b_whole_high, b_block, dx4; \
2493 vdup.u32 dx8, rg_dx8[0]; \
2495 vmlal.u8 pixels, r_whole_8, d64_1; \
2496 vmlal.u8 pixels, g_whole_8, d64_4; \
2497 vmlal.u8 pixels, b_whole_8, d64_128; \
2499 vadd.u32 r_block, r_block, dx8; \
2500 vdup.u32 dx8, rg_dx8[1]; \
2502 vadd.u32 g_block, g_block, dx8; \
2503 vdup.u32 dx8, b_dx8; \
2505 vadd.u32 b_block, b_block, dx8; \
2507 vmovn.u16 r_whole_8, r_whole; \
2508 vmovn.u16 g_whole_8, g_whole; \
2509 vmovn.u16 b_whole_8, b_whole; \
2511 vst1.u32 { pixels }, [ fb_ptr ]!; \
2512 subs span_num_blocks, span_num_blocks, #1; \
2516 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2518 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2519 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2521 vshr.u8 r_whole_8, r_whole_8, #3; \
2522 rbit right_mask, right_mask; \
2523 vmov pixels, msb_mask; \
2524 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2525 clz right_mask, right_mask; \
2527 vmlal.u8 pixels, r_whole_8, d64_1; \
2528 vmlal.u8 pixels, g_whole_8, d64_4; \
2529 vmlal.u8 pixels, b_whole_8, d64_128; \
2531 ldr pc, [ pc, right_mask, lsl #2 ]; \
2544 vst1.u16 { pixels_low[0] }, [ fb_ptr ]; \
2548 vst1.u32 { pixels_low[0] }, [ fb_ptr ]; \
2552 vst1.u32 { pixels_low[0] }, [ fb_ptr ]!; \
2553 vst1.u16 { pixels_low[2] }, [ fb_ptr ]; \
2557 vst1.u32 { pixels_low }, [ fb_ptr ]; \
2561 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2562 vst1.u16 { pixels_high[0] }, [ fb_ptr ]; \
2566 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2567 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2571 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2572 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2573 vst1.u16 { pixels_high[2] }, [ fb_ptr ]; \
2577 vst1.u32 { pixels }, [ fb_ptr ]; \
2581 add span_uvrg_offset, span_uvrg_offset, #16; \
2582 add span_b_offset, span_b_offset, #4; \
2584 add span_edge_data, span_edge_data, #8; \
2585 subs num_spans, num_spans, #1; \
2589 ldmia sp!, { r4 - r11, pc } \
2591 setup_blocks_shaded_untextured_direct_builder(undithered)
2592 setup_blocks_shaded_untextured_direct_builder(dithered)
2601 #define block_ptr r1
2602 #define num_blocks r2
2615 #define texture_ptr r11
2629 #define pixels_d r10
2633 #define clut_ptr r12
2634 #define current_texture_mask r5
2635 #define dirty_textures_mask r6
2639 #define clut_low_a d2
2640 #define clut_low_b d3
2641 #define clut_high_a d4
2642 #define clut_high_b d5
2647 #define texels_low d6
2648 #define texels_high d7
2652 function(texture_blocks_untextured)
2658 function(texture_blocks_4bpp)
2659 stmdb sp!, { r3 - r11, r14 }
2660 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2662 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2663 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2665 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2666 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]
2668 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2669 vuzp.u8 clut_a, clut_b
2671 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
2672 tst dirty_textures_mask, current_texture_mask
2678 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2680 uxtah uv_0, texture_ptr, uv_01
2681 uxtah uv_1, texture_ptr, uv_01, ror #16
2683 uxtah uv_2, texture_ptr, uv_23
2684 uxtah uv_3, texture_ptr, uv_23, ror #16
2686 uxtah uv_4, texture_ptr, uv_45
2687 ldrb pixel_0, [ uv_0 ]
2689 uxtah uv_5, texture_ptr, uv_45, ror #16
2690 ldrb pixel_1, [ uv_1 ]
2692 uxtah uv_6, texture_ptr, uv_67
2693 ldrb pixel_2, [ uv_2 ]
2695 uxtah uv_7, texture_ptr, uv_67, ror #16
2696 ldrb pixel_3, [ uv_3 ]
2698 ldrb pixel_4, [ uv_4 ]
2699 subs num_blocks, num_blocks, #1
2701 ldrb pixel_5, [ uv_5 ]
2702 orr pixels_a, pixel_0, pixel_1, lsl #8
2704 ldrb pixel_6, [ uv_6 ]
2705 orr pixels_b, pixel_4, pixel_5, lsl #8
2707 ldrb pixel_7, [ uv_7 ]
2708 orr pixels_a, pixels_a, pixel_2, lsl #16
2710 orr pixels_b, pixels_b, pixel_6, lsl #16
2711 orr pixels_a, pixels_a, pixel_3, lsl #24
2713 orr pixels_b, pixels_b, pixel_7, lsl #24
2714 vmov.u32 texels, pixels_a, pixels_b
2716 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2717 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2719 vst2.u8 { texels_low, texels_high }, [ block_ptr, :128 ], c_64
2722 ldmia sp!, { r3 - r11, pc }
2725 stmdb sp!, { r1 - r2 }
2726 bl update_texture_4bpp_cache
2729 ldmia sp!, { r1 - r2 }
2735 function(texture_blocks_8bpp)
2736 stmdb sp!, { r3 - r11, r14 }
2737 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2739 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2740 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2742 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2743 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2745 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]
2746 tst dirty_textures_mask, current_texture_mask
2752 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2754 uxtah uv_0, texture_ptr, uv_01
2755 uxtah uv_1, texture_ptr, uv_01, ror #16
2757 uxtah uv_2, texture_ptr, uv_23
2758 uxtah uv_3, texture_ptr, uv_23, ror #16
2760 uxtah uv_4, texture_ptr, uv_45
2761 ldrb pixel_0, [ uv_0 ]
2763 uxtah uv_5, texture_ptr, uv_45, ror #16
2764 ldrb pixel_1, [ uv_1 ]
2766 uxtah uv_6, texture_ptr, uv_67
2767 ldrb pixel_2, [ uv_2 ]
2769 uxtah uv_7, texture_ptr, uv_67, ror #16
2770 ldrb pixel_3, [ uv_3 ]
2772 ldrb pixel_4, [ uv_4 ]
2773 add pixel_0, pixel_0, pixel_0
2775 ldrb pixel_5, [ uv_5 ]
2776 add pixel_1, pixel_1, pixel_1
2778 ldrb pixel_6, [ uv_6 ]
2779 add pixel_2, pixel_2, pixel_2
2781 ldrb pixel_7, [ uv_7 ]
2782 add pixel_3, pixel_3, pixel_3
2784 ldrh pixel_0, [ clut_ptr, pixel_0 ]
2785 add pixel_4, pixel_4, pixel_4
2787 ldrh pixel_1, [ clut_ptr, pixel_1 ]
2788 add pixel_5, pixel_5, pixel_5
2790 ldrh pixel_2, [ clut_ptr, pixel_2 ]
2791 add pixel_6, pixel_6, pixel_6
2793 ldrh pixel_3, [ clut_ptr, pixel_3 ]
2794 add pixel_7, pixel_7, pixel_7
2796 ldrh pixel_4, [ clut_ptr, pixel_4 ]
2797 orr pixels_a, pixel_0, pixel_1, lsl #16
2799 ldrh pixel_5, [ clut_ptr, pixel_5 ]
2800 orr pixels_c, pixel_2, pixel_3, lsl #16
2802 ldrh pixel_6, [ clut_ptr, pixel_6 ]
2803 subs num_blocks, num_blocks, #1
2805 ldrh pixel_7, [ clut_ptr, pixel_7 ]
2806 orr pixels_b, pixel_4, pixel_5, lsl #16
2808 orr pixels_d, pixel_6, pixel_7, lsl #16
2809 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2811 add block_ptr, block_ptr, #64
2814 ldmia sp!, { r3 - r11, pc }
2817 stmdb sp!, { r1 - r2, r12 }
2819 bl update_texture_8bpp_cache
2821 ldmia sp!, { r1 - r2, r12 }
2851 #define block_ptr r1
2852 #define num_blocks r2
2896 #define texture_ptr r12
2901 function(texture_blocks_16bpp)
2902 stmdb sp!, { r3 - r11, r14 }
2903 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2905 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2906 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2909 ldrh uv_0, [ block_ptr ]
2910 subs num_blocks, num_blocks, #1
2912 ldrh uv_1, [ block_ptr, #2 ]
2914 and v_0, uv_0, #0xFF00
2915 and v_1, uv_1, #0xFF00
2917 and u_0, uv_0, #0xFF
2918 and u_1, uv_1, #0xFF
2920 add uv_0, u_0, v_0, lsl #2
2921 ldrh uv_2, [ block_ptr, #4 ]
2923 add uv_1, u_1, v_1, lsl #2
2924 ldrh uv_3, [ block_ptr, #6 ]
2926 add uv_0, uv_0, uv_0
2927 add uv_1, uv_1, uv_1
2929 and v_2, uv_2, #0xFF00
2930 and v_3, uv_3, #0xFF00
2932 and u_2, uv_2, #0xFF
2933 and u_3, uv_3, #0xFF
2935 add uv_2, u_2, v_2, lsl #2
2936 ldrh uv_4, [ block_ptr, #8 ]
2938 add uv_3, u_3, v_3, lsl #2
2939 ldrh uv_5, [ block_ptr, #10 ]
2941 add uv_2, uv_2, uv_2
2942 add uv_3, uv_3, uv_3
2944 and v_4, uv_4, #0xFF00
2945 and v_5, uv_5, #0xFF00
2947 and u_4, uv_4, #0xFF
2948 and u_5, uv_5, #0xFF
2950 add uv_4, u_4, v_4, lsl #2
2951 ldrh uv_6, [ block_ptr, #12 ]
2953 add uv_5, u_5, v_5, lsl #2
2954 ldrh uv_7, [ block_ptr, #14 ]
2956 add uv_4, uv_4, uv_4
2957 ldrh pixel_0, [ texture_ptr, uv_0 ]
2959 add uv_5, uv_5, uv_5
2960 ldrh pixel_1, [ texture_ptr, uv_1 ]
2962 and v_6, uv_6, #0xFF00
2963 ldrh pixel_2, [ texture_ptr, uv_2 ]
2965 and v_7, uv_7, #0xFF00
2966 ldrh pixel_3, [ texture_ptr, uv_3 ]
2968 and u_6, uv_6, #0xFF
2969 ldrh pixel_4, [ texture_ptr, uv_4 ]
2971 and u_7, uv_7, #0xFF
2972 ldrh pixel_5, [ texture_ptr, uv_5 ]
2974 add uv_6, u_6, v_6, lsl #2
2975 add uv_7, u_7, v_7, lsl #2
2977 add uv_6, uv_6, uv_6
2978 add uv_7, uv_7, uv_7
2980 orr pixels_a, pixel_0, pixel_1, lsl #16
2981 orr pixels_b, pixel_2, pixel_3, lsl #16
2983 ldrh pixel_6, [ texture_ptr, uv_6 ]
2984 orr pixels_c, pixel_4, pixel_5, lsl #16
2986 ldrh pixel_7, [ texture_ptr, uv_7 ]
2987 orr pixels_d, pixel_6, pixel_7, lsl #16
2989 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
2990 add block_ptr, block_ptr, #64
2994 ldmia sp!, { r3 - r11, pc }
3009 #undef msb_mask_high
3017 #define num_blocks r1
3018 #define color_ptr r2
3019 #define colors_scalar r2
3020 #define colors_scalar_compare r3
3021 #define mask_msb_ptr r2
3023 #define block_ptr_load_a r0
3024 #define block_ptr_store r3
3025 #define block_ptr_load_b r12
3030 #define draw_mask_bits_scalar r5
3032 #define d128_0x07 q0
3033 #define d128_0x1F q1
3034 #define d128_0x8000 q2
3035 #define test_mask q3
3037 #define colors_rg q5
3038 #define colors_b_dm_bits q6
3039 #define texels_rg q7
3042 #define pixels_b q10
3044 #define zero_mask q4
3045 #define draw_mask q12
3046 #define msb_mask q13
3048 #define fb_pixels q8
3050 #define pixels_gb_low q9
3052 #define colors_r d10
3053 #define colors_g d11
3054 #define colors_b d12
3055 #define draw_mask_bits d13
3056 #define texels_r d14
3057 #define texels_g d15
3058 #define pixels_r_low d16
3059 #define pixels_g_low d18
3060 #define pixels_b_low d19
3061 #define msb_mask_low d26
3062 #define msb_mask_high d27
3067 #define texels_b d31
3069 #define shade_blocks_textured_modulated_prologue_indirect() \
3071 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3073 #define shade_blocks_textured_modulated_prologue_direct() \
3074 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3075 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ] \
3078 #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \
3080 #define shade_blocks_textured_false_modulation_check_undithered(target) \
3081 ldr colors_scalar, [ psx_gpu, #psx_gpu_triangle_color_offset ]; \
3082 movw colors_scalar_compare, #0x8080; \
3084 movt colors_scalar_compare, #0x80; \
3085 cmp colors_scalar, colors_scalar_compare; \
3086 beq shade_blocks_textured_unmodulated_##target \
3088 #define shade_blocks_textured_false_modulation_check_dithered(target) \
3090 #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \
3091 shade_blocks_textured_false_modulation_check_##dithering(target); \
3092 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3093 vld1.u32 { colors_r[] }, [ color_ptr, :32 ]; \
3094 vdup.u8 colors_g, colors_r[1]; \
3095 vdup.u8 colors_b, colors_r[2]; \
3096 vdup.u8 colors_r, colors_r[0] \
3099 #define shade_blocks_textured_modulated_load_dithered(target) \
3100 vld1.u32 { target }, [ block_ptr_load_b, :128 ] \
3102 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3103 vld1.u32 { target }, [ block_ptr_load_b, :128 ], c_32 \
3105 #define shade_blocks_textured_modulated_load_undithered(target) \
3107 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3108 add block_ptr_load_b, block_ptr_load_b, #32 \
3110 #define shade_blocks_textured_modulate_dithered(channel) \
3111 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3113 #define shade_blocks_textured_modulate_undithered(channel) \
3114 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3117 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3118 vst1.u32 { draw_mask }, [ block_ptr_store, :128 ]! \
3120 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3121 ldr fb_ptr, [ block_ptr_load_b, #(offset - 64) ]; \
3122 vld1.u32 { fb_pixels }, [ fb_ptr ]; \
3123 vbit.u16 pixels, fb_pixels, draw_mask \
3125 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3126 vst1.u32 { pixels }, [ block_ptr_store, :128 ], c_48 \
3128 #define shade_blocks_textured_modulated_store_pixels_direct() \
3129 vst1.u32 { pixels }, [ fb_ptr ] \
3132 #define shade_blocks_textured_modulated_load_rg_shaded() \
3133 vld1.u32 { colors_r, colors_g }, [ block_ptr_load_b, :128 ], c_32 \
3135 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3136 add block_ptr_load_b, block_ptr_load_b, #32 \
3138 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3139 vld1.u32 { colors_b, draw_mask_bits }, [ block_ptr_load_a, :128 ], c_32 \
3141 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3142 ldr draw_mask_bits_scalar, [ block_ptr_load_a, #8 ]; \
3143 add block_ptr_load_a, block_ptr_load_a, #32 \
3145 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3146 vdup.u16 draw_mask, draw_mask_bits[0] \
3148 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3149 vdup.u16 draw_mask, draw_mask_bits_scalar \
3152 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3154 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3155 vorr.u16 pixels, pixels, msb_mask \
3158 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3161 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3162 shade_blocks_textured_modulated_prologue_##shading(dithering, target); \
3163 stmdb sp!, { r4 - r5, lr }; \
3164 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3166 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
3168 shade_blocks_textured_modulated_prologue_##target(); \
3170 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3173 add block_ptr_load_b, block_ptr_load_a, #16; \
3174 vmov.u8 d64_1, #1; \
3175 vmov.u8 d64_4, #4; \
3176 vmov.u8 d64_128, #128; \
3178 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3179 vmov.u8 d128_0x07, #0x07; \
3181 shade_blocks_textured_modulated_load_rg_##shading(); \
3182 vmov.u8 d128_0x1F, #0x1F; \
3184 shade_blocks_textured_modulated_load_bdm_##shading(); \
3185 vmov.u16 d128_0x8000, #0x8000; \
3187 vmovn.u16 texels_r, texels; \
3188 vshrn.u16 texels_g, texels, #5; \
3190 vshrn.u16 texels_b, texels, #7; \
3191 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3193 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3194 vtst.u16 draw_mask, draw_mask, test_mask; \
3196 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3197 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3199 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3200 vshr.u8 texels_b, texels_b, #3; \
3202 shade_blocks_textured_modulate_##dithering(r); \
3203 shade_blocks_textured_modulate_##dithering(g); \
3204 shade_blocks_textured_modulate_##dithering(b); \
3206 vand.u16 pixels, texels, d128_0x8000; \
3207 vceq.u16 zero_mask, texels, #0; \
3209 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3210 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3211 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3213 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3214 vorr.u16 draw_mask, draw_mask, zero_mask; \
3215 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3216 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3218 subs num_blocks, num_blocks, #1; \
3224 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3225 shade_blocks_textured_modulated_load_rg_##shading(); \
3226 vshrn.u16 texels_g, texels, #5; \
3228 shade_blocks_textured_modulated_load_bdm_##shading(); \
3229 vshrn.u16 texels_b, texels, #7; \
3231 vmovn.u16 texels_r, texels; \
3232 vmlal.u8 pixels, pixels_r_low, d64_1; \
3234 vmlal.u8 pixels, pixels_g_low, d64_4; \
3235 vmlal.u8 pixels, pixels_b_low, d64_128; \
3236 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3238 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3239 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3241 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3242 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3244 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3245 vtst.u16 draw_mask, draw_mask, test_mask; \
3247 shade_blocks_textured_modulated_store_pixels_##target(); \
3248 vshr.u8 texels_b, texels_b, #3; \
3250 shade_blocks_textured_modulate_##dithering(r); \
3251 shade_blocks_textured_modulate_##dithering(g); \
3252 shade_blocks_textured_modulate_##dithering(b); \
3254 vand.u16 pixels, texels, d128_0x8000; \
3255 vceq.u16 zero_mask, texels, #0; \
3257 subs num_blocks, num_blocks, #1; \
3259 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3260 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3261 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3263 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3264 vorr.u16 draw_mask, draw_mask, zero_mask; \
3265 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3266 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3271 vmlal.u8 pixels, pixels_r_low, d64_1; \
3272 vmlal.u8 pixels, pixels_g_low, d64_4; \
3273 vmlal.u8 pixels, pixels_b_low, d64_128; \
3275 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3276 shade_blocks_textured_modulated_store_pixels_##target(); \
3278 ldmia sp!, { r4 - r5, pc } \
3281 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3282 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3283 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3284 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3286 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3287 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3288 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3289 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3307 #undef msb_mask_high
3310 #define num_blocks r1
3311 #define mask_msb_ptr r2
3312 #define color_ptr r3
3314 #define block_ptr_load r0
3315 #define draw_mask_store_ptr r3
3316 #define draw_mask_bits_ptr r12
3317 #define draw_mask_ptr r12
3318 #define pixel_store_ptr r14
3320 #define fb_ptr_cmp r4
3323 #define fb_ptr_next r14
3327 #define test_mask q0
3329 #define draw_mask q2
3330 #define zero_mask q3
3331 #define draw_mask_combined q4
3332 #define fb_pixels q5
3333 #define fb_pixels_next q6
3336 #define draw_mask_low d4
3337 #define draw_mask_high d5
3338 #define msb_mask_low d14
3339 #define msb_mask_high d15
3342 function(shade_blocks_textured_unmodulated_indirect)
3343 str r14, [ sp, #-4 ]
3344 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3346 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3347 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3349 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3350 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3353 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3355 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3356 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3357 [ draw_mask_bits_ptr, :16 ], c_64
3358 vceq.u16 zero_mask, pixels, #0
3360 vtst.u16 draw_mask, draw_mask, test_mask
3361 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3363 subs num_blocks, num_blocks, #1
3367 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3368 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3370 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3371 [ draw_mask_bits_ptr, :16 ], c_64
3372 vceq.u16 zero_mask, pixels, #0
3374 vtst.u16 draw_mask, draw_mask, test_mask
3375 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3377 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3378 subs num_blocks, num_blocks, #1
3383 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3384 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3391 function(shade_blocks_textured_unmodulated_direct)
3392 stmdb sp!, { r4, r14 }
3393 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3395 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3396 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3398 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3401 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3402 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3404 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3405 [ draw_mask_bits_ptr, :16 ], c_64
3406 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3408 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3409 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3410 vceq.u16 zero_mask, pixels, #0
3411 vtst.u16 draw_mask, draw_mask, test_mask
3413 subs num_blocks, num_blocks, #1
3417 mov fb_ptr, fb_ptr_next
3418 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3420 vorr.u16 pixels, pixels, msb_mask
3422 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3423 vmov fb_pixels, fb_pixels_next
3425 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3426 [ draw_mask_bits_ptr, :16 ], c_64
3427 vbif.u16 fb_pixels, pixels, draw_mask_combined
3429 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3431 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3432 add fb_ptr_cmp, fb_ptr_cmp, #14
3436 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3437 vceq.u16 zero_mask, pixels, #0
3439 vst1.u16 { fb_pixels }, [ fb_ptr ]
3440 vtst.u16 draw_mask, draw_mask, test_mask
3443 subs num_blocks, num_blocks, #1
3447 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3448 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3450 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3452 ldmia sp!, { r4, pc }
3455 vst1.u16 { fb_pixels }, [ fb_ptr ]
3456 vceq.u16 zero_mask, pixels, #0
3458 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3459 vtst.u16 draw_mask, draw_mask, test_mask
3464 function(shade_blocks_unshaded_untextured_indirect)
3469 function(shade_blocks_unshaded_untextured_direct)
3470 stmdb sp!, { r4, r14 }
3471 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3473 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3474 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3476 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3477 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3479 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3480 vld1.u16 { pixels }, [ color_ptr, :128 ]
3483 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3485 vorr.u16 pixels, pixels, msb_mask
3486 subs num_blocks, num_blocks, #1
3488 ldr fb_ptr_next, [ block_ptr_load ], #64
3490 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3494 vmov fb_pixels, fb_pixels_next
3495 mov fb_ptr, fb_ptr_next
3496 ldr fb_ptr_next, [ block_ptr_load ], #64
3498 vbif.u16 fb_pixels, pixels, draw_mask
3499 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3501 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3502 add fb_ptr_cmp, fb_ptr_cmp, #14
3506 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3507 vst1.u16 { fb_pixels }, [ fb_ptr ]
3510 subs num_blocks, num_blocks, #1
3514 vbif.u16 fb_pixels_next, pixels, draw_mask
3515 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3517 ldmia sp!, { r4, pc }
3520 vst1.u16 { fb_pixels }, [ fb_ptr ]
3521 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3525 #undef draw_mask_ptr
3532 #define num_blocks r1
3533 #define msb_mask_ptr r2
3534 #define pixel_ptr r3
3535 #define draw_mask_ptr r0
3538 #define fb_ptr_next r14
3539 #define fb_ptr_cmp r4
3547 #undef msb_mask_high
3548 #undef draw_mask_next
3551 #undef fb_pixels_next
3554 #define draw_mask q1
3556 #define fb_pixels q3
3557 #define blend_pixels q4
3558 #define pixels_no_msb q5
3559 #define blend_mask q6
3560 #define fb_pixels_no_msb q7
3561 #define d128_0x8000 q8
3562 #define d128_0x0421 q9
3563 #define fb_pixels_next q10
3564 #define blend_pixels_next q11
3565 #define pixels_next q12
3566 #define draw_mask_next q13
3567 #define write_mask q14
3569 #define pixels_rb q5
3570 #define pixels_mg q7
3572 #define d128_0x7C1F q8
3573 #define d128_0x03E0 q9
3574 #define fb_pixels_rb q10
3575 #define fb_pixels_g q11
3576 #define fb_pixels_masked q11
3577 #define d128_0x83E0 q15
3578 #define pixels_fourth q7
3579 #define d128_0x1C07 q12
3580 #define d128_0x00E0 q13
3581 #define d128_0x80E0 q13
3583 #define msb_mask_low d0
3584 #define msb_mask_high d1
3586 #define blend_blocks_average_set_blend_mask_textured(source) \
3587 vclt.s16 blend_mask, source, #0 \
3589 #define blend_blocks_average_set_stp_bit_textured() \
3590 vorr.u16 blend_pixels, #0x8000 \
3592 #define blend_blocks_average_combine_textured(source) \
3593 vbif.u16 blend_pixels, source, blend_mask \
3595 #define blend_blocks_average_set_blend_mask_untextured(source) \
3597 #define blend_blocks_average_set_stp_bit_untextured() \
3599 #define blend_blocks_average_combine_untextured(source) \
3601 #define blend_blocks_average_mask_set_on() \
3602 vclt.s16 write_mask, fb_pixels_next, #0 \
3604 #define blend_blocks_average_mask_copy_on() \
3605 vorr.u16 draw_mask, draw_mask_next, write_mask \
3607 #define blend_blocks_average_mask_copy_b_on() \
3608 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3610 #define blend_blocks_average_mask_set_off() \
3612 #define blend_blocks_average_mask_copy_off() \
3613 vmov draw_mask, draw_mask_next \
3615 #define blend_blocks_average_mask_copy_b_off() \
3617 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3620 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3621 stmdb sp!, { r4, r14 }; \
3622 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3623 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3625 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3626 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3628 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3631 vmov.u16 d128_0x8000, #0x8000; \
3632 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3633 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3635 vmov.u16 d128_0x0421, #0x0400; \
3636 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3638 vorr.u16 d128_0x0421, #0x0021; \
3639 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3641 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3642 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3643 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3644 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3645 blend_blocks_average_mask_set_##mask_evaluate(); \
3646 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3648 subs num_blocks, num_blocks, #1; \
3652 mov fb_ptr, fb_ptr_next; \
3653 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3655 vmov pixels, pixels_next; \
3656 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3658 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3660 blend_blocks_average_mask_copy_##mask_evaluate(); \
3661 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3663 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3664 blend_blocks_average_set_stp_bit_##texturing(); \
3665 vmov fb_pixels, fb_pixels_next; \
3666 blend_blocks_average_combine_##texturing(pixels); \
3668 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3669 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3670 cmp fb_ptr_cmp, #28; \
3673 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3674 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3676 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3677 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3679 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3680 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3682 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3683 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3684 blend_blocks_average_mask_set_##mask_evaluate(); \
3685 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3688 subs num_blocks, num_blocks, #1; \
3692 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3693 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3695 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3696 blend_blocks_average_set_stp_bit_##texturing(); \
3697 blend_blocks_average_combine_##texturing(pixels_next); \
3699 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3700 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3701 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3703 ldmia sp!, { r4, pc }; \
3706 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3707 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3708 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3710 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3711 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3712 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3713 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3714 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3715 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3719 blend_blocks_average_builder(textured, off)
3720 blend_blocks_average_builder(untextured, off)
3721 blend_blocks_average_builder(textured, on)
3722 blend_blocks_average_builder(untextured, on)
3725 #define blend_blocks_add_mask_set_on() \
3726 vclt.s16 write_mask, fb_pixels, #0 \
3728 #define blend_blocks_add_mask_copy_on() \
3729 vorr.u16 draw_mask, draw_mask, write_mask \
3731 #define blend_blocks_add_mask_set_off() \
3733 #define blend_blocks_add_mask_copy_off() \
3736 #define blend_blocks_add_textured_builder(mask_evaluate) \
3739 function(blend_blocks_textured_add_##mask_evaluate) \
3740 stmdb sp!, { r4, r14 }; \
3741 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3742 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3744 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3745 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3747 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3750 vmov.u16 d128_0x7C1F, #0x7C00; \
3751 vmov.u16 d128_0x03E0, #0x0300; \
3752 vmov.u16 d128_0x83E0, #0x8000; \
3753 vorr.u16 d128_0x03E0, #0x00E0; \
3754 vorr.u16 d128_0x7C1F, #0x001F; \
3755 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3757 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3758 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3759 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3760 vclt.s16 blend_mask, pixels, #0; \
3761 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3762 blend_blocks_add_mask_set_##mask_evaluate(); \
3763 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3765 blend_blocks_add_mask_copy_##mask_evaluate(); \
3766 vorr.u16 pixels, pixels, msb_mask; \
3767 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3768 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3769 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3770 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3771 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3772 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3773 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3774 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3776 subs num_blocks, num_blocks, #1; \
3780 mov fb_ptr, fb_ptr_next; \
3782 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3784 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3785 vclt.s16 blend_mask, pixels, #0; \
3787 vorr.u16 pixels, pixels, msb_mask; \
3788 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3789 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3791 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3792 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3794 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3795 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3796 cmp fb_ptr_cmp, #28; \
3799 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3800 blend_blocks_add_mask_set_##mask_evaluate(); \
3801 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3802 blend_blocks_add_mask_copy_##mask_evaluate(); \
3803 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3804 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3805 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3808 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3809 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3810 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3811 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3812 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3814 subs num_blocks, num_blocks, #1; \
3818 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3819 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3820 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3822 ldmia sp!, { r4, pc }; \
3825 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3826 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3828 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3829 blend_blocks_add_mask_set_##mask_evaluate(); \
3830 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3831 blend_blocks_add_mask_copy_##mask_evaluate(); \
3832 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3836 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3839 function(blend_blocks_untextured_add_##mask_evaluate) \
3840 stmdb sp!, { r4, r14 }; \
3841 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3842 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3844 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3845 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3847 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3850 vmov.u16 d128_0x7C1F, #0x7C00; \
3851 vmov.u16 d128_0x03E0, #0x0300; \
3852 vorr.u16 d128_0x7C1F, #0x001F; \
3853 vorr.u16 d128_0x03E0, #0x00E0; \
3855 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3856 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3857 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3858 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3859 blend_blocks_add_mask_set_##mask_evaluate(); \
3860 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3862 blend_blocks_add_mask_copy_##mask_evaluate(); \
3863 vand.u16 pixels_g, pixels, d128_0x03E0; \
3864 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3865 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3866 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3867 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3868 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3869 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3871 subs num_blocks, num_blocks, #1; \
3875 mov fb_ptr, fb_ptr_next; \
3877 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3879 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3881 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3882 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3883 vand.u16 pixels_g, pixels, d128_0x03E0; \
3885 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3886 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3888 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3889 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3890 cmp fb_ptr_cmp, #28; \
3893 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3894 blend_blocks_add_mask_set_##mask_evaluate(); \
3895 blend_blocks_add_mask_copy_##mask_evaluate(); \
3896 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3897 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3898 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3901 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3902 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3903 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3904 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3905 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3907 subs num_blocks, num_blocks, #1; \
3911 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3912 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3913 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3914 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3916 ldmia sp!, { r4, pc }; \
3919 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3920 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3922 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3923 blend_blocks_add_mask_set_##mask_evaluate(); \
3924 blend_blocks_add_mask_copy_##mask_evaluate(); \
3925 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3929 blend_blocks_add_textured_builder(off)
3930 blend_blocks_add_textured_builder(on)
3931 blend_blocks_add_untextured_builder(off)
3932 blend_blocks_add_untextured_builder(on)
3934 #define blend_blocks_subtract_set_blend_mask_textured() \
3935 vclt.s16 blend_mask, pixels_next, #0 \
3937 #define blend_blocks_subtract_combine_textured() \
3938 vbif.u16 blend_pixels, pixels, blend_mask \
3940 #define blend_blocks_subtract_set_stb_textured() \
3941 vorr.u16 blend_pixels, #0x8000 \
3943 #define blend_blocks_subtract_msb_mask_textured() \
3944 vorr.u16 pixels, pixels_next, msb_mask \
3946 #define blend_blocks_subtract_set_blend_mask_untextured() \
3948 #define blend_blocks_subtract_combine_untextured() \
3950 #define blend_blocks_subtract_set_stb_untextured() \
3951 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3953 #define blend_blocks_subtract_msb_mask_untextured() \
3956 #define blend_blocks_subtract_mask_set_on() \
3957 vclt.s16 write_mask, fb_pixels, #0 \
3959 #define blend_blocks_subtract_mask_copy_on() \
3960 vorr.u16 draw_mask, draw_mask_next, write_mask \
3962 #define blend_blocks_subtract_mask_set_off() \
3964 #define blend_blocks_subtract_mask_copy_off() \
3965 vmov draw_mask, draw_mask_next \
3968 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
3971 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
3972 stmdb sp!, { r4, r14 }; \
3973 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3974 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3976 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3977 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3979 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3982 vmov.u16 d128_0x7C1F, #0x7C00; \
3983 vmov.u16 d128_0x03E0, #0x0300; \
3984 vorr.u16 d128_0x7C1F, #0x001F; \
3985 vorr.u16 d128_0x03E0, #0x00E0; \
3987 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3988 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3989 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3990 blend_blocks_subtract_set_blend_mask_##texturing(); \
3991 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3992 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3993 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3995 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3996 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3997 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3998 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3999 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4001 subs num_blocks, num_blocks, #1; \
4005 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4006 mov fb_ptr, fb_ptr_next; \
4007 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4009 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
4010 blend_blocks_subtract_msb_mask_##texturing(); \
4012 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
4013 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4014 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4015 blend_blocks_subtract_set_stb_##texturing(); \
4016 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4017 blend_blocks_subtract_combine_##texturing(); \
4018 blend_blocks_subtract_set_blend_mask_##texturing(); \
4019 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4021 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4022 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4023 cmp fb_ptr_cmp, #28; \
4026 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4027 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4028 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4029 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4030 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4031 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4032 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4035 subs num_blocks, num_blocks, #1; \
4039 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4041 blend_blocks_subtract_msb_mask_##texturing(); \
4042 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4043 blend_blocks_subtract_set_stb_##texturing(); \
4044 blend_blocks_subtract_combine_##texturing(); \
4045 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4046 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4048 ldmia sp!, { r4, pc }; \
4051 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4052 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4053 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4054 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4055 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4056 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4057 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4061 blend_blocks_subtract_builder(textured, off)
4062 blend_blocks_subtract_builder(textured, on)
4063 blend_blocks_subtract_builder(untextured, off)
4064 blend_blocks_subtract_builder(untextured, on)
4067 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4070 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4071 stmdb sp!, { r4, r14 }; \
4072 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4073 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4075 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4076 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4078 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4081 vmov.u16 d128_0x7C1F, #0x7C00; \
4082 vmov.u16 d128_0x03E0, #0x0300; \
4083 vmov.u16 d128_0x83E0, #0x8300; \
4084 vmov.u16 d128_0x1C07, #0x1C00; \
4085 vmov.u16 d128_0x80E0, #0x8000; \
4086 vorr.u16 d128_0x7C1F, #0x001F; \
4087 vorr.u16 d128_0x03E0, #0x00E0; \
4088 vorr.u16 d128_0x83E0, #0x00E0; \
4089 vorr.u16 d128_0x1C07, #0x0007; \
4090 vorr.u16 d128_0x80E0, #0x00E0; \
4092 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4093 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4094 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4095 vclt.s16 blend_mask, pixels, #0; \
4096 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4097 blend_blocks_add_mask_set_##mask_evaluate(); \
4098 vshr.s16 pixels_fourth, pixels, #2; \
4100 blend_blocks_add_mask_copy_##mask_evaluate(); \
4101 vorr.u16 pixels, pixels, msb_mask; \
4102 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4103 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
4104 vand.u16 pixels_mg, pixels_fourth, d128_0x80E0; \
4105 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
4106 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
4107 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4108 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
4109 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4110 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
4112 subs num_blocks, num_blocks, #1; \
4116 mov fb_ptr, fb_ptr_next; \
4118 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4120 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4121 vclt.s16 blend_mask, pixels, #0; \
4123 vshr.s16 pixels_fourth, pixels, #2; \
4124 vorr.u16 pixels, pixels, msb_mask; \
4125 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4126 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4128 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4129 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4131 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4132 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4133 cmp fb_ptr_cmp, #28; \
4136 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4137 blend_blocks_add_mask_set_##mask_evaluate(); \
4138 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
4139 blend_blocks_add_mask_copy_##mask_evaluate(); \
4140 vand.u16 pixels_mg, pixels_fourth, d128_0x80E0; \
4141 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
4142 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4145 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
4146 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4147 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
4148 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4149 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
4151 subs num_blocks, num_blocks, #1; \
4155 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4156 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4157 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4159 ldmia sp!, { r4, pc }; \
4162 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4163 vand.u16 pixels_mg, pixels_fourth, d128_0x80E0; \
4165 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4166 blend_blocks_add_mask_set_##mask_evaluate(); \
4167 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
4168 blend_blocks_add_mask_copy_##mask_evaluate(); \
4169 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
4173 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4176 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4177 stmdb sp!, { r4, r14 }; \
4178 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4179 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4181 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4182 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4184 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4187 vmov.u16 d128_0x7C1F, #0x7C00; \
4188 vmov.u16 d128_0x03E0, #0x0300; \
4189 vmov.u16 d128_0x83E0, #0x8300; \
4190 vmov.u16 d128_0x1C07, #0x1C00; \
4191 vmov.u16 d128_0x00E0, #0x00E0; \
4192 vorr.u16 d128_0x7C1F, #0x001F; \
4193 vorr.u16 d128_0x03E0, #0x00E0; \
4194 vorr.u16 d128_0x83E0, #0x00E0; \
4195 vorr.u16 d128_0x1C07, #0x0007; \
4197 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4198 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4199 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4200 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4201 blend_blocks_add_mask_set_##mask_evaluate(); \
4202 vshr.s16 pixels_fourth, pixels, #2; \
4203 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4205 blend_blocks_add_mask_copy_##mask_evaluate(); \
4206 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4207 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4208 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4209 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4210 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4211 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4212 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4214 subs num_blocks, num_blocks, #1; \
4218 mov fb_ptr, fb_ptr_next; \
4220 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4222 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4224 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4225 vshr.s16 pixels_fourth, pixels, #2; \
4226 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4227 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4229 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4230 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4232 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4233 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4234 cmp fb_ptr_cmp, #28; \
4237 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4238 blend_blocks_add_mask_set_##mask_evaluate(); \
4239 blend_blocks_add_mask_copy_##mask_evaluate(); \
4240 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4241 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4242 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4245 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4246 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4247 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4248 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4249 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4251 subs num_blocks, num_blocks, #1; \
4255 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4256 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4257 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4258 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4260 ldmia sp!, { r4, pc }; \
4263 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4264 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4266 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4267 blend_blocks_add_mask_set_##mask_evaluate(); \
4268 blend_blocks_add_mask_copy_##mask_evaluate(); \
4269 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4273 blend_blocks_add_fourth_textured_builder(off)
4274 blend_blocks_add_fourth_textured_builder(on)
4275 blend_blocks_add_fourth_untextured_builder(off)
4276 blend_blocks_add_fourth_untextured_builder(on)
4278 // TODO: Optimize this more. Need a scene that actually uses it for
4283 function(blend_blocks_textured_unblended_on)
4284 stmdb sp!, { r4, r14 }
4285 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4286 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
4288 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4289 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
4291 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4294 ldr fb_ptr, [ pixel_ptr, #28 ]
4295 vld1.u16 { fb_pixels }, [ fb_ptr ]
4296 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4297 vclt.s16 write_mask, fb_pixels, #0
4298 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4300 subs num_blocks, num_blocks, #1
4304 vorr.u16 draw_mask, draw_mask, write_mask
4305 vbif.u16 fb_pixels, pixels, draw_mask
4306 vst1.u16 { fb_pixels }, [ fb_ptr ]
4308 ldr fb_ptr, [ pixel_ptr, #28 ]
4309 vld1.u16 { fb_pixels }, [ fb_ptr ]
4310 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4311 vclt.s16 write_mask, fb_pixels, #0
4312 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4314 subs num_blocks, num_blocks, #1
4318 vorr.u16 draw_mask, draw_mask, write_mask
4319 vbif.u16 fb_pixels, pixels, draw_mask
4320 vst1.u16 { fb_pixels }, [ fb_ptr ]
4322 ldmia sp!, { r4, pc }
4325 function(blend_blocks_textured_unblended_off)
4335 vld1.u32 { u_whole_8, v_whole_8 }, [ r1, :128 ], r3
4355 #define parameter_width_offset 0
4356 #define parameter_height_offset 4
4362 #define left_unaligned r14
4363 #define right_unaligned r4
4365 #define num_unaligned r2
4366 #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 add left_unaligned, x, #0x7
4393 bic left_unaligned, left_unaligned, #0x7
4394 vdup.u16 colors, color
4396 sub left_unaligned, left_unaligned, x
4399 sub pitch, pitch, width, lsl #1
4400 sub width, width, left_unaligned
4402 and right_unaligned, width, #0x7
4403 bic width, width, #0x7
4406 mov num_width, width, lsr #3
4408 movs num_unaligned, left_unaligned
4412 strh color, [ vram_ptr ], #2
4414 subs num_unaligned, num_unaligned, #1
4418 vst1.u32 { colors }, [ vram_ptr, :128 ]!
4419 subs num_width, num_width, #1
4422 movs num_unaligned, right_unaligned
4426 strh color, [ vram_ptr ], #2
4428 subs num_unaligned, num_unaligned, #1
4432 add vram_ptr, vram_ptr, pitch
4433 subs height, height, #1
4436 ldmia sp!, { r4 - r6, pc }
4447 #undef dirty_textures_mask
4449 #undef current_texture_mask
4460 #define offset_u_right r10
4461 #define width_rounded r11
4462 #define height_rounded r12
4464 #define texture_offset_base r1
4465 #define tile_width r2
4466 #define tile_height r3
4467 #define num_blocks r4
4469 #define sub_tile_height r6
4471 #define texture_mask r8
4472 #define column_data r9
4473 #define texture_offset r10
4474 #define tiles_remaining r11
4475 #define fb_ptr_advance_column r12
4476 #define texture_block_ptr r14
4478 #define texture_page_ptr r3
4479 #define left_block_mask r4
4480 #define right_block_mask r5
4481 #define texture_mask_rev r10
4482 #define control_mask r11
4484 #define dirty_textures_mask r4
4486 #define current_texture_mask r6
4500 #define draw_masks_fb_ptrs q1
4502 #define draw_mask_fb_ptr_left d2
4503 #define draw_mask_fb_ptr_right d3
4505 #define clut_low_a d4
4506 #define clut_low_b d5
4507 #define clut_high_a d6
4508 #define clut_high_b d7
4510 #define block_masks d8
4511 #define block_masks_shifted d9
4516 #define texels_low d10
4517 #define texels_high d11
4520 setup_sprite_flush_blocks_single:
4523 stmdb sp!, { r0 - r3, r12, r14 }
4524 bl flush_render_block_buffer
4525 ldmia sp!, { r0 - r3, r12, r14 }
4529 add block, psx_gpu, #psx_gpu_blocks_offset
4531 mov num_blocks, sub_tile_height
4535 setup_sprite_flush_blocks_double:
4538 stmdb sp!, { r0 - r3, r12, r14 }
4539 bl flush_render_block_buffer
4540 ldmia sp!, { r0 - r3, r12, r14 }
4544 add block, psx_gpu, #psx_gpu_blocks_offset
4546 mov num_blocks, sub_tile_height, lsl #1
4550 setup_sprite_update_texture_4bpp_cache:
4551 stmdb sp!, { r0 - r3, r14 }
4552 bl update_texture_4bpp_cache
4553 ldmia sp!, { r0 - r3, pc }
4556 setup_sprite_update_texture_8bpp_cache:
4557 stmdb sp!, { r0 - r3, r14 }
4558 bl update_texture_8bpp_cache
4559 ldmia sp!, { r0 - r3, pc }
4562 #define setup_sprite_tiled_initialize_4bpp() \
4563 ldr dirty_textures_mask, \
4564 [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]; \
4565 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4567 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4568 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4570 tst current_texture_mask, dirty_textures_mask; \
4571 vuzp.u8 clut_a, clut_b; \
4573 blne setup_sprite_update_texture_4bpp_cache \
4575 #define setup_sprite_tiled_initialize_8bpp() \
4576 ldr dirty_textures_mask, \
4577 [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]; \
4578 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4580 tst current_texture_mask, dirty_textures_mask; \
4581 blne setup_sprite_update_texture_8bpp_cache \
4584 #define setup_sprite_tile_setup_block_no(side, offset, texture_mode) \
4586 #define setup_sprite_block_count_single() \
4589 #define setup_sprite_block_count_double() \
4590 sub_tile_height, lsl #1 \
4592 #define setup_sprite_tile_add_blocks(type) \
4593 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4594 cmp num_blocks, #MAX_BLOCKS; \
4596 blgt setup_sprite_flush_blocks_##type \
4599 #define setup_sprite_tile_full_4bpp(edge) \
4600 setup_sprite_tile_add_blocks(double); \
4603 and texture_block_ptr, texture_offset, texture_mask; \
4604 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4607 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4608 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4610 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4611 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4613 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4614 add texture_block_ptr, texture_offset, #8; \
4616 and texture_block_ptr, texture_block_ptr, texture_mask; \
4617 add block, block, #40; \
4619 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4620 add fb_ptr, fb_ptr, #16; \
4622 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4623 add block, block, #24; \
4625 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4626 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4629 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4630 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4632 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4633 add block, block, #40; \
4635 add texture_offset, texture_offset, #0x10; \
4636 add fb_ptr, fb_ptr, #(2048 - 16); \
4638 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4639 add block, block, #24; \
4641 subs sub_tile_height, sub_tile_height, #1; \
4644 add texture_offset, texture_offset, #0xF00; \
4645 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4648 #define setup_sprite_tile_half_4bpp(edge) \
4649 setup_sprite_tile_add_blocks(single); \
4652 and texture_block_ptr, texture_offset, texture_mask; \
4653 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4656 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4657 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4659 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4660 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4662 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4663 add block, block, #40; \
4665 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4666 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4668 add block, block, #24; \
4669 add texture_offset, texture_offset, #0x10; \
4671 add fb_ptr, fb_ptr, #2048; \
4672 subs sub_tile_height, sub_tile_height, #1; \
4676 add texture_offset, texture_offset, #0xF00; \
4677 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4680 #define setup_sprite_tile_full_8bpp(edge) \
4681 setup_sprite_tile_add_blocks(double); \
4682 add block, block, #16; \
4685 and texture_block_ptr, texture_offset, texture_mask; \
4686 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4689 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4690 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4692 add texture_block_ptr, texture_offset, #8; \
4693 vst1.u32 { texels }, [ block, :64 ]; \
4695 and texture_block_ptr, texture_block_ptr, texture_mask; \
4696 add block, block, #24; \
4698 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4700 add fb_ptr, fb_ptr, #16; \
4701 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4703 add block, block, #40; \
4704 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4707 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4708 vst1.u32 { texels }, [ block, :64 ]; \
4709 add block, block, #24; \
4711 add texture_offset, texture_offset, #0x10; \
4712 add fb_ptr, fb_ptr, #(2048 - 16); \
4714 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4715 add block, block, #40; \
4717 subs sub_tile_height, sub_tile_height, #1; \
4720 sub block, block, #16; \
4721 add texture_offset, texture_offset, #0xF00; \
4722 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4725 #define setup_sprite_tile_half_8bpp(edge) \
4726 setup_sprite_tile_add_blocks(single); \
4727 add block, block, #16; \
4730 and texture_block_ptr, texture_offset, texture_mask; \
4731 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4734 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4735 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4737 vst1.u32 { texels }, [ block, :64 ]; \
4738 add block, block, #24; \
4740 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4741 add block, block, #40; \
4743 add texture_offset, texture_offset, #0x10; \
4744 add fb_ptr, fb_ptr, #2048; \
4746 subs sub_tile_height, sub_tile_height, #1; \
4749 sub block, block, #16; \
4750 add texture_offset, texture_offset, #0xF00; \
4751 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4754 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4755 add texture_offset, texture_offset_base, #8; \
4756 add fb_ptr, fb_ptr, #16 \
4758 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4759 mov texture_offset, texture_offset_base \
4761 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4762 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4764 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4765 mov texture_offset, texture_offset_base \
4767 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4768 sub fb_ptr, fb_ptr, #16 \
4770 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4772 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4773 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4775 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4778 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode) \
4779 mov sub_tile_height, column_data; \
4780 setup_sprite_tile_column_edge_pre_adjust_##edge_mode(edge); \
4781 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4782 setup_sprite_tile_column_edge_post_adjust_##edge_mode(edge) \
4784 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode) \
4785 and sub_tile_height, column_data, #0xFF; \
4786 mov tiles_remaining, column_data, lsr #16; \
4787 setup_sprite_tile_column_edge_pre_adjust_##edge_mode(edge); \
4788 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4790 subs tiles_remaining, tiles_remaining, #1; \
4794 mov sub_tile_height, #16; \
4795 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4796 subs tiles_remaining, tiles_remaining, #1; \
4800 uxtb sub_tile_height, column_data, ror #8; \
4801 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4802 setup_sprite_tile_column_edge_post_adjust_##edge_mode(edge) \
4805 #define setup_sprite_column_data_single() \
4806 mov column_data, height; \
4807 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] \
4809 #define setup_sprite_column_data_multi() \
4810 and height_rounded, height_rounded, #0xF; \
4811 rsb column_data, offset_v, #16; \
4813 add height_rounded, height_rounded, #1; \
4814 sub tile_height, tile_height, #1; \
4816 orr column_data, column_data, tile_height, lsl #16; \
4817 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]; \
4819 orr column_data, column_data, height_rounded, lsl #8 \
4821 #define setup_sprite_tile_column_width_single(texture_mode, multi_height, \
4823 setup_sprite_##texture_mode##_single_##multi_height##_##edge_mode##_##edge: \
4824 setup_sprite_column_data_##multi_height(); \
4825 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4826 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4827 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4828 vdup.u8 draw_mask_fb_ptr_right, block_masks[1]; \
4830 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, \
4832 ldmia sp!, { r4 - r11, pc } \
4834 #define setup_sprite_tiled_advance_column() \
4835 add texture_offset_base, texture_offset_base, #0x100; \
4836 tst texture_offset_base, #0xF00; \
4837 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4839 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4841 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode: \
4842 setup_sprite_column_data_##multi_height(); \
4843 mov fb_ptr_advance_column, #32; \
4845 sub fb_ptr_advance_column, height, lsl #11; \
4846 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4848 vdup.u8 draw_mask_fb_ptr_right, block_masks[1]; \
4849 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm); \
4851 subs tile_width, tile_width, #2; \
4852 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4854 vmov.u8 draw_masks_fb_ptrs, #0; \
4858 setup_sprite_tiled_advance_column(); \
4859 setup_sprite_tile_column_height_##multi_height(full, none, tm); \
4860 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4861 subs tile_width, tile_width, #1; \
4865 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4866 vdup.u8 draw_mask_fb_ptr_right, block_masks[5]; \
4868 setup_sprite_tiled_advance_column(); \
4869 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm); \
4870 ldmia sp!, { r4 - r11, pc } \
4878 // [ sp + 4 ]: width
4879 // [ sp + 8 ]: height
4880 // [ sp + 12 ]: color (unused)
4882 #define setup_sprite_tiled_builder(texture_mode) \
4884 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full); \
4885 setup_sprite_tile_column_width_single(texture_mode, multi, full, none); \
4886 setup_sprite_tile_column_width_multi(texture_mode, single, full, full); \
4887 setup_sprite_tile_column_width_single(texture_mode, single, full, none); \
4888 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full); \
4889 setup_sprite_tile_column_width_single(texture_mode, multi, half, right); \
4890 setup_sprite_tile_column_width_multi(texture_mode, single, half, full); \
4891 setup_sprite_tile_column_width_single(texture_mode, single, half, right); \
4892 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half); \
4893 setup_sprite_tile_column_width_single(texture_mode, multi, half, left); \
4894 setup_sprite_tile_column_width_multi(texture_mode, single, full, half); \
4895 setup_sprite_tile_column_width_single(texture_mode, single, half, left); \
4896 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half); \
4897 setup_sprite_tile_column_width_multi(texture_mode, single, half, half); \
4901 function(setup_sprite_##texture_mode) \
4902 stmdb sp!, { r4 - r11, r14 }; \
4903 setup_sprite_tiled_initialize_##texture_mode(); \
4905 ldr v, [ sp, #36 ]; \
4906 and offset_u, u, #0xF; \
4908 ldr width, [ sp, #40 ]; \
4909 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]; \
4911 ldr height, [ sp, #44 ]; \
4912 add fb_ptr, fb_ptr, y, lsl #11; \
4914 add fb_ptr, fb_ptr, x, lsl #1; \
4915 and offset_v, v, #0xF; \
4917 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4918 add width_rounded, offset_u, width; \
4920 add height_rounded, offset_v, height; \
4921 add width_rounded, width_rounded, #15; \
4923 add height_rounded, height_rounded, #15; \
4924 mov tile_width, width_rounded, lsr #4; \
4926 /* texture_offset_base = VH-VL-00-00 */\
4927 mov texture_offset_base, v, lsl #8; \
4928 and offset_u_right, width_rounded, #0xF; \
4930 /* texture_offset_base = VH-UH-UL-00 */\
4931 bfi texture_offset_base, u, #4, #8; \
4932 movw right_block_mask, #0xFFFE; \
4934 /* texture_offset_base = VH-UH-VL-00 */\
4935 bfi texture_offset_base, v, #4, #4; \
4936 movw left_block_mask, #0xFFFF; \
4938 mov tile_height, height_rounded, lsr #4; \
4939 mvn left_block_mask, left_block_mask, lsl offset_u; \
4941 /* texture_mask = HH-HL-WH-WL */\
4942 ldrh texture_mask, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]; \
4943 mov right_block_mask, right_block_mask, lsl offset_u_right; \
4945 /* texture_mask_rev = WH-WL-HH-HL */\
4946 rev16 texture_mask_rev, texture_mask; \
4947 vmov block_masks, left_block_mask, right_block_mask; \
4949 /* texture_mask = HH-HL-HL-WL */\
4950 bfi texture_mask, texture_mask_rev, #4, #4; \
4951 /* texture_mask_rev = 00-00-00-WH */\
4952 mov texture_mask_rev, texture_mask_rev, lsr #12; \
4954 /* texture_mask = HH-WH-HL-WL */\
4955 bfi texture_mask, texture_mask_rev, #8, #4; \
4956 and left_block_mask, left_block_mask, #0xFF; \
4958 mov control_mask, #0; \
4959 cmp left_block_mask, #0xFF; \
4961 uxtb right_block_mask, right_block_mask, ror #8; \
4962 orreq control_mask, control_mask, #0x4; \
4964 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4965 cmp right_block_mask, #0xFF; \
4967 orreq control_mask, control_mask, #0x8; \
4968 cmp tile_width, #1; \
4970 add block, psx_gpu, #psx_gpu_blocks_offset; \
4971 orreq control_mask, control_mask, #0x1; \
4973 cmp tile_height, #1; \
4974 add block, block, num_blocks, lsl #6; \
4976 orreq control_mask, control_mask, #0x2; \
4977 ldr pc, [ pc, control_mask, lsl #2 ]; \
4980 .word setup_sprite_##texture_mode##_multi_multi_full_full; \
4981 .word setup_sprite_##texture_mode##_single_multi_full_none; \
4982 .word setup_sprite_##texture_mode##_multi_single_full_full; \
4983 .word setup_sprite_##texture_mode##_single_single_full_none; \
4984 .word setup_sprite_##texture_mode##_multi_multi_half_full; \
4985 .word setup_sprite_##texture_mode##_single_multi_half_right; \
4986 .word setup_sprite_##texture_mode##_multi_single_half_full; \
4987 .word setup_sprite_##texture_mode##_single_single_half_right; \
4988 .word setup_sprite_##texture_mode##_multi_multi_full_half; \
4989 .word setup_sprite_##texture_mode##_single_multi_half_left; \
4990 .word setup_sprite_##texture_mode##_multi_single_full_half; \
4991 .word setup_sprite_##texture_mode##_single_single_half_left; \
4992 .word setup_sprite_##texture_mode##_multi_multi_half_half; \
4994 .word setup_sprite_##texture_mode##_multi_single_half_half \
4997 setup_sprite_tiled_builder(4bpp);
4998 setup_sprite_tiled_builder(8bpp);
5006 #define block_ptr r0
5007 #define num_blocks r1
5009 #define texel_shift_mask r3
5010 #define block_pixels_a r4
5011 #define block_pixels_b r5
5020 #define texels_01 r6
5021 #define texels_23 r7
5022 #define texels_45 r8
5023 #define texels_67 r9
5025 function(texture_sprite_blocks_8bpp)
5026 stmdb sp!, { r4 - r11, r14 }
5027 movw texel_shift_mask, #(0xFF << 1)
5029 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5030 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
5032 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
5033 ldr block_pixels_a, [ block_ptr, #16 ]
5036 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
5037 ldr block_pixels_b, [ block_ptr, #20 ]
5039 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
5040 ldrh texel_0, [ clut_ptr, texel_0 ]
5042 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
5043 ldrh texel_1, [ clut_ptr, texel_1 ]
5045 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
5046 ldr block_pixels_a, [ block_ptr, #(64 + 16) ]
5048 ldrh texel_2, [ clut_ptr, texel_2 ]
5049 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
5051 ldrh texel_3, [ clut_ptr, texel_3 ]
5052 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
5054 ldrh texel_4, [ clut_ptr, texel_4 ]
5055 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
5057 ldrh texel_5, [ clut_ptr, texel_5 ]
5058 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
5060 ldrh texel_6, [ clut_ptr, texel_6 ]
5061 orr texels_01, texel_0, texel_1, lsl #16
5063 ldrh texel_7, [ clut_ptr, texel_7 ]
5064 orr texels_23, texel_2, texel_3, lsl #16
5066 orr texels_45, texel_4, texel_5, lsl #16
5067 str texels_01, [ block_ptr, #0 ]
5069 orr texels_67, texel_6, texel_7, lsl #16
5070 str texels_23, [ block_ptr, #4 ]
5072 subs num_blocks, num_blocks, #1
5073 str texels_45, [ block_ptr, #8 ]
5075 str texels_67, [ block_ptr, #12 ]
5076 add block_ptr, block_ptr, #64
5080 ldmia sp!, { r4 - r11, pc }
5083 #undef width_rounded
5086 #undef texture_offset
5095 #define left_offset r8
5096 #define width_rounded r9
5097 #define right_width r10
5098 #define block_width r11
5100 #define texture_offset_base r1
5101 #define texture_mask r2
5102 #define texture_page_ptr r3
5103 #define num_blocks r4
5106 #define texture_offset r8
5107 #define blocks_remaining r9
5108 #define fb_ptr_pitch r12
5109 #define texture_block_ptr r14
5111 #define texture_mask_width r2
5112 #define texture_mask_height r3
5113 #define left_mask_bits r4
5114 #define right_mask_bits r5
5118 #undef block_masks_shifted
5121 #define block_masks d0
5122 #define block_masks_shifted d1
5123 #define draw_mask_fb_ptr d2
5127 setup_sprites_16bpp_flush_single:
5130 stmdb sp!, { r0 - r3, r12, r14 }
5131 bl flush_render_block_buffer
5132 ldmia sp!, { r0 - r3, r12, r14 }
5136 add block, psx_gpu, #psx_gpu_blocks_offset
5141 setup_sprites_16bpp_flush_row:
5144 stmdb sp!, { r0 - r3, r12, r14 }
5145 bl flush_render_block_buffer
5146 ldmia sp!, { r0 - r3, r12, r14 }
5150 add block, psx_gpu, #psx_gpu_blocks_offset
5151 mov num_blocks, block_width
5155 function(setup_sprite_16bpp)
5156 stmdb sp!, { r4 - r11, r14 }
5157 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5160 add fb_ptr, fb_ptr, y, lsl #11
5162 ldr width, [ sp, #40 ]
5163 add fb_ptr, fb_ptr, x, lsl #1
5165 ldr height, [ sp, #44 ]
5166 and left_offset, u, #0x7
5168 add texture_offset_base, u, u
5169 add width_rounded, width, #7
5171 add texture_offset_base, v, lsl #11
5172 mov left_mask_bits, #0xFF
5174 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5175 add width_rounded, width_rounded, left_offset
5177 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5178 sub fb_ptr, fb_ptr, left_offset, lsl #1
5180 add texture_mask, texture_mask_width, texture_mask_width
5181 mov right_mask_bits, #0xFE
5183 and right_width, width_rounded, #0x7
5184 mvn left_mask_bits, left_mask_bits, lsl left_offset
5186 add texture_mask, texture_mask_height, lsl #11
5187 mov block_width, width_rounded, lsr #3
5189 mov right_mask_bits, right_mask_bits, lsl right_width
5190 movw fb_ptr_pitch, #(2048 + 16)
5192 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5193 vmov block_masks, left_mask_bits, right_mask_bits
5195 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5196 add block, psx_gpu, #psx_gpu_blocks_offset
5198 bic texture_offset_base, texture_offset_base, #0xF
5201 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5202 add block, block, num_blocks, lsl #6
5206 vext.32 block_masks_shifted, block_masks, block_masks, #1
5207 vorr.u32 block_masks, block_masks, block_masks_shifted
5208 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5211 add num_blocks, num_blocks, #1
5212 cmp num_blocks, #MAX_BLOCKS
5213 blgt setup_sprites_16bpp_flush_single
5215 and texture_block_ptr, texture_offset_base, texture_mask
5216 subs height, height, #1
5218 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5219 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5221 vst1.u32 { texels }, [ block, :128 ]
5222 add block, block, #40
5224 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5227 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5229 add block, block, #24
5230 add texture_offset_base, texture_offset_base, #2048
5231 add fb_ptr, fb_ptr, #2048
5232 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5235 ldmia sp!, { r4 - r11, pc }
5238 add num_blocks, num_blocks, block_width
5239 mov texture_offset, texture_offset_base
5241 cmp num_blocks, #MAX_BLOCKS
5242 blgt setup_sprites_16bpp_flush_row
5244 add texture_offset_base, texture_offset_base, #2048
5245 and texture_block_ptr, texture_offset, texture_mask
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 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5254 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5257 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5258 subs blocks_remaining, block_width, #2
5260 add texture_offset, texture_offset, #16
5261 add fb_ptr, fb_ptr, #16
5263 vmov.u8 draw_mask_fb_ptr, #0
5265 add block, block, #24
5269 and texture_block_ptr, texture_offset, texture_mask
5270 subs blocks_remaining, blocks_remaining, #1
5272 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5273 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5275 vst1.u32 { texels }, [ block, :128 ]
5276 add block, block, #40
5278 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5281 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5283 add texture_offset, texture_offset, #16
5284 add fb_ptr, fb_ptr, #16
5286 add block, block, #24
5290 and texture_block_ptr, texture_offset, texture_mask
5291 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5293 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5294 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5296 vst1.u32 { texels }, [ block, :128 ]
5297 add block, block, #40
5299 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5300 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5302 add block, block, #24
5303 subs height, height, #1
5305 add fb_ptr, fb_ptr, fb_ptr_pitch
5306 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5310 ldmia sp!, { r4 - r11, pc }
5313 #undef texture_page_ptr
5315 #undef dirty_textures_mask
5316 #undef current_texture_mask
5319 #define current_texture_page r1
5320 #define texture_page_ptr r2
5321 #define vram_ptr_a r3
5322 #define current_texture_page_x r12
5323 #define current_texture_page_y r4
5324 #define dirty_textures_mask r5
5328 #define current_texture_mask r9
5330 #define vram_ptr_b r11
5332 #define texel_block_a d0
5333 #define texel_block_b d1
5334 #define texel_block_expanded_a q1
5335 #define texel_block_expanded_b q2
5336 #define texel_block_expanded_ab q2
5337 #define texel_block_expanded_c q3
5338 #define texel_block_expanded_d q4
5339 #define texel_block_expanded_cd q3
5341 function(update_texture_4bpp_cache)
5342 stmdb sp!, { r4 - r11, r14 }
5345 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5347 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5348 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5350 and current_texture_page_x, current_texture_page, #0xF
5351 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
5353 mov current_texture_page_y, current_texture_page, lsr #4
5354 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5356 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5359 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5360 bic dirty_textures_mask, current_texture_mask
5363 str dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5368 add vram_ptr_b, vram_ptr_a, #2048
5371 vld1.u32 { texel_block_a }, [ vram_ptr_a, :64 ], c_4096
5372 vld1.u32 { texel_block_b }, [ vram_ptr_b, :64 ], c_4096
5374 vmovl.u8 texel_block_expanded_a, texel_block_a
5375 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5376 vmovl.u8 texel_block_expanded_c, texel_block_b
5377 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5379 vbic.u16 texel_block_expanded_a, #0x00F0
5380 vbic.u16 texel_block_expanded_b, #0x00F0
5381 vbic.u16 texel_block_expanded_c, #0x00F0
5382 vbic.u16 texel_block_expanded_d, #0x00F0
5384 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
5385 texel_block_expanded_b
5386 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
5387 texel_block_expanded_d
5389 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
5390 [ texture_page_ptr, :256 ]!
5392 subs sub_y, sub_y, #1
5396 add vram_ptr_a, vram_ptr_a, #8
5397 add vram_ptr_b, vram_ptr_b, #8
5399 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5400 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5402 subs tile_x, tile_x, #1
5406 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5407 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5409 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5410 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5412 subs tile_y, tile_y, #1
5416 ldmia sp!, { r4 - r11, pc }
5419 #undef current_texture_page
5422 #define texture_page r1
5423 #define texture_page_ptr r2
5424 #define vram_ptr_a r3
5425 #define texture_page_x r12
5426 #define texture_page_y r4
5427 #define current_texture_page r5
5432 #define vram_ptr_b r11
5444 function(update_texture_8bpp_cache_slice)
5445 stmdb sp!, { r4 - r11, r14 }
5448 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5449 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5451 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5454 and texture_page_x, texture_page, #0xF
5455 mov texture_page_y, texture_page, lsr #4
5457 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
5460 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
5461 eor current_texture_page, current_texture_page, texture_page
5463 ands current_texture_page, current_texture_page, #0x1
5466 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
5469 add vram_ptr_b, vram_ptr_a, #2048
5472 vld1.u32 { texels_a }, [ vram_ptr_a, :128 ], c_4096
5473 vld1.u32 { texels_b }, [ vram_ptr_b, :128 ], c_4096
5474 vld1.u32 { texels_c }, [ vram_ptr_a, :128 ], c_4096
5475 vld1.u32 { texels_d }, [ vram_ptr_b, :128 ], c_4096
5477 vst1.u32 { texels_a, texels_b }, [ texture_page_ptr, :256 ]!
5478 vst1.u32 { texels_c, texels_d }, [ texture_page_ptr, :256 ]!
5480 subs sub_y, sub_y, #1
5485 add vram_ptr_a, vram_ptr_a, #16
5486 add vram_ptr_b, vram_ptr_b, #16
5488 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5489 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5491 subs tile_x, tile_x, #1
5496 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5497 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5499 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5500 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5502 subs tile_y, tile_y, #1
5503 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
5508 ldmia sp!, { r4 - r11, pc }
notaz.gp2x.de / pcsx_rearmed.git / blob