2 * Copyright (C) 2011 Gilead Kutnick "Exophase" <exophase@gmail.com>
3 * Copyright (C) 2012 GraÅžvydas Ignotas "notaz" <notasas@gmail.com>
5 * This program is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU General Public License as
7 * published by the Free Software Foundation; either version 2 of
8 * the License, or (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License for more details.
18 #define MAX_BLOCKS_PER_ROW 128
20 #define RENDER_STATE_MASK_EVALUATE 0x20
21 #define RENDER_FLAGS_MODULATE_TEXELS 0x1
22 #define RENDER_FLAGS_BLEND 0x2
23 #define RENDER_INTERLACE_ENABLED 0x1
25 #include "psx_gpu_offsets.h"
27 #define psx_gpu_b_dx_offset (psx_gpu_b_block_span_offset + 4)
29 #define edge_data_left_x_offset 0
30 #define edge_data_num_blocks_offset 2
31 #define edge_data_right_mask_offset 4
32 #define edge_data_y_offset 6
76 #define gw_by_l gw_bx_l
101 #define uvrg_xxxx1 q4
106 #define uvrg_xxxx2 q5
111 #define yyyy_uvrg0 q6
116 #define yyyy_uvrg1 q7
121 #define yyyy_uvrg2 q8
148 #define uvrg_base q11
191 #define uvrgb_phase q13
195 /* FIXME: users of this should be in psx_gpu instead */
197 #define load_pointer(register, pointer) \
198 movw register, :lower16:pointer; \
199 movt register, :upper16:pointer; \
202 #define load_pointer(register, pointer) \
203 ldr register, =pointer \
207 #define function(name) \
216 function(compute_all_gradients)
217 // First compute the triangle area reciprocal and shift. The division will
218 // happen concurrently with much of the work which follows.
219 @ r12 = psx_gpu->triangle_area
220 ldr r12, [ psx_gpu, #psx_gpu_triangle_area_offset ]
221 stmdb sp!, { r4 - r11, lr }
223 @ load exponent of 62 into upper half of double
225 clz r14, r12 @ r14 = shift
227 movt r4, #((62 + 1023) << 4)
228 mov r12, r12, lsl r14 @ r12 = triangle_area_normalized
230 @ load area normalized into lower half of double
232 vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n
234 movt r4, #((1022 + 31) << 4)
237 add r4, r4, r12, lsr #11
240 vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n
242 // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) =
243 // ( d0 * d1 ) - ( d2 * d3 ) =
244 // ( m0 ) - ( m1 ) = gradient
246 // This is split to do 12 elements at a time over three sets: a, b, and c.
247 // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so
248 // two of the slots are unused.
250 // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as
253 // First type is: uvrg bxxx xxxx
254 // Second type is: yyyy ybyy uvrg
255 // Since x_a and y_c are the same the same variable is used for both.
257 vld1.u32 { v0 }, [ v_a, : 128 ] @ v0 = { uvrg0, b0, x0, y0 }
258 ldrsh x0, [ v_a, #8 ] @ load x0
260 vld1.u32 { v1 }, [ v_b, : 128 ] @ v1 = { uvrg1, b1, x1, y1}
261 ldrh x1, [ v_b, #8 ] @ load x1
263 vld1.u32 { v2 }, [ v_c, : 128 ] @ v2 = { uvrg2, b2, x2, y2 }
264 ldrh x2, [ v_c, #8 ] @ load x2
266 vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- }
267 ldrh y0, [ v_a, #10 ] @ load y0
269 vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- }
270 ldrh y1, [ v_b, #10 ] @ load y1
272 vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- }
273 ldrh y2, [ v_c, #10 ] @ load y2
275 vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 }
276 vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 }
278 orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 }
279 pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 }
281 vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 }
282 vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 }
284 vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 }
285 vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 }
287 ldrb b2, [ v_c, #4 ] @ load b2
288 orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 }
290 ldrb b1, [ v_b, #4 ] @ load b1
291 orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 }
293 vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 }
294 vsub.s16 d0_ab, x1_ab, x0_ab
296 ldrb b0, [ v_a, #4 ] @ load b0
297 orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 }
299 vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 }
300 vsub.s16 d2_ab, x2_ab, x1_ab
302 vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 }
303 vsub.s16 d1_ab, y2_ab, y1_ab
305 orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 }
306 ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 }
308 ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 }
309 ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 }
311 vsub.s16 d3_ab, y1_ab, y0_ab
312 smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) -
313 @ ((x2 - X1) * (b1 - b0))
314 vmull.s16 ga_uvrg_x, d0_a, d1_a
315 smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) -
316 @ ((b2 - b1) * (y1 - y0))
317 vmlsl.s16 ga_uvrg_x, d2_a, d3_a
318 movs gs_bx, ga_bx, asr #31
320 vmull.s16 ga_uvrg_y, d0_b, d1_b
321 rsbmi ga_bx, ga_bx, #0
323 @ r12 = psx_gpu->uvrgb_phase
324 ldr r12, [ psx_gpu, #psx_gpu_uvrgb_phase_offset ]
326 vmlsl.s16 ga_uvrg_y, d2_b, d3_b
327 movs gs_by, ga_by, asr #31
329 vshr.u64 d0, d30, #22
330 add b_base, r12, b0, lsl #16
332 vdup.u32 uvrgb_phase, r12
334 rsbmi ga_by, ga_by, #0
335 vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0
337 @ r12 = psx_gpu->triangle_winding_offset
338 ldrb r12, [ psx_gpu, #psx_gpu_triangle_winding_offset ]
339 vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0
341 rsb r12, r12, #0 @ r12 = -(triangle->winding)
343 vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w }
344 sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS)
346 vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16
347 vdup.u32 r_shift, r14 @ r_shift = { shift, shift, shift, shift }
349 vadd.u32 uvrg_base, uvrgb_phase
350 vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x)
352 vmov area_r_s, s0 @ area_r_s = triangle_reciprocal
353 vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y)
355 vmull.u32 gw_rg_x, ga_rg_x, d0[0]
356 vmull.u32 gw_uv_x, ga_uv_x, d0[0]
357 vmull.u32 gw_rg_y, ga_rg_y, d0[0]
358 vmull.u32 gw_uv_y, ga_uv_y, d0[0]
360 vshl.u64 gw_rg_x, gw_rg_x, r_shift
361 vshl.u64 gw_uv_x, gw_uv_x, r_shift
362 vshl.u64 gw_rg_y, gw_rg_y, r_shift
363 vshl.u64 gw_uv_y, gw_uv_y, r_shift
365 veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask
366 vmovn.u64 g_uv_x, gw_uv_x
368 veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask
369 vmovn.u64 g_rg_x, gw_rg_x
371 veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
372 vmovn.u64 g_uv_y, gw_uv_y
374 vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
375 vmovn.u64 g_rg_y, gw_rg_y
377 veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
378 mov ga_bx, ga_bx, lsl #13
380 vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
381 mov ga_by, ga_by, lsl #13
384 umull gw_bx_l, gw_bx_h, ga_bx, area_r_s
386 vshl.u32 g_uvrg_x, g_uvrg_x, #4
387 vshl.u32 g_uvrg_y, g_uvrg_y, #4
389 umull gw_by_l, gw_by_h, ga_by, area_r_s
390 vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0]
392 eor gs_bx, gs_bx, r12
393 vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1
395 veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0
396 eor gs_by, gs_by, r12
398 rsb r11, r14, #0 @ r11 = negative shift for scalar lsr
399 add store_a, psx_gpu, #psx_gpu_uvrg_offset
401 sub r11, r11, #(32 - 13)
403 add store_b, store_a, #16
406 vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1
407 vst1.u32 { uvrg_base }, [ store_a, : 128 ], store_inc
409 vst1.u32 { uvrg_dx1 }, [ store_b, : 128 ], store_inc
410 mov g_bx, gw_bx_h, lsr r11
412 vst1.u32 { g_uvrg_y }, [ store_a, : 128 ], store_inc
413 mov g_by, gw_by_h, lsr r11
415 vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \
416 [ store_b, : 128 ], store_inc
417 eor g_bx, g_bx, gs_bx
419 vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \
420 [ store_b, : 128 ], store_inc
421 sub g_bx, g_bx, gs_bx
424 eor g_by, g_by, gs_by
426 mls b_base, g_bx, x0, b_base
427 sub g_by, g_by, gs_by
432 add g_bx2, g_bx, g_bx
433 add g_bx3, g_bx, g_bx2
435 stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by }
437 ldmia sp!, { r4 - r11, pc }
454 #define height_minor_a r7
455 #define height_minor_b r8
456 #define height_major r9
459 #define reciprocal_table_ptr r10
461 #define edge_alt_low r4
462 #define edge_alt_high r5
463 #define edge_dx_dy_alt r6
464 #define edge_shift_alt r10
466 #define edge_dx_dy_alt_low r4
467 #define edge_dx_dy_alt_high r5
469 #define span_edge_data r4
470 #define span_uvrg_offset r5
471 #define span_b_offset r6
479 #define alternate_x q0
480 #define alternate_dx_dy q1
481 #define alternate_x_32 q2
483 #define alternate_x_low d0
484 #define alternate_x_high d1
485 #define alternate_dx_dy_low d2
486 #define alternate_dx_dy_high d3
487 #define alternate_x_32_low d4
488 #define alternate_x_32_high d5
492 #define left_dx_dy q5
493 #define right_dx_dy q6
495 #define right_edge q8
497 #define left_x_low d6
498 #define left_x_high d7
499 #define right_x_low d8
500 #define right_x_high d9
501 #define left_dx_dy_low d10
502 #define left_dx_dy_high d11
503 #define right_dx_dy_low d12
504 #define right_dx_dy_high d13
505 #define left_edge_low d14
506 #define left_edge_high d15
507 #define right_edge_low d16
508 #define right_edge_high d17
510 #define y_mid_point d18
513 #define left_right_x_16 q11
514 #define span_shifts_y q12
517 #define span_shifts d24
522 #define left_right_x_16_low d22
523 #define left_right_x_16_high d23
528 #define alternate_x_16 d4
531 #define v_clip_low d6
533 #define right_x_32 q10
534 #define left_x_32 q11
535 #define alternate_select d24
537 #define right_x_32_low d20
538 #define right_x_32_high d21
539 #define left_x_32_low d22
540 #define left_x_32_high d23
543 #define edges_dx_dy d2
544 #define edge_shifts d3
545 #define edge_shifts_64 q2
547 #define edges_xy_left d0
548 #define edges_xy_right d1
550 #define height_reciprocals d6
558 #define heights_b d12
559 #define edges_dx_dy_64 q10
561 #define edges_dx_dy_64_left d20
562 #define edges_dx_dy_64_right d21
565 #define setup_spans_prologue() \
566 stmdb sp!, { r4 - r11, lr }; \
568 ldrsh x_a, [ v_a, #8 ]; \
569 ldrsh x_b, [ v_b, #8 ]; \
570 ldrsh x_c, [ v_c, #8 ]; \
571 ldrsh y_a, [ v_a, #10 ]; \
572 ldrsh y_b, [ v_b, #10 ]; \
573 ldrsh y_c, [ v_c, #10 ]; \
575 add temp, psx_gpu, #psx_gpu_uvrg_offset; \
576 vld1.32 { uvrg }, [ temp ]; \
577 add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \
578 vld1.32 { uvrg_dy }, [ temp ]; \
579 load_pointer(reciprocal_table_ptr, reciprocal_table); \
581 vmov.u32 c_0x01, #0x01 \
583 #define setup_spans_load_b() \
584 ldr b, [ psx_gpu, #psx_gpu_b_offset ]; \
585 ldr b_dy, [ psx_gpu, #psx_gpu_b_dy_offset ] \
587 #define setup_spans_prologue_b() \
588 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
589 add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \
591 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
592 vmov.u16 c_0x0004, #0x0004; \
594 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
595 vmov.u16 c_0x0001, #0x0001; \
597 vld1.u16 { left_edge_low[], left_edge_high[] }, [ temp ]; \
598 add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \
600 vld1.u16 { right_edge_low[], right_edge_high[] }, [ temp ]; \
601 vadd.u16 right_edge, right_edge, c_0x0001; \
603 vmov.u16 c_0x0007, #0x0007; \
604 vmvn.u16 c_0xFFFE, #0x0001 \
607 #define compute_edge_delta_x2() \
608 ldr temp, [ reciprocal_table_ptr, height, lsl #2 ]; \
610 vdup.u32 heights, height; \
611 vsub.u32 widths, x_ends, x_starts; \
613 vdup.u32 edge_shifts, temp; \
614 vsub.u32 heights_b, heights, c_0x01; \
615 vshr.u32 height_reciprocals, edge_shifts, #10; \
617 vmla.s32 heights_b, x_starts, heights; \
618 vbic.u16 edge_shifts, #0xE0; \
619 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
620 vmull.s32 edges_xy, heights_b, height_reciprocals \
623 #define height_reciprocal_alt r11
624 #define height_b_alt r12
626 #define compute_edge_delta_x3(start_c, height_a, height_b) \
627 vmov.u32 heights, height_a, height_b; \
628 ldr temp, [ reciprocal_table_ptr, height_a, lsl #2 ]; \
629 vmov.u32 edge_shifts[0], temp; \
630 ldr temp, [ reciprocal_table_ptr, height_b, lsl #2 ]; \
631 vmov.u32 edge_shifts[1], temp; \
632 ldr edge_shift_alt, [ reciprocal_table_ptr, height_minor_b, lsl #2 ]; \
634 vsub.u32 widths, x_ends, x_starts; \
635 sub width_alt, x_c, start_c; \
637 vsub.u32 heights_b, heights, c_0x01; \
638 sub height_b_alt, height_minor_b, #1; \
640 vshr.u32 height_reciprocals, edge_shifts, #10; \
641 lsr height_reciprocal_alt, edge_shift_alt, #10; \
643 vmla.s32 heights_b, x_starts, heights; \
644 mla height_b_alt, height_minor_b, start_c, height_b_alt; \
646 vbic.u16 edge_shifts, #0xE0; \
647 and edge_shift_alt, edge_shift_alt, #0x1F; \
649 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
650 mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \
652 vmull.s32 edges_xy, heights_b, height_reciprocals; \
653 smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \
656 #define setup_spans_adjust_y_up() \
657 vsub.u32 y_x4, y_x4, c_0x0004 \
659 #define setup_spans_adjust_y_down() \
660 vadd.u32 y_x4, y_x4, c_0x0004 \
662 #define setup_spans_adjust_interpolants_up() \
663 vsub.u32 uvrg, uvrg, uvrg_dy; \
666 #define setup_spans_adjust_interpolants_down() \
667 vadd.u32 uvrg, uvrg, uvrg_dy; \
671 #define setup_spans_clip_interpolants_increment() \
672 mla b, b_dy, clip, b; \
673 vmla.s32 uvrg, uvrg_dy, v_clip \
675 #define setup_spans_clip_interpolants_decrement() \
676 mls b, b_dy, clip, b; \
677 vmls.s32 uvrg, uvrg_dy, v_clip \
679 #define setup_spans_clip_alternate_yes() \
680 smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \
682 #define setup_spans_clip_alternate_no() \
684 #define setup_spans_clip(direction, alternate_active) \
685 vdup.u32 v_clip, clip; \
686 setup_spans_clip_alternate_##alternate_active(); \
687 setup_spans_clip_interpolants_##direction(); \
688 vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \
691 #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \
692 vmovl.s32 edge_shifts_64, edge_shifts; \
693 vmovl.s32 edges_dx_dy_64, edges_dx_dy; \
695 vshl.s64 edges_xy, edges_xy, edge_shifts_64; \
696 vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \
698 vmov left_x_low, edges_xy_##left_index; \
699 vmov right_x_low, edges_xy_##right_index; \
701 vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \
702 vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \
703 vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \
704 vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \
706 vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \
707 vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \
709 vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \
710 vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \
713 #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \
714 setup_spans_adjust_edges_alternate_no(left_index, right_index); \
716 vdup.u16 y_mid_point, y_b; \
717 rsb temp, edge_shift_alt, #32; \
719 lsl edge_alt_high, edge_alt_high, edge_shift_alt; \
720 orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \
721 lsl edge_alt_low, edge_alt_low, edge_shift_alt; \
722 vmov alternate_x_low, edge_alt_low, edge_alt_high; \
724 asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \
725 lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \
726 vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \
727 vmov alternate_dx_dy_high, alternate_dx_dy_low; \
729 vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \
730 vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \
733 #define setup_spans_y_select_up() \
734 vclt.s16 alternate_select, y_x4, y_mid_point \
736 #define setup_spans_y_select_down() \
737 vcgt.s16 alternate_select, y_x4, y_mid_point \
740 #define setup_spans_alternate_select_left() \
741 vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \
743 #define setup_spans_alternate_select_right() \
744 vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \
747 #define setup_spans_set_x4_alternate_yes(alternate, direction) \
748 vshrn.s64 alternate_x_32_low, alternate_x, #32; \
749 vshrn.s64 left_x_32_low, left_x, #32; \
750 vshrn.s64 right_x_32_low, right_x, #32; \
752 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
753 vadd.u64 left_x, left_x, left_dx_dy; \
754 vadd.u64 right_x, right_x, right_dx_dy; \
756 vshrn.s64 alternate_x_32_high, alternate_x, #32; \
757 vshrn.s64 left_x_32_high, left_x, #32; \
758 vshrn.s64 right_x_32_high, right_x, #32; \
760 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
761 vadd.u64 left_x, left_x, left_dx_dy; \
762 vadd.u64 right_x, right_x, right_dx_dy; \
764 vmovn.u32 alternate_x_16, alternate_x_32; \
765 setup_spans_y_select_##direction(); \
766 vmovn.u32 left_right_x_16_low, left_x_32; \
768 vmovn.u32 left_right_x_16_high, right_x_32; \
769 setup_spans_alternate_select_##alternate(); \
771 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
772 str b, [ span_b_offset ], #4; \
773 setup_spans_adjust_interpolants_##direction(); \
775 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
777 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
778 str b, [ span_b_offset ], #4; \
779 setup_spans_adjust_interpolants_##direction(); \
781 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
783 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
784 str b, [ span_b_offset ], #4; \
785 setup_spans_adjust_interpolants_##direction(); \
787 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
788 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
789 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
791 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
792 str b, [ span_b_offset ], #4; \
793 setup_spans_adjust_interpolants_##direction(); \
795 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
796 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
798 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
800 setup_spans_adjust_y_##direction() \
803 #define setup_spans_set_x4_alternate_no(alternate, direction) \
804 vshrn.s64 left_x_32_low, left_x, #32; \
805 vshrn.s64 right_x_32_low, right_x, #32; \
807 vadd.u64 left_x, left_x, left_dx_dy; \
808 vadd.u64 right_x, right_x, right_dx_dy; \
810 vshrn.s64 left_x_32_high, left_x, #32; \
811 vshrn.s64 right_x_32_high, right_x, #32; \
813 vadd.u64 left_x, left_x, left_dx_dy; \
814 vadd.u64 right_x, right_x, right_dx_dy; \
816 vmovn.u32 left_right_x_16_low, left_x_32; \
817 vmovn.u32 left_right_x_16_high, right_x_32; \
819 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
820 str b, [ span_b_offset ], #4; \
821 setup_spans_adjust_interpolants_##direction(); \
823 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
825 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
826 str b, [ span_b_offset ], #4; \
827 setup_spans_adjust_interpolants_##direction(); \
829 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
831 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
832 str b, [ span_b_offset ], #4; \
833 setup_spans_adjust_interpolants_##direction(); \
835 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
836 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
837 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
839 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
840 str b, [ span_b_offset ], #4; \
841 setup_spans_adjust_interpolants_##direction(); \
843 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
844 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
846 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
848 setup_spans_adjust_y_##direction() \
851 #define edge_adjust_low r11
852 #define edge_adjust_high r12
854 #define setup_spans_alternate_adjust_yes() \
855 smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \
856 subs edge_alt_low, edge_alt_low, edge_adjust_low; \
857 sbc edge_alt_high, edge_alt_high, edge_adjust_high \
859 #define setup_spans_alternate_adjust_no() \
862 #define setup_spans_down(left_index, right_index, alternate, alternate_active) \
863 setup_spans_alternate_adjust_##alternate_active(); \
864 setup_spans_load_b(); \
866 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
867 subs y_c, y_c, temp; \
868 subgt height, height, y_c; \
869 addgt height, height, #1; \
871 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
872 subs clip, temp, y_a; \
875 sub height, height, clip; \
876 add y_a, y_a, clip; \
877 setup_spans_clip(increment, alternate_active); \
883 orr temp, y_a, y_a, lsl #16; \
884 add temp, temp, #(1 << 16); \
886 add y_a, y_a, #(2 << 16); \
887 vmov.u32 y_x4, temp, y_a; \
889 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
891 setup_spans_prologue_b(); \
893 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
896 setup_spans_set_x4_alternate_##alternate_active(alternate, down); \
897 subs height, height, #4; \
903 #define setup_spans_alternate_pre_increment_yes() \
904 adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \
905 adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \
907 #define setup_spans_alternate_pre_increment_no() \
910 #define setup_spans_up_decrement_yes() \
911 suble height, height, #1 \
913 #define setup_spans_up_decrement_no() \
916 #define setup_spans_up(left_index, right_index, alternate, alternate_active) \
917 setup_spans_alternate_adjust_##alternate_active(); \
918 setup_spans_load_b(); \
921 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
922 subs temp, temp, y_c; \
923 subgt height, height, temp; \
924 setup_spans_up_decrement_##alternate_active(); \
926 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
927 subs clip, y_a, temp; \
930 sub height, height, clip; \
931 sub y_a, y_a, clip; \
932 setup_spans_clip(decrement, alternate_active); \
938 orr temp, y_a, y_a, lsl #16; \
939 sub temp, temp, #(1 << 16); \
941 sub y_a, y_a, #(2 << 16); \
942 vmov.u32 y_x4, temp, y_a; \
944 vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \
946 setup_spans_alternate_pre_increment_##alternate_active(); \
947 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
949 setup_spans_adjust_interpolants_up(); \
950 setup_spans_prologue_b(); \
952 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
955 setup_spans_set_x4_alternate_##alternate_active(alternate, up); \
956 subs height, height, #4; \
962 #define setup_spans_epilogue() \
963 ldmia sp!, { r4 - r11, pc } \
966 #define setup_spans_up_up(minor, major) \
967 setup_spans_prologue(); \
968 sub height_minor_a, y_a, y_b; \
969 sub height_minor_b, y_b, y_c; \
970 sub height, y_a, y_c; \
972 vdup.u32 x_starts, x_a; \
973 vmov.u32 x_ends, x_c, x_b; \
975 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
976 setup_spans_up(major, minor, minor, yes); \
977 setup_spans_epilogue() \
979 function(setup_spans_up_left)
980 setup_spans_up_up(left, right)
982 function(setup_spans_up_right)
983 setup_spans_up_up(right, left)
987 #define setup_spans_down_down(minor, major) \
988 setup_spans_prologue(); \
989 sub height_minor_a, y_b, y_a; \
990 sub height_minor_b, y_c, y_b; \
991 sub height, y_c, y_a; \
993 vdup.u32 x_starts, x_a; \
994 vmov.u32 x_ends, x_c, x_b; \
996 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
997 setup_spans_down(major, minor, minor, yes); \
998 setup_spans_epilogue() \
1000 function(setup_spans_down_left)
1001 setup_spans_down_down(left, right)
1003 function(setup_spans_down_right)
1004 setup_spans_down_down(right, left)
1007 #define setup_spans_up_flat() \
1008 sub height, y_a, y_c; \
1010 compute_edge_delta_x2(); \
1011 setup_spans_up(left, right, none, no); \
1012 setup_spans_epilogue() \
1014 function(setup_spans_up_a)
1015 setup_spans_prologue()
1017 vmov.u32 x_starts, x_a, x_b
1018 vdup.u32 x_ends, x_c
1020 setup_spans_up_flat()
1022 function(setup_spans_up_b)
1023 setup_spans_prologue()
1025 vdup.u32 x_starts, x_a
1026 vmov.u32 x_ends, x_b, x_c
1028 setup_spans_up_flat()
1030 #define setup_spans_down_flat() \
1031 sub height, y_c, y_a; \
1033 compute_edge_delta_x2(); \
1034 setup_spans_down(left, right, none, no); \
1035 setup_spans_epilogue() \
1037 function(setup_spans_down_a)
1038 setup_spans_prologue()
1040 vmov.u32 x_starts, x_a, x_b
1041 vdup.u32 x_ends, x_c
1043 setup_spans_down_flat()
1045 function(setup_spans_down_b)
1046 setup_spans_prologue()
1048 vdup.u32 x_starts, x_a
1049 vmov.u32 x_ends, x_b, x_c
1051 setup_spans_down_flat()
1056 #define edges_xy_b q11
1057 #define edges_dx_dy_b d26
1058 #define edge_shifts_b d27
1059 #define edges_dx_dy_and_shifts_b q13
1060 #define height_increment d20
1062 #define edges_dx_dy_and_shifts q1
1064 #define edges_xy_b_left d22
1065 #define edges_xy_b_right d23
1067 #define setup_spans_up_down_load_edge_set_b() \
1068 vmov edges_xy, edges_xy_b; \
1069 vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \
1072 function(setup_spans_up_down)
1073 setup_spans_prologue()
1075 // s32 middle_y = y_a;
1076 sub height_minor_a, y_a, y_b
1077 sub height_minor_b, y_c, y_a
1078 sub height_major, y_c, y_b
1080 vmov.u32 x_starts, x_a, x_c
1081 vdup.u32 x_ends, x_b
1083 compute_edge_delta_x3(x_a, height_minor_a, height_major)
1086 vmov.u32 height_increment, temp, height_minor_b
1087 vmlal.s32 edges_xy, edges_dx_dy, height_increment
1089 vmov edges_xy_b_left, edge_alt_low, edge_alt_high
1090 vmov edges_xy_b_right, edges_xy_right
1092 vmov edge_shifts_b, edge_shifts
1093 vmov.u32 edge_shifts_b[0], edge_shift_alt
1095 vneg.s32 edges_dx_dy_b, edges_dx_dy
1096 vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt
1100 setup_spans_load_b()
1103 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1104 subs temp, temp, y_b
1105 subgt height_minor_a, height_minor_a, temp
1107 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1108 subs clip, y_a, temp
1111 sub height_minor_a, height_minor_a, clip
1113 setup_spans_clip(decrement, no)
1116 cmp height_minor_a, #0
1119 orr temp, y_a, y_a, lsl #16
1120 sub temp, temp, #(1 << 16)
1122 sub y_a, y_a, #(2 << 16)
1123 vmov.u32 y_x4, temp, y_a
1125 vaddw.s32 edges_xy, edges_xy, edges_dx_dy
1127 strh height_minor_a, [ psx_gpu, #psx_gpu_num_spans_offset ]
1129 setup_spans_adjust_edges_alternate_no(left, right);
1130 setup_spans_adjust_interpolants_up()
1131 setup_spans_up_down_load_edge_set_b()
1133 setup_spans_prologue_b()
1137 setup_spans_set_x4_alternate_no(none, up)
1138 subs height_minor_a, height_minor_a, #4
1141 add span_edge_data, span_edge_data, height_minor_a, lsl #3
1142 add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4
1143 add span_b_offset, span_b_offset, height_minor_a, lsl #2
1146 add temp, psx_gpu, #psx_gpu_uvrg_offset
1147 vld1.32 { uvrg }, [ temp ]
1150 setup_spans_load_b()
1152 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1154 subgt height_minor_b, height_minor_b, y_c
1155 addgt height_minor_b, height_minor_b, #1
1157 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1158 subs clip, temp, y_a
1161 sub height_minor_b, height_minor_b, clip
1163 setup_spans_clip(increment, no)
1166 cmp height_minor_b, #0
1169 orr temp, y_a, y_a, lsl #16
1170 add temp, temp, #(1 << 16)
1172 add y_a, y_a, #(2 << 16)
1173 vmov.u32 y_x4, temp, y_a
1175 setup_spans_adjust_edges_alternate_no(left, right)
1177 ldrh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1178 add temp, temp, height_minor_b
1180 cmp temp, #MAX_SPANS
1183 strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1186 setup_spans_set_x4_alternate_no(none, down)
1187 subs height_minor_b, height_minor_b, #4
1191 setup_spans_epilogue()
1194 setup_spans_up_down_load_edge_set_b()
1195 setup_spans_prologue_b()
1199 // FIXME: overflow corner case
1200 sub temp, temp, height_minor_b
1201 bics height_minor_b, #3
1202 add temp, temp, height_minor_b
1203 strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1209 #undef span_uvrg_offset
1210 #undef span_edge_data
1211 #undef span_b_offset
1216 #define num_spans r1
1217 #define span_uvrg_offset r2
1218 #define span_edge_data r3
1219 #define span_b_offset r4
1221 #define span_num_blocks r6
1225 #define dither_offset_ptr r10
1226 #define block_ptr_a r11
1228 #define num_blocks r14
1230 #define uvrg_dx_ptr r2
1231 #define texture_mask_ptr r3
1232 #define dither_shift r8
1233 #define dither_row r10
1238 #define block_ptr_b r10
1240 #define block_span_ptr r10
1241 #define right_mask r8
1260 #define b_whole_8 d14
1261 #define fb_mask_ptrs d15
1274 #define u_whole_low d16
1275 #define u_whole_high d17
1276 #define v_whole_low d18
1277 #define v_whole_high d19
1278 #define r_whole_low d20
1279 #define r_whole_high d21
1280 #define g_whole_low d22
1281 #define g_whole_high d23
1282 #define b_whole_low d24
1283 #define b_whole_high d25
1288 #define u_whole_8 d26
1289 #define v_whole_8 d27
1290 #define u_whole_8b d24
1291 #define r_whole_8 d24
1292 #define g_whole_8 d25
1294 #define uv_whole_8 q13
1295 #define uv_whole_8b q14
1297 #define dither_offsets q14
1298 #define texture_mask q15
1299 #define texture_mask_u d30
1300 #define texture_mask_v d31
1302 #define dither_offsets_short d28
1306 #define block_span q10
1311 #define draw_mask q1
1312 #define draw_mask_edge q13
1313 #define test_mask q0
1319 #define setup_blocks_texture_swizzled() \
1320 vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \
1321 vsli.u8 u_whole_8, v_whole_8, #4; \
1322 vsri.u8 v_whole_8, u_whole_8b, #4 \
1324 #define setup_blocks_texture_unswizzled() \
1327 #define setup_blocks_shaded_textured_builder(swizzling) \
1330 function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \
1331 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1332 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1334 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1335 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1337 cmp num_spans, #0; \
1340 stmdb sp!, { r4 - r11, r14 }; \
1341 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1343 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
1344 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1346 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1347 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1349 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1350 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1352 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
1353 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1355 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1358 vmov.u8 fb_mask_ptrs, #0; \
1360 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1361 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1363 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1364 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
1366 cmp span_num_blocks, #0; \
1369 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1370 add num_blocks, span_num_blocks, num_blocks; \
1372 cmp num_blocks, #MAX_BLOCKS; \
1376 ldr b, [ span_b_offset ]; \
1377 add fb_ptr, fb_ptr, y, lsl #11; \
1379 vdup.u32 v_left_x, left_x; \
1382 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1383 add fb_ptr, fb_ptr, left_x, lsl #1; \
1385 mla b, b_dx, left_x, b; \
1386 and dither_shift, left_x, #0x03; \
1388 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1389 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1391 mov dither_shift, dither_shift, lsl #3; \
1392 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1395 subs span_num_blocks, span_num_blocks, #1; \
1397 mov dither_row, dither_row, ror dither_shift; \
1398 mov b_dx4, b_dx, lsl #2; \
1400 vdup.u32 dither_offsets_short, dither_row; \
1401 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1403 vdup.u32 b_block, b; \
1404 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1406 vdup.u32 u_block, uv[0]; \
1407 mov b_dx8, b_dx, lsl #3; \
1409 vdup.u32 v_block, uv[1]; \
1410 vdup.u32 r_block, rg[0]; \
1411 vdup.u32 g_block, rg[1]; \
1413 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1415 vadd.u32 u_block, u_block, block_span; \
1416 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1418 vadd.u32 v_block, v_block, block_span; \
1419 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1421 vadd.u32 r_block, r_block, block_span; \
1422 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1424 vadd.u32 g_block, g_block, block_span; \
1425 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
1427 vadd.u32 b_block, b_block, block_span; \
1428 add block_ptr_b, block_ptr_a, #16; \
1430 vshrn.u32 u_whole_low, u_block, #16; \
1431 vshrn.u32 v_whole_low, v_block, #16; \
1432 vshrn.u32 r_whole_low, r_block, #16; \
1433 vshrn.u32 g_whole_low, g_block, #16; \
1435 vdup.u32 dx4, uv_dx4[0]; \
1436 vshrn.u32 b_whole_low, b_block, #16; \
1438 vaddhn.u32 u_whole_high, u_block, dx4; \
1439 vdup.u32 dx4, uv_dx4[1]; \
1441 vaddhn.u32 v_whole_high, v_block, dx4; \
1442 vdup.u32 dx4, rg_dx4[0]; \
1444 vaddhn.u32 r_whole_high, r_block, dx4; \
1445 vdup.u32 dx4, rg_dx4[1]; \
1447 vaddhn.u32 g_whole_high, g_block, dx4; \
1448 vdup.u32 dx4, b_dx4; \
1450 vaddhn.u32 b_whole_high, b_block, dx4; \
1451 vdup.u32 dx8, uv_dx8[0]; \
1453 vadd.u32 u_block, u_block, dx8; \
1454 vdup.u32 dx8, uv_dx8[1]; \
1456 vadd.u32 v_block, v_block, dx8; \
1457 vdup.u32 dx8, rg_dx8[0]; \
1459 vadd.u32 r_block, r_block, dx8; \
1460 vdup.u32 dx8, rg_dx8[1]; \
1462 vadd.u32 g_block, g_block, dx8; \
1463 vdup.u32 dx8, b_dx8; \
1465 vadd.u32 b_block, b_block, dx8; \
1466 vmovn.u16 u_whole_8, u_whole; \
1468 vmovn.u16 v_whole_8, v_whole; \
1470 vmovn.u16 b_whole_8, b_whole; \
1472 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1474 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1475 setup_blocks_texture_##swizzling(); \
1477 vmovn.u16 r_whole_8, r_whole; \
1481 vmovn.u16 g_whole_8, g_whole; \
1482 vshrn.u32 u_whole_low, u_block, #16; \
1484 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1485 vshrn.u32 v_whole_low, v_block, #16; \
1487 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1488 vshrn.u32 r_whole_low, r_block, #16; \
1490 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1491 vshrn.u32 g_whole_low, g_block, #16; \
1493 vdup.u32 dx4, uv_dx4[0]; \
1494 vshrn.u32 b_whole_low, b_block, #16; \
1496 vaddhn.u32 u_whole_high, u_block, dx4; \
1497 vdup.u32 dx4, uv_dx4[1]; \
1499 vaddhn.u32 v_whole_high, v_block, dx4; \
1500 vdup.u32 dx4, rg_dx4[0]; \
1502 vaddhn.u32 r_whole_high, r_block, dx4; \
1503 vdup.u32 dx4, rg_dx4[1]; \
1505 vaddhn.u32 g_whole_high, g_block, dx4; \
1506 vdup.u32 dx4, b_dx4; \
1508 vaddhn.u32 b_whole_high, b_block, dx4; \
1509 vdup.u32 dx8, uv_dx8[0]; \
1511 vadd.u32 u_block, u_block, dx8; \
1512 vdup.u32 dx8, uv_dx8[1]; \
1514 vadd.u32 v_block, v_block, dx8; \
1515 vdup.u32 dx8, rg_dx8[0]; \
1517 vadd.u32 r_block, r_block, dx8; \
1518 vdup.u32 dx8, rg_dx8[1]; \
1520 vadd.u32 g_block, g_block, dx8; \
1521 vdup.u32 dx8, b_dx8; \
1523 vadd.u32 b_block, b_block, dx8; \
1524 vmovn.u16 u_whole_8, u_whole; \
1526 add fb_ptr, fb_ptr, #16; \
1527 vmovn.u16 v_whole_8, v_whole; \
1529 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1530 vmovn.u16 b_whole_8, b_whole; \
1534 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1535 subs span_num_blocks, span_num_blocks, #1; \
1537 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1538 setup_blocks_texture_##swizzling(); \
1540 vmovn.u16 r_whole_8, r_whole; \
1544 vmovn.u16 g_whole_8, g_whole; \
1545 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1547 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1548 vdup.u8 draw_mask, right_mask; \
1550 vmov.u32 fb_mask_ptrs[0], right_mask; \
1551 vtst.u16 draw_mask, draw_mask, test_mask; \
1552 vzip.u8 u_whole_8, v_whole_8; \
1554 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1555 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1556 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1557 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1558 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1561 add span_uvrg_offset, span_uvrg_offset, #16; \
1562 add span_b_offset, span_b_offset, #4; \
1564 add span_edge_data, span_edge_data, #8; \
1565 subs num_spans, num_spans, #1; \
1567 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1570 ldmia sp!, { r4 - r11, pc }; \
1573 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1574 vpush { texture_mask }; \
1575 vpush { uvrg_dx4 }; \
1577 stmdb sp!, { r0 - r3, r12, r14 }; \
1578 bl flush_render_block_buffer; \
1579 ldmia sp!, { r0 - r3, r12, r14 }; \
1581 vpop { uvrg_dx4 }; \
1582 vpop { texture_mask }; \
1584 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1585 vmov.u8 fb_mask_ptrs, #0; \
1587 mov num_blocks, span_num_blocks; \
1588 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1592 setup_blocks_shaded_textured_builder(swizzled)
1593 setup_blocks_shaded_textured_builder(unswizzled)
1596 #define setup_blocks_unshaded_textured_builder(swizzling) \
1599 function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \
1600 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1601 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1603 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1604 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1606 cmp num_spans, #0; \
1609 stmdb sp!, { r4 - r11, r14 }; \
1610 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1612 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1614 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1615 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1617 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1618 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1620 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1622 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1625 vmov.u8 fb_mask_ptrs, #0; \
1627 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1628 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1630 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1631 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
1633 cmp span_num_blocks, #0; \
1636 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1637 add num_blocks, span_num_blocks, num_blocks; \
1639 cmp num_blocks, #MAX_BLOCKS; \
1643 add fb_ptr, fb_ptr, y, lsl #11; \
1645 vdup.u32 v_left_x, left_x; \
1648 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1649 add fb_ptr, fb_ptr, left_x, lsl #1; \
1651 and dither_shift, left_x, #0x03; \
1653 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1654 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1656 mov dither_shift, dither_shift, lsl #3; \
1657 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1660 subs span_num_blocks, span_num_blocks, #1; \
1662 mov dither_row, dither_row, ror dither_shift; \
1664 vdup.u32 dither_offsets_short, dither_row; \
1665 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1667 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1669 vdup.u32 u_block, uv[0]; \
1671 vdup.u32 v_block, uv[1]; \
1672 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1674 vadd.u32 u_block, u_block, block_span; \
1675 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1677 vadd.u32 v_block, v_block, block_span; \
1678 add block_ptr_b, block_ptr_a, #16; \
1680 vshrn.u32 u_whole_low, u_block, #16; \
1681 vshrn.u32 v_whole_low, v_block, #16; \
1683 vdup.u32 dx4, uv_dx4[0]; \
1685 vaddhn.u32 u_whole_high, u_block, dx4; \
1686 vdup.u32 dx4, uv_dx4[1]; \
1688 vaddhn.u32 v_whole_high, v_block, dx4; \
1689 vdup.u32 dx8, uv_dx8[0]; \
1691 vadd.u32 u_block, u_block, dx8; \
1692 vdup.u32 dx8, uv_dx8[1]; \
1694 vadd.u32 v_block, v_block, dx8; \
1695 vmovn.u16 u_whole_8, u_whole; \
1697 vmovn.u16 v_whole_8, v_whole; \
1700 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1702 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1703 setup_blocks_texture_##swizzling(); \
1708 vshrn.u32 u_whole_low, u_block, #16; \
1710 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1711 vshrn.u32 v_whole_low, v_block, #16; \
1713 add block_ptr_b, block_ptr_b, #32; \
1714 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1716 vdup.u32 dx4, uv_dx4[0]; \
1717 vaddhn.u32 u_whole_high, u_block, dx4; \
1718 vdup.u32 dx4, uv_dx4[1]; \
1720 vaddhn.u32 v_whole_high, v_block, dx4; \
1721 vdup.u32 dx8, uv_dx8[0]; \
1723 vadd.u32 u_block, u_block, dx8; \
1724 vdup.u32 dx8, uv_dx8[1]; \
1726 vadd.u32 v_block, v_block, dx8; \
1727 vmovn.u16 u_whole_8, u_whole; \
1729 add fb_ptr, fb_ptr, #16; \
1730 vmovn.u16 v_whole_8, v_whole; \
1732 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1735 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1736 subs span_num_blocks, span_num_blocks, #1; \
1738 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1739 setup_blocks_texture_##swizzling(); \
1744 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1746 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1747 vdup.u8 draw_mask, right_mask; \
1749 vmov.u32 fb_mask_ptrs[0], right_mask; \
1750 vtst.u16 draw_mask, draw_mask, test_mask; \
1751 vzip.u8 u_whole_8, v_whole_8; \
1753 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1754 add block_ptr_b, block_ptr_b, #32; \
1755 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1756 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1757 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1760 add span_uvrg_offset, span_uvrg_offset, #16; \
1761 add span_edge_data, span_edge_data, #8; \
1762 subs num_spans, num_spans, #1; \
1764 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1767 ldmia sp!, { r4 - r11, pc }; \
1770 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1771 vpush { texture_mask }; \
1772 vpush { uvrg_dx4 }; \
1774 stmdb sp!, { r0 - r3, r12, r14 }; \
1775 bl flush_render_block_buffer; \
1776 ldmia sp!, { r0 - r3, r12, r14 }; \
1778 vpop { uvrg_dx4 }; \
1779 vpop { texture_mask }; \
1781 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1782 vmov.u8 fb_mask_ptrs, #0; \
1784 mov num_blocks, span_num_blocks; \
1785 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1789 setup_blocks_unshaded_textured_builder(swizzled)
1790 setup_blocks_unshaded_textured_builder(unswizzled)
1795 function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect)
1796 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1797 veor.u32 draw_mask, draw_mask, draw_mask
1802 stmdb sp!, { r4 - r11, r14 }
1803 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1805 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1807 ubfx color_r, color, #3, #5
1808 ubfx color_g, color, #11, #5
1809 ubfx color_b, color, #19, #5
1811 orr color, color_r, color_b, lsl #10
1812 orr color, color, color_g, lsl #5
1814 vdup.u16 colors, color
1816 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1817 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1819 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1820 add block_ptr_a, block_ptr_a, num_blocks, lsl #6
1823 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1824 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1826 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
1828 cmp span_num_blocks, #0
1831 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1832 add num_blocks, span_num_blocks, num_blocks
1834 cmp num_blocks, #MAX_BLOCKS
1838 add fb_ptr, fb_ptr, y, lsl #11
1841 add fb_ptr, fb_ptr, left_x, lsl #1
1844 subs span_num_blocks, span_num_blocks, #1
1846 add block_ptr_b, block_ptr_a, #16
1849 vmov.u32 fb_mask_ptrs[1], fb_ptr
1853 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_32
1854 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1855 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1857 add fb_ptr, fb_ptr, #16
1858 add block_ptr_b, block_ptr_b, #32
1862 vmov.u32 fb_mask_ptrs[1], fb_ptr
1863 subs span_num_blocks, span_num_blocks, #1
1868 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1870 vdup.u8 draw_mask_edge, right_mask
1871 vtst.u16 draw_mask_edge, draw_mask_edge, test_mask
1873 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1874 vst1.u32 { draw_mask_edge }, [ block_ptr_a, :128 ], c_32
1875 add block_ptr_b, block_ptr_b, #32
1876 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1879 add span_edge_data, span_edge_data, #8
1880 subs num_spans, num_spans, #1
1882 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1885 ldmia sp!, { r4 - r11, pc }
1890 stmdb sp!, { r0 - r3, r12, r14 }
1891 bl flush_render_block_buffer
1892 ldmia sp!, { r0 - r3, r12, r14 }
1896 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1897 veor.u32 draw_mask, draw_mask, draw_mask
1899 mov num_blocks, span_num_blocks
1900 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1904 #define mask_msb_scalar r14
1906 #define msb_mask q15
1908 #define pixels_low d16
1910 #define msb_mask_low d30
1911 #define msb_mask_high d31
1916 function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct)
1917 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1922 stmdb sp!, { r4 - r11, r14 }
1924 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1926 ubfx color_r, color, #3, #5
1927 ubfx color_g, color, #11, #5
1929 ldrh mask_msb_scalar, [ psx_gpu, #psx_gpu_mask_msb_offset ]
1930 ubfx color_b, color, #19, #5
1932 orr color, color_r, color_b, lsl #10
1933 orr color, color, color_g, lsl #5
1934 orr color, color, mask_msb_scalar
1936 vdup.u16 colors, color
1938 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1939 orr color, color, lsl #16
1943 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1944 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1946 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
1948 cmp span_num_blocks, #0
1951 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1953 add fb_ptr, fb_ptr, y, lsl #11
1954 subs span_num_blocks, span_num_blocks, #1
1956 add fb_ptr, fb_ptr, left_x, lsl #1
1960 vst1.u32 { colors }, [ fb_ptr ]!
1961 subs span_num_blocks, span_num_blocks, #1
1966 ldrb right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1968 cmp right_mask, #0x0
1971 tst right_mask, #0xF
1972 streq color, [ fb_ptr ], #4
1973 moveq right_mask, right_mask, lsr #4
1974 streq color, [ fb_ptr ], #4
1976 tst right_mask, #0x3
1977 streq color, [ fb_ptr ], #4
1978 moveq right_mask, right_mask, lsr #2
1980 tst right_mask, #0x1
1981 streqh color, [ fb_ptr ]
1984 add span_edge_data, span_edge_data, #8
1985 subs num_spans, num_spans, #1
1988 ldmia sp!, { r4 - r11, pc }
1991 vst1.u32 { colors }, [ fb_ptr ]
1998 #define rg_dx_ptr r2
2016 #undef dither_offsets
2036 #define r_whole_low d6
2037 #define r_whole_high d7
2038 #define g_whole_low d8
2039 #define g_whole_high d9
2040 #define b_whole_low d10
2041 #define b_whole_high d11
2043 #define gb_whole_8 q6
2045 #define g_whole_8 d12
2046 #define b_whole_8 d13
2048 #define r_whole_8 d14
2060 #define block_span q5
2068 #define d128_0x7 q13
2072 #define dither_offsets q14
2073 #define draw_mask q15
2075 #define dither_offsets_low d28
2078 #define test_mask q10
2081 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2082 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2083 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2085 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2086 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2087 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2089 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2091 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2094 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2097 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2098 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2099 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2101 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2103 cmp num_spans, #0; \
2106 stmdb sp!, { r4 - r11, r14 }; \
2107 vshl.u32 rg_dx4, rg_dx, #2; \
2109 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2110 vshl.u32 rg_dx8, rg_dx, #3; \
2112 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2114 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2115 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2117 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2118 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2120 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2121 vmov.u8 d64_1, #1; \
2123 vmov.u8 d128_4, #4; \
2124 vmov.u8 d64_128, #128; \
2126 vmov.u8 d128_0x7, #0x7; \
2129 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2130 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2132 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2133 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
2135 cmp span_num_blocks, #0; \
2138 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2139 add num_blocks, span_num_blocks, num_blocks; \
2141 cmp num_blocks, #MAX_BLOCKS; \
2145 ldr b, [ span_b_offset ]; \
2146 add fb_ptr, fb_ptr, y, lsl #11; \
2148 vdup.u32 v_left_x, left_x; \
2151 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2152 add fb_ptr, fb_ptr, left_x, lsl #1; \
2154 mla b, b_dx, left_x, b; \
2155 and dither_shift, left_x, #0x03; \
2157 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2158 vshr.u32 rg_dx, rg_dx4, #2; \
2160 mov dither_shift, dither_shift, lsl #3; \
2161 vmla.u32 rg, rg_dx, v_left_x; \
2164 subs span_num_blocks, span_num_blocks, #1; \
2166 mov dither_row, dither_row, ror dither_shift; \
2167 mov b_dx4, b_dx, lsl #2; \
2169 vdup.u32 dither_offsets, dither_row; \
2170 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2172 vdup.u32 b_block, b; \
2173 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2175 mov b_dx8, b_dx, lsl #3; \
2176 vdup.u32 r_block, rg[0]; \
2177 vdup.u32 g_block, rg[1]; \
2179 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2181 vadd.u32 r_block, r_block, block_span; \
2182 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2184 vadd.u32 g_block, g_block, block_span; \
2185 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2187 vadd.u32 b_block, b_block, block_span; \
2188 add block_ptr_b, block_ptr_a, #16; \
2190 vshrn.u32 r_whole_low, r_block, #16; \
2191 vshrn.u32 g_whole_low, g_block, #16; \
2192 vshrn.u32 b_whole_low, b_block, #16; \
2193 vdup.u32 dx4, rg_dx4[0]; \
2195 vaddhn.u32 r_whole_high, r_block, dx4; \
2196 vdup.u32 dx4, rg_dx4[1]; \
2198 vaddhn.u32 g_whole_high, g_block, dx4; \
2199 vdup.u32 dx4, b_dx4; \
2201 vaddhn.u32 b_whole_high, b_block, dx4; \
2202 vdup.u32 dx8, rg_dx8[0]; \
2204 vadd.u32 r_block, r_block, dx8; \
2205 vdup.u32 dx8, rg_dx8[1]; \
2207 vadd.u32 g_block, g_block, dx8; \
2208 vdup.u32 dx8, b_dx8; \
2210 vadd.u32 b_block, b_block, dx8; \
2212 vmovn.u16 r_whole_8, r_whole; \
2213 vmovn.u16 g_whole_8, g_whole; \
2214 vmovn.u16 b_whole_8, b_whole; \
2217 veor.u32 draw_mask, draw_mask, draw_mask; \
2220 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2221 vshrn.u32 r_whole_low, r_block, #16; \
2223 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2224 vshrn.u32 g_whole_low, g_block, #16; \
2226 vshrn.u32 b_whole_low, b_block, #16; \
2227 str fb_ptr, [ block_ptr_a, #44 ]; \
2229 vdup.u32 dx4, rg_dx4[0]; \
2230 vshr.u8 r_whole_8, r_whole_8, #3; \
2231 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2233 vaddhn.u32 r_whole_high, r_block, dx4; \
2234 vdup.u32 dx4, rg_dx4[1]; \
2236 vaddhn.u32 g_whole_high, g_block, dx4; \
2237 vdup.u32 dx4, b_dx4; \
2239 vaddhn.u32 b_whole_high, b_block, dx4; \
2240 vdup.u32 dx8, rg_dx8[0]; \
2242 vmull.u8 pixels, r_whole_8, d64_1; \
2243 vmlal.u8 pixels, g_whole_8, d64_4; \
2244 vmlal.u8 pixels, b_whole_8, d64_128; \
2246 vadd.u32 r_block, r_block, dx8; \
2247 vdup.u32 dx8, rg_dx8[1]; \
2249 vadd.u32 g_block, g_block, dx8; \
2250 vdup.u32 dx8, b_dx8; \
2252 vadd.u32 b_block, b_block, dx8; \
2253 add fb_ptr, fb_ptr, #16; \
2255 vmovn.u16 r_whole_8, r_whole; \
2256 vmovn.u16 g_whole_8, g_whole; \
2257 vmovn.u16 b_whole_8, b_whole; \
2259 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2260 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2264 subs span_num_blocks, span_num_blocks, #1; \
2268 str fb_ptr, [ block_ptr_a, #44 ]; \
2269 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2271 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2272 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2274 vshr.u8 r_whole_8, r_whole_8, #3; \
2275 vdup.u8 draw_mask, right_mask; \
2277 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2278 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
2280 vtst.u16 draw_mask, draw_mask, test_mask; \
2282 vmull.u8 pixels, r_whole_8, d64_1; \
2283 vmlal.u8 pixels, g_whole_8, d64_4; \
2284 vmlal.u8 pixels, b_whole_8, d64_128; \
2286 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2287 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2290 add span_uvrg_offset, span_uvrg_offset, #16; \
2291 add span_b_offset, span_b_offset, #4; \
2293 add span_edge_data, span_edge_data, #8; \
2294 subs num_spans, num_spans, #1; \
2296 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2299 ldmia sp!, { r4 - r11, pc }; \
2302 /* TODO: Load from psx_gpu instead of saving/restoring these */\
2305 stmdb sp!, { r0 - r3, r12, r14 }; \
2306 bl flush_render_block_buffer; \
2307 ldmia sp!, { r0 - r3, r12, r14 }; \
2311 vmov.u8 d64_1, #1; \
2312 vmov.u8 d128_4, #4; \
2313 vmov.u8 d64_128, #128; \
2314 vmov.u8 d128_0x7, #0x7; \
2316 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2318 mov num_blocks, span_num_blocks; \
2319 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2323 setup_blocks_shaded_untextured_indirect_builder(undithered)
2324 setup_blocks_shaded_untextured_indirect_builder(dithered)
2329 #define mask_msb_ptr r14
2331 #define draw_mask q0
2332 #define pixels_low d16
2333 #define pixels_high d17
2337 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2340 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2341 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2342 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2344 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2346 cmp num_spans, #0; \
2349 stmdb sp!, { r4 - r11, r14 }; \
2350 vshl.u32 rg_dx4, rg_dx, #2; \
2352 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2353 vshl.u32 rg_dx8, rg_dx, #3; \
2355 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2356 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2358 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2359 vmov.u8 d64_1, #1; \
2361 vmov.u8 d128_4, #4; \
2362 vmov.u8 d64_128, #128; \
2364 vmov.u8 d128_0x7, #0x7; \
2365 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2366 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
2369 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2370 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2372 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2373 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
2375 cmp span_num_blocks, #0; \
2378 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2379 add fb_ptr, fb_ptr, y, lsl #11; \
2381 ldr b, [ span_b_offset ]; \
2382 vdup.u32 v_left_x, left_x; \
2385 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2386 add fb_ptr, fb_ptr, left_x, lsl #1; \
2388 mla b, b_dx, left_x, b; \
2389 and dither_shift, left_x, #0x03; \
2391 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2392 vshr.u32 rg_dx, rg_dx4, #2; \
2394 mov dither_shift, dither_shift, lsl #3; \
2395 vmla.u32 rg, rg_dx, v_left_x; \
2397 subs span_num_blocks, span_num_blocks, #1; \
2399 mov dither_row, dither_row, ror dither_shift; \
2400 mov b_dx4, b_dx, lsl #2; \
2402 vdup.u32 dither_offsets, dither_row; \
2403 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2405 vdup.u32 b_block, b; \
2406 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2408 mov b_dx8, b_dx, lsl #3; \
2409 vdup.u32 r_block, rg[0]; \
2410 vdup.u32 g_block, rg[1]; \
2412 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2414 vadd.u32 r_block, r_block, block_span; \
2415 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2417 vadd.u32 g_block, g_block, block_span; \
2418 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2420 vadd.u32 b_block, b_block, block_span; \
2421 add block_ptr_b, block_ptr_a, #16; \
2423 vshrn.u32 r_whole_low, r_block, #16; \
2424 vshrn.u32 g_whole_low, g_block, #16; \
2425 vshrn.u32 b_whole_low, b_block, #16; \
2426 vdup.u32 dx4, rg_dx4[0]; \
2428 vaddhn.u32 r_whole_high, r_block, dx4; \
2429 vdup.u32 dx4, rg_dx4[1]; \
2431 vaddhn.u32 g_whole_high, g_block, dx4; \
2432 vdup.u32 dx4, b_dx4; \
2434 vaddhn.u32 b_whole_high, b_block, dx4; \
2435 vdup.u32 dx8, rg_dx8[0]; \
2437 vadd.u32 r_block, r_block, dx8; \
2438 vdup.u32 dx8, rg_dx8[1]; \
2440 vadd.u32 g_block, g_block, dx8; \
2441 vdup.u32 dx8, b_dx8; \
2443 vadd.u32 b_block, b_block, dx8; \
2445 vmovn.u16 r_whole_8, r_whole; \
2446 vmovn.u16 g_whole_8, g_whole; \
2447 vmovn.u16 b_whole_8, b_whole; \
2452 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2453 vshrn.u32 r_whole_low, r_block, #16; \
2455 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2456 vshrn.u32 g_whole_low, g_block, #16; \
2458 vshrn.u32 b_whole_low, b_block, #16; \
2460 vdup.u32 dx4, rg_dx4[0]; \
2461 vshr.u8 r_whole_8, r_whole_8, #3; \
2462 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2464 vaddhn.u32 r_whole_high, r_block, dx4; \
2465 vdup.u32 dx4, rg_dx4[1]; \
2467 vmov pixels, msb_mask; \
2468 vaddhn.u32 g_whole_high, g_block, dx4; \
2469 vdup.u32 dx4, b_dx4; \
2471 vaddhn.u32 b_whole_high, b_block, dx4; \
2472 vdup.u32 dx8, rg_dx8[0]; \
2474 vmlal.u8 pixels, r_whole_8, d64_1; \
2475 vmlal.u8 pixels, g_whole_8, d64_4; \
2476 vmlal.u8 pixels, b_whole_8, d64_128; \
2478 vadd.u32 r_block, r_block, dx8; \
2479 vdup.u32 dx8, rg_dx8[1]; \
2481 vadd.u32 g_block, g_block, dx8; \
2482 vdup.u32 dx8, b_dx8; \
2484 vadd.u32 b_block, b_block, dx8; \
2486 vmovn.u16 r_whole_8, r_whole; \
2487 vmovn.u16 g_whole_8, g_whole; \
2488 vmovn.u16 b_whole_8, b_whole; \
2490 vst1.u32 { pixels }, [ fb_ptr ]!; \
2491 subs span_num_blocks, span_num_blocks, #1; \
2495 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2497 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2498 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2500 vshr.u8 r_whole_8, r_whole_8, #3; \
2501 rbit right_mask, right_mask; \
2502 vmov pixels, msb_mask; \
2503 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2504 clz right_mask, right_mask; \
2506 vmlal.u8 pixels, r_whole_8, d64_1; \
2507 vmlal.u8 pixels, g_whole_8, d64_4; \
2508 vmlal.u8 pixels, b_whole_8, d64_128; \
2510 ldr pc, [ pc, right_mask, lsl #2 ]; \
2523 vst1.u16 { pixels_low[0] }, [ fb_ptr ]; \
2527 vst1.u32 { pixels_low[0] }, [ fb_ptr ]; \
2531 vst1.u32 { pixels_low[0] }, [ fb_ptr ]!; \
2532 vst1.u16 { pixels_low[2] }, [ fb_ptr ]; \
2536 vst1.u32 { pixels_low }, [ fb_ptr ]; \
2540 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2541 vst1.u16 { pixels_high[0] }, [ fb_ptr ]; \
2545 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2546 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2550 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2551 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2552 vst1.u16 { pixels_high[2] }, [ fb_ptr ]; \
2556 vst1.u32 { pixels }, [ fb_ptr ]; \
2560 add span_uvrg_offset, span_uvrg_offset, #16; \
2561 add span_b_offset, span_b_offset, #4; \
2563 add span_edge_data, span_edge_data, #8; \
2564 subs num_spans, num_spans, #1; \
2568 ldmia sp!, { r4 - r11, pc } \
2570 setup_blocks_shaded_untextured_direct_builder(undithered)
2571 setup_blocks_shaded_untextured_direct_builder(dithered)
2580 #define block_ptr r1
2581 #define num_blocks r2
2594 #define texture_ptr r11
2608 #define pixels_d r10
2612 #define clut_ptr r12
2613 #define current_texture_mask r5
2614 #define dirty_textures_mask r6
2618 #define clut_low_a d2
2619 #define clut_low_b d3
2620 #define clut_high_a d4
2621 #define clut_high_b d5
2626 #define texels_low d6
2627 #define texels_high d7
2631 function(texture_blocks_untextured)
2637 function(texture_blocks_4bpp)
2638 stmdb sp!, { r3 - r11, r14 }
2639 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2641 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2642 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2644 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2645 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]
2647 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2648 vuzp.u8 clut_a, clut_b
2650 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
2651 tst dirty_textures_mask, current_texture_mask
2657 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2659 uxtah uv_0, texture_ptr, uv_01
2660 uxtah uv_1, texture_ptr, uv_01, ror #16
2662 uxtah uv_2, texture_ptr, uv_23
2663 uxtah uv_3, texture_ptr, uv_23, ror #16
2665 uxtah uv_4, texture_ptr, uv_45
2666 ldrb pixel_0, [ uv_0 ]
2668 uxtah uv_5, texture_ptr, uv_45, ror #16
2669 ldrb pixel_1, [ uv_1 ]
2671 uxtah uv_6, texture_ptr, uv_67
2672 ldrb pixel_2, [ uv_2 ]
2674 uxtah uv_7, texture_ptr, uv_67, ror #16
2675 ldrb pixel_3, [ uv_3 ]
2677 ldrb pixel_4, [ uv_4 ]
2678 subs num_blocks, num_blocks, #1
2680 ldrb pixel_5, [ uv_5 ]
2681 orr pixels_a, pixel_0, pixel_1, lsl #8
2683 ldrb pixel_6, [ uv_6 ]
2684 orr pixels_b, pixel_4, pixel_5, lsl #8
2686 ldrb pixel_7, [ uv_7 ]
2687 orr pixels_a, pixels_a, pixel_2, lsl #16
2689 orr pixels_b, pixels_b, pixel_6, lsl #16
2690 orr pixels_a, pixels_a, pixel_3, lsl #24
2692 orr pixels_b, pixels_b, pixel_7, lsl #24
2693 vmov.u32 texels, pixels_a, pixels_b
2695 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2696 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2698 vst2.u8 { texels_low, texels_high }, [ block_ptr, :128 ], c_64
2701 ldmia sp!, { r3 - r11, pc }
2704 stmdb sp!, { r1 - r2 }
2705 bl update_texture_4bpp_cache
2708 ldmia sp!, { r1 - r2 }
2714 function(texture_blocks_8bpp)
2715 stmdb sp!, { r3 - r11, r14 }
2716 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2718 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2719 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2721 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2722 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2724 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]
2725 tst dirty_textures_mask, current_texture_mask
2731 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2733 uxtah uv_0, texture_ptr, uv_01
2734 uxtah uv_1, texture_ptr, uv_01, ror #16
2736 uxtah uv_2, texture_ptr, uv_23
2737 uxtah uv_3, texture_ptr, uv_23, ror #16
2739 uxtah uv_4, texture_ptr, uv_45
2740 ldrb pixel_0, [ uv_0 ]
2742 uxtah uv_5, texture_ptr, uv_45, ror #16
2743 ldrb pixel_1, [ uv_1 ]
2745 uxtah uv_6, texture_ptr, uv_67
2746 ldrb pixel_2, [ uv_2 ]
2748 uxtah uv_7, texture_ptr, uv_67, ror #16
2749 ldrb pixel_3, [ uv_3 ]
2751 ldrb pixel_4, [ uv_4 ]
2752 add pixel_0, pixel_0, pixel_0
2754 ldrb pixel_5, [ uv_5 ]
2755 add pixel_1, pixel_1, pixel_1
2757 ldrb pixel_6, [ uv_6 ]
2758 add pixel_2, pixel_2, pixel_2
2760 ldrb pixel_7, [ uv_7 ]
2761 add pixel_3, pixel_3, pixel_3
2763 ldrh pixel_0, [ clut_ptr, pixel_0 ]
2764 add pixel_4, pixel_4, pixel_4
2766 ldrh pixel_1, [ clut_ptr, pixel_1 ]
2767 add pixel_5, pixel_5, pixel_5
2769 ldrh pixel_2, [ clut_ptr, pixel_2 ]
2770 add pixel_6, pixel_6, pixel_6
2772 ldrh pixel_3, [ clut_ptr, pixel_3 ]
2773 add pixel_7, pixel_7, pixel_7
2775 ldrh pixel_4, [ clut_ptr, pixel_4 ]
2776 orr pixels_a, pixel_0, pixel_1, lsl #16
2778 ldrh pixel_5, [ clut_ptr, pixel_5 ]
2779 orr pixels_c, pixel_2, pixel_3, lsl #16
2781 ldrh pixel_6, [ clut_ptr, pixel_6 ]
2782 subs num_blocks, num_blocks, #1
2784 ldrh pixel_7, [ clut_ptr, pixel_7 ]
2785 orr pixels_b, pixel_4, pixel_5, lsl #16
2787 orr pixels_d, pixel_6, pixel_7, lsl #16
2788 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2790 add block_ptr, block_ptr, #64
2793 ldmia sp!, { r3 - r11, pc }
2796 stmdb sp!, { r1 - r2, r12 }
2798 bl update_texture_8bpp_cache
2800 ldmia sp!, { r1 - r2, r12 }
2830 #define block_ptr r1
2831 #define num_blocks r2
2875 #define texture_ptr r12
2880 function(texture_blocks_16bpp)
2881 stmdb sp!, { r3 - r11, r14 }
2882 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2884 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2885 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2888 ldrh uv_0, [ block_ptr ]
2889 subs num_blocks, num_blocks, #1
2891 ldrh uv_1, [ block_ptr, #2 ]
2893 and v_0, uv_0, #0xFF00
2894 and v_1, uv_1, #0xFF00
2896 and u_0, uv_0, #0xFF
2897 and u_1, uv_1, #0xFF
2899 add uv_0, u_0, v_0, lsl #2
2900 ldrh uv_2, [ block_ptr, #4 ]
2902 add uv_1, u_1, v_1, lsl #2
2903 ldrh uv_3, [ block_ptr, #6 ]
2905 add uv_0, uv_0, uv_0
2906 add uv_1, uv_1, uv_1
2908 and v_2, uv_2, #0xFF00
2909 and v_3, uv_3, #0xFF00
2911 and u_2, uv_2, #0xFF
2912 and u_3, uv_3, #0xFF
2914 add uv_2, u_2, v_2, lsl #2
2915 ldrh uv_4, [ block_ptr, #8 ]
2917 add uv_3, u_3, v_3, lsl #2
2918 ldrh uv_5, [ block_ptr, #10 ]
2920 add uv_2, uv_2, uv_2
2921 add uv_3, uv_3, uv_3
2923 and v_4, uv_4, #0xFF00
2924 and v_5, uv_5, #0xFF00
2926 and u_4, uv_4, #0xFF
2927 and u_5, uv_5, #0xFF
2929 add uv_4, u_4, v_4, lsl #2
2930 ldrh uv_6, [ block_ptr, #12 ]
2932 add uv_5, u_5, v_5, lsl #2
2933 ldrh uv_7, [ block_ptr, #14 ]
2935 add uv_4, uv_4, uv_4
2936 ldrh pixel_0, [ texture_ptr, uv_0 ]
2938 add uv_5, uv_5, uv_5
2939 ldrh pixel_1, [ texture_ptr, uv_1 ]
2941 and v_6, uv_6, #0xFF00
2942 ldrh pixel_2, [ texture_ptr, uv_2 ]
2944 and v_7, uv_7, #0xFF00
2945 ldrh pixel_3, [ texture_ptr, uv_3 ]
2947 and u_6, uv_6, #0xFF
2948 ldrh pixel_4, [ texture_ptr, uv_4 ]
2950 and u_7, uv_7, #0xFF
2951 ldrh pixel_5, [ texture_ptr, uv_5 ]
2953 add uv_6, u_6, v_6, lsl #2
2954 add uv_7, u_7, v_7, lsl #2
2956 add uv_6, uv_6, uv_6
2957 add uv_7, uv_7, uv_7
2959 orr pixels_a, pixel_0, pixel_1, lsl #16
2960 orr pixels_b, pixel_2, pixel_3, lsl #16
2962 ldrh pixel_6, [ texture_ptr, uv_6 ]
2963 orr pixels_c, pixel_4, pixel_5, lsl #16
2965 ldrh pixel_7, [ texture_ptr, uv_7 ]
2966 orr pixels_d, pixel_6, pixel_7, lsl #16
2968 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
2969 add block_ptr, block_ptr, #64
2973 ldmia sp!, { r3 - r11, pc }
2988 #undef msb_mask_high
2996 #define num_blocks r1
2997 #define color_ptr r2
2998 #define colors_scalar r2
2999 #define colors_scalar_compare r3
3000 #define mask_msb_ptr r2
3002 #define block_ptr_load_a r0
3003 #define block_ptr_store r3
3004 #define block_ptr_load_b r12
3009 #define draw_mask_bits_scalar r5
3011 #define d128_0x07 q0
3012 #define d128_0x1F q1
3013 #define d128_0x8000 q2
3014 #define test_mask q3
3016 #define colors_rg q5
3017 #define colors_b_dm_bits q6
3018 #define texels_rg q7
3021 #define pixels_b q10
3023 #define zero_mask q4
3024 #define draw_mask q12
3025 #define msb_mask q13
3027 #define fb_pixels q8
3029 #define pixels_gb_low q9
3031 #define colors_r d10
3032 #define colors_g d11
3033 #define colors_b d12
3034 #define draw_mask_bits d13
3035 #define texels_r d14
3036 #define texels_g d15
3037 #define pixels_r_low d16
3038 #define pixels_g_low d18
3039 #define pixels_b_low d19
3040 #define msb_mask_low d26
3041 #define msb_mask_high d27
3046 #define texels_b d31
3048 #define shade_blocks_textured_modulated_prologue_indirect() \
3050 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3052 #define shade_blocks_textured_modulated_prologue_direct() \
3053 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3054 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ] \
3057 #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \
3059 #define shade_blocks_textured_false_modulation_check_undithered(target) \
3060 ldr colors_scalar, [ psx_gpu, #psx_gpu_triangle_color_offset ]; \
3061 movw colors_scalar_compare, #0x8080; \
3063 movt colors_scalar_compare, #0x80; \
3064 cmp colors_scalar, colors_scalar_compare; \
3065 beq shade_blocks_textured_unmodulated_##target \
3067 #define shade_blocks_textured_false_modulation_check_dithered(target) \
3069 #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \
3070 shade_blocks_textured_false_modulation_check_##dithering(target); \
3071 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3072 vld1.u32 { colors_r[] }, [ color_ptr, :32 ]; \
3073 vdup.u8 colors_g, colors_r[1]; \
3074 vdup.u8 colors_b, colors_r[2]; \
3075 vdup.u8 colors_r, colors_r[0] \
3078 #define shade_blocks_textured_modulated_load_dithered(target) \
3079 vld1.u32 { target }, [ block_ptr_load_b, :128 ] \
3081 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3082 vld1.u32 { target }, [ block_ptr_load_b, :128 ], c_32 \
3084 #define shade_blocks_textured_modulated_load_undithered(target) \
3086 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3087 add block_ptr_load_b, block_ptr_load_b, #32 \
3089 #define shade_blocks_textured_modulate_dithered(channel) \
3090 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3092 #define shade_blocks_textured_modulate_undithered(channel) \
3093 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3096 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3097 vst1.u32 { draw_mask }, [ block_ptr_store, :128 ]! \
3099 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3100 ldr fb_ptr, [ block_ptr_load_b, #(offset - 64) ]; \
3101 vld1.u32 { fb_pixels }, [ fb_ptr ]; \
3102 vbit.u16 pixels, fb_pixels, draw_mask \
3104 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3105 vst1.u32 { pixels }, [ block_ptr_store, :128 ], c_48 \
3107 #define shade_blocks_textured_modulated_store_pixels_direct() \
3108 vst1.u32 { pixels }, [ fb_ptr ] \
3111 #define shade_blocks_textured_modulated_load_rg_shaded() \
3112 vld1.u32 { colors_r, colors_g }, [ block_ptr_load_b, :128 ], c_32 \
3114 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3115 add block_ptr_load_b, block_ptr_load_b, #32 \
3117 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3118 vld1.u32 { colors_b, draw_mask_bits }, [ block_ptr_load_a, :128 ], c_32 \
3120 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3121 ldr draw_mask_bits_scalar, [ block_ptr_load_a, #8 ]; \
3122 add block_ptr_load_a, block_ptr_load_a, #32 \
3124 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3125 vdup.u16 draw_mask, draw_mask_bits[0] \
3127 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3128 vdup.u16 draw_mask, draw_mask_bits_scalar \
3131 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3133 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3134 vorr.u16 pixels, pixels, msb_mask \
3137 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3140 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3141 shade_blocks_textured_modulated_prologue_##shading(dithering, target); \
3142 stmdb sp!, { r4 - r5, lr }; \
3143 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3145 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
3147 shade_blocks_textured_modulated_prologue_##target(); \
3149 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3152 add block_ptr_load_b, block_ptr_load_a, #16; \
3153 vmov.u8 d64_1, #1; \
3154 vmov.u8 d64_4, #4; \
3155 vmov.u8 d64_128, #128; \
3157 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3158 vmov.u8 d128_0x07, #0x07; \
3160 shade_blocks_textured_modulated_load_rg_##shading(); \
3161 vmov.u8 d128_0x1F, #0x1F; \
3163 shade_blocks_textured_modulated_load_bdm_##shading(); \
3164 vmov.u16 d128_0x8000, #0x8000; \
3166 vmovn.u16 texels_r, texels; \
3167 vshrn.u16 texels_g, texels, #5; \
3169 vshrn.u16 texels_b, texels, #7; \
3170 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3172 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3173 vtst.u16 draw_mask, draw_mask, test_mask; \
3175 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3176 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3178 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3179 vshr.u8 texels_b, texels_b, #3; \
3181 shade_blocks_textured_modulate_##dithering(r); \
3182 shade_blocks_textured_modulate_##dithering(g); \
3183 shade_blocks_textured_modulate_##dithering(b); \
3185 vand.u16 pixels, texels, d128_0x8000; \
3186 vceq.u16 zero_mask, texels, #0; \
3188 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3189 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3190 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3192 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3193 vorr.u16 draw_mask, draw_mask, zero_mask; \
3194 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3195 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3197 subs num_blocks, num_blocks, #1; \
3203 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3204 shade_blocks_textured_modulated_load_rg_##shading(); \
3205 vshrn.u16 texels_g, texels, #5; \
3207 shade_blocks_textured_modulated_load_bdm_##shading(); \
3208 vshrn.u16 texels_b, texels, #7; \
3210 pld [ block_ptr_load_a ]; \
3211 vmovn.u16 texels_r, texels; \
3212 vmlal.u8 pixels, pixels_r_low, d64_1; \
3214 vmlal.u8 pixels, pixels_g_low, d64_4; \
3215 vmlal.u8 pixels, pixels_b_low, d64_128; \
3216 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3218 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3219 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3221 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3222 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3224 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3225 vtst.u16 draw_mask, draw_mask, test_mask; \
3227 shade_blocks_textured_modulated_store_pixels_##target(); \
3228 vshr.u8 texels_b, texels_b, #3; \
3230 shade_blocks_textured_modulate_##dithering(r); \
3231 shade_blocks_textured_modulate_##dithering(g); \
3232 shade_blocks_textured_modulate_##dithering(b); \
3234 vand.u16 pixels, texels, d128_0x8000; \
3235 vceq.u16 zero_mask, texels, #0; \
3237 subs num_blocks, num_blocks, #1; \
3239 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3240 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3241 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3243 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3244 vorr.u16 draw_mask, draw_mask, zero_mask; \
3245 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3246 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3251 vmlal.u8 pixels, pixels_r_low, d64_1; \
3252 vmlal.u8 pixels, pixels_g_low, d64_4; \
3253 vmlal.u8 pixels, pixels_b_low, d64_128; \
3255 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3256 shade_blocks_textured_modulated_store_pixels_##target(); \
3258 ldmia sp!, { r4 - r5, pc } \
3261 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3262 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3263 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3264 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3266 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3267 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3268 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3269 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3287 #undef msb_mask_high
3290 #define num_blocks r1
3291 #define mask_msb_ptr r2
3292 #define color_ptr r3
3294 #define block_ptr_load r0
3295 #define draw_mask_store_ptr r3
3296 #define draw_mask_bits_ptr r12
3297 #define draw_mask_ptr r12
3298 #define pixel_store_ptr r14
3300 #define fb_ptr_cmp r4
3303 #define fb_ptr_next r14
3307 #define test_mask q0
3309 #define draw_mask q2
3310 #define zero_mask q3
3311 #define draw_mask_combined q4
3312 #define fb_pixels q5
3313 #define fb_pixels_next q6
3316 #define draw_mask_low d4
3317 #define draw_mask_high d5
3318 #define msb_mask_low d14
3319 #define msb_mask_high d15
3322 function(shade_blocks_textured_unmodulated_indirect)
3323 str r14, [ sp, #-4 ]
3324 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3326 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3327 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3329 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3330 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3333 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3335 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3336 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3337 [ draw_mask_bits_ptr, :16 ], c_64
3338 vceq.u16 zero_mask, pixels, #0
3340 vtst.u16 draw_mask, draw_mask, test_mask
3341 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3343 subs num_blocks, num_blocks, #1
3347 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3348 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3350 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3351 [ draw_mask_bits_ptr, :16 ], c_64
3352 vceq.u16 zero_mask, pixels, #0
3354 vtst.u16 draw_mask, draw_mask, test_mask
3355 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3357 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3358 subs num_blocks, num_blocks, #1
3363 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3364 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3371 function(shade_blocks_textured_unmodulated_direct)
3372 stmdb sp!, { r4, r14 }
3373 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3375 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3376 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3378 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3381 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3382 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3384 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3385 [ draw_mask_bits_ptr, :16 ], c_64
3386 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3388 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3389 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3390 vceq.u16 zero_mask, pixels, #0
3391 vtst.u16 draw_mask, draw_mask, test_mask
3393 subs num_blocks, num_blocks, #1
3397 mov fb_ptr, fb_ptr_next
3398 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3400 vorr.u16 pixels, pixels, msb_mask
3402 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3403 vmov fb_pixels, fb_pixels_next
3405 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3406 [ draw_mask_bits_ptr, :16 ], c_64
3407 vbif.u16 fb_pixels, pixels, draw_mask_combined
3409 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3410 pld [ fb_ptr_next, #64 ]
3412 add fb_ptr_cmp, fb_ptr_cmp, #14
3413 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3418 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3419 vceq.u16 zero_mask, pixels, #0
3421 vst1.u16 { fb_pixels }, [ fb_ptr ]
3422 vtst.u16 draw_mask, draw_mask, test_mask
3425 subs num_blocks, num_blocks, #1
3429 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3430 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3432 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3434 ldmia sp!, { r4, pc }
3437 vst1.u16 { fb_pixels }, [ fb_ptr ]
3438 vceq.u16 zero_mask, pixels, #0
3440 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3441 vtst.u16 draw_mask, draw_mask, test_mask
3446 function(shade_blocks_unshaded_untextured_indirect)
3451 function(shade_blocks_unshaded_untextured_direct)
3452 stmdb sp!, { r4, r14 }
3453 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3455 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3456 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3458 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3459 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3461 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3462 vld1.u16 { pixels }, [ color_ptr, :128 ]
3465 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3467 vorr.u16 pixels, pixels, msb_mask
3468 subs num_blocks, num_blocks, #1
3470 ldr fb_ptr_next, [ block_ptr_load ], #64
3472 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3476 vmov fb_pixels, fb_pixels_next
3477 mov fb_ptr, fb_ptr_next
3478 ldr fb_ptr_next, [ block_ptr_load ], #64
3480 vbif.u16 fb_pixels, pixels, draw_mask
3481 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3483 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3484 add fb_ptr_cmp, fb_ptr_cmp, #14
3488 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3489 vst1.u16 { fb_pixels }, [ fb_ptr ]
3492 subs num_blocks, num_blocks, #1
3496 vbif.u16 fb_pixels_next, pixels, draw_mask
3497 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3499 ldmia sp!, { r4, pc }
3502 vst1.u16 { fb_pixels }, [ fb_ptr ]
3503 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3507 #undef draw_mask_ptr
3514 #define num_blocks r1
3515 #define msb_mask_ptr r2
3516 #define pixel_ptr r3
3517 #define draw_mask_ptr r0
3520 #define fb_ptr_next r14
3521 #define fb_ptr_cmp r4
3529 #undef msb_mask_high
3530 #undef draw_mask_next
3533 #undef fb_pixels_next
3536 #define draw_mask q1
3538 #define fb_pixels q3
3539 #define blend_pixels q4
3540 #define pixels_no_msb q5
3541 #define blend_mask q6
3542 #define fb_pixels_no_msb q7
3543 #define d128_0x8000 q8
3544 #define d128_0x0421 q9
3545 #define fb_pixels_next q10
3546 #define blend_pixels_next q11
3547 #define pixels_next q12
3548 #define draw_mask_next q13
3549 #define write_mask q14
3551 #define pixels_rb q5
3552 #define pixels_mg q7
3554 #define d128_0x7C1F q8
3555 #define d128_0x03E0 q9
3556 #define fb_pixels_rb q10
3557 #define fb_pixels_g q11
3558 #define fb_pixels_masked q11
3559 #define d128_0x83E0 q15
3560 #define pixels_fourth q7
3561 #define d128_0x1C07 q12
3562 #define d128_0x00E0 q13
3563 #define d128_0x80E0 q13
3565 #define msb_mask_low d0
3566 #define msb_mask_high d1
3568 #define blend_blocks_average_set_blend_mask_textured(source) \
3569 vclt.s16 blend_mask, source, #0 \
3571 #define blend_blocks_average_set_stp_bit_textured() \
3572 vorr.u16 blend_pixels, #0x8000 \
3574 #define blend_blocks_average_combine_textured(source) \
3575 vbif.u16 blend_pixels, source, blend_mask \
3577 #define blend_blocks_average_set_blend_mask_untextured(source) \
3579 #define blend_blocks_average_set_stp_bit_untextured() \
3581 #define blend_blocks_average_combine_untextured(source) \
3583 #define blend_blocks_average_mask_set_on() \
3584 vclt.s16 write_mask, fb_pixels_next, #0 \
3586 #define blend_blocks_average_mask_copy_on() \
3587 vorr.u16 draw_mask, draw_mask_next, write_mask \
3589 #define blend_blocks_average_mask_copy_b_on() \
3590 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3592 #define blend_blocks_average_mask_set_off() \
3594 #define blend_blocks_average_mask_copy_off() \
3595 vmov draw_mask, draw_mask_next \
3597 #define blend_blocks_average_mask_copy_b_off() \
3599 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3602 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3603 stmdb sp!, { r4, r14 }; \
3604 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3605 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3607 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3608 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3610 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3613 vmov.u16 d128_0x8000, #0x8000; \
3614 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3615 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3617 vmov.u16 d128_0x0421, #0x0400; \
3618 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3620 vorr.u16 d128_0x0421, #0x0021; \
3621 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3623 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3624 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3625 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3626 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3627 blend_blocks_average_mask_set_##mask_evaluate(); \
3628 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3630 subs num_blocks, num_blocks, #1; \
3634 mov fb_ptr, fb_ptr_next; \
3635 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3637 vmov pixels, pixels_next; \
3638 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3640 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3642 blend_blocks_average_mask_copy_##mask_evaluate(); \
3643 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3645 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3646 blend_blocks_average_set_stp_bit_##texturing(); \
3647 vmov fb_pixels, fb_pixels_next; \
3648 blend_blocks_average_combine_##texturing(pixels); \
3650 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3651 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3652 cmp fb_ptr_cmp, #28; \
3655 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3656 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3658 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3659 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3661 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3662 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3664 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3665 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3666 blend_blocks_average_mask_set_##mask_evaluate(); \
3667 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3670 subs num_blocks, num_blocks, #1; \
3674 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3675 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3677 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3678 blend_blocks_average_set_stp_bit_##texturing(); \
3679 blend_blocks_average_combine_##texturing(pixels_next); \
3681 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3682 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3683 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3685 ldmia sp!, { r4, pc }; \
3688 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3689 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3690 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3692 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3693 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3694 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3695 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3696 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3697 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3701 blend_blocks_average_builder(textured, off)
3702 blend_blocks_average_builder(untextured, off)
3703 blend_blocks_average_builder(textured, on)
3704 blend_blocks_average_builder(untextured, on)
3707 #define blend_blocks_add_mask_set_on() \
3708 vclt.s16 write_mask, fb_pixels, #0 \
3710 #define blend_blocks_add_mask_copy_on() \
3711 vorr.u16 draw_mask, draw_mask, write_mask \
3713 #define blend_blocks_add_mask_set_off() \
3715 #define blend_blocks_add_mask_copy_off() \
3718 #define blend_blocks_add_textured_builder(mask_evaluate) \
3721 function(blend_blocks_textured_add_##mask_evaluate) \
3722 stmdb sp!, { r4, r14 }; \
3723 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3724 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3726 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3727 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3729 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3732 vmov.u16 d128_0x7C1F, #0x7C00; \
3733 vmov.u16 d128_0x03E0, #0x0300; \
3734 vmov.u16 d128_0x83E0, #0x8000; \
3735 vorr.u16 d128_0x03E0, #0x00E0; \
3736 vorr.u16 d128_0x7C1F, #0x001F; \
3737 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3739 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3740 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3741 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3742 vclt.s16 blend_mask, pixels, #0; \
3743 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3744 blend_blocks_add_mask_set_##mask_evaluate(); \
3745 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3747 blend_blocks_add_mask_copy_##mask_evaluate(); \
3748 vorr.u16 pixels, pixels, msb_mask; \
3749 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3750 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3751 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3752 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3753 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3754 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3755 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3756 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3758 subs num_blocks, num_blocks, #1; \
3762 mov fb_ptr, fb_ptr_next; \
3764 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3766 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3767 vclt.s16 blend_mask, pixels, #0; \
3769 vorr.u16 pixels, pixels, msb_mask; \
3770 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3771 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3773 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3774 pld [ fb_ptr_next, #64 ]; \
3776 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3777 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3779 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3780 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3782 cmp fb_ptr_cmp, #28; \
3785 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3786 blend_blocks_add_mask_set_##mask_evaluate(); \
3787 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3788 blend_blocks_add_mask_copy_##mask_evaluate(); \
3789 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3790 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3791 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3794 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3795 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3796 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3797 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3798 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3800 subs num_blocks, num_blocks, #1; \
3804 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3805 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3806 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3808 ldmia sp!, { r4, pc }; \
3811 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3812 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3814 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3815 blend_blocks_add_mask_set_##mask_evaluate(); \
3816 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3817 blend_blocks_add_mask_copy_##mask_evaluate(); \
3818 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3822 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3825 function(blend_blocks_untextured_add_##mask_evaluate) \
3826 stmdb sp!, { r4, r14 }; \
3827 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3828 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3830 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3831 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3833 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3836 vmov.u16 d128_0x7C1F, #0x7C00; \
3837 vmov.u16 d128_0x03E0, #0x0300; \
3838 vorr.u16 d128_0x7C1F, #0x001F; \
3839 vorr.u16 d128_0x03E0, #0x00E0; \
3841 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3842 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3843 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3844 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3845 blend_blocks_add_mask_set_##mask_evaluate(); \
3846 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3848 blend_blocks_add_mask_copy_##mask_evaluate(); \
3849 vand.u16 pixels_g, pixels, d128_0x03E0; \
3850 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3851 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3852 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3853 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3854 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3855 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3857 subs num_blocks, num_blocks, #1; \
3861 mov fb_ptr, fb_ptr_next; \
3863 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3865 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3867 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3868 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3869 vand.u16 pixels_g, pixels, d128_0x03E0; \
3871 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3872 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3874 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3875 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3876 cmp fb_ptr_cmp, #28; \
3879 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3880 blend_blocks_add_mask_set_##mask_evaluate(); \
3881 blend_blocks_add_mask_copy_##mask_evaluate(); \
3882 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3883 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3884 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3887 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3888 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3889 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3890 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3891 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3893 subs num_blocks, num_blocks, #1; \
3897 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3898 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3899 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3900 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3902 ldmia sp!, { r4, pc }; \
3905 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3906 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3908 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3909 blend_blocks_add_mask_set_##mask_evaluate(); \
3910 blend_blocks_add_mask_copy_##mask_evaluate(); \
3911 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3915 blend_blocks_add_textured_builder(off)
3916 blend_blocks_add_textured_builder(on)
3917 blend_blocks_add_untextured_builder(off)
3918 blend_blocks_add_untextured_builder(on)
3920 #define blend_blocks_subtract_set_blend_mask_textured() \
3921 vclt.s16 blend_mask, pixels_next, #0 \
3923 #define blend_blocks_subtract_combine_textured() \
3924 vbif.u16 blend_pixels, pixels, blend_mask \
3926 #define blend_blocks_subtract_set_stb_textured() \
3927 vorr.u16 blend_pixels, #0x8000 \
3929 #define blend_blocks_subtract_msb_mask_textured() \
3930 vorr.u16 pixels, pixels_next, msb_mask \
3932 #define blend_blocks_subtract_set_blend_mask_untextured() \
3934 #define blend_blocks_subtract_combine_untextured() \
3936 #define blend_blocks_subtract_set_stb_untextured() \
3937 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3939 #define blend_blocks_subtract_msb_mask_untextured() \
3942 #define blend_blocks_subtract_mask_set_on() \
3943 vclt.s16 write_mask, fb_pixels, #0 \
3945 #define blend_blocks_subtract_mask_copy_on() \
3946 vorr.u16 draw_mask, draw_mask_next, write_mask \
3948 #define blend_blocks_subtract_mask_set_off() \
3950 #define blend_blocks_subtract_mask_copy_off() \
3951 vmov draw_mask, draw_mask_next \
3954 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
3957 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
3958 stmdb sp!, { r4, r14 }; \
3959 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3960 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3962 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3963 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3965 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3968 vmov.u16 d128_0x7C1F, #0x7C00; \
3969 vmov.u16 d128_0x03E0, #0x0300; \
3970 vorr.u16 d128_0x7C1F, #0x001F; \
3971 vorr.u16 d128_0x03E0, #0x00E0; \
3973 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3974 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3975 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3976 blend_blocks_subtract_set_blend_mask_##texturing(); \
3977 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3978 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3979 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3981 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3982 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3983 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3984 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3985 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3987 subs num_blocks, num_blocks, #1; \
3991 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
3992 mov fb_ptr, fb_ptr_next; \
3993 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3995 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3996 blend_blocks_subtract_msb_mask_##texturing(); \
3998 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3999 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4000 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4001 blend_blocks_subtract_set_stb_##texturing(); \
4002 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4003 blend_blocks_subtract_combine_##texturing(); \
4004 blend_blocks_subtract_set_blend_mask_##texturing(); \
4005 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4007 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4008 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4009 cmp fb_ptr_cmp, #28; \
4012 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4013 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4014 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4015 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4016 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4017 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4018 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4021 subs num_blocks, num_blocks, #1; \
4025 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4027 blend_blocks_subtract_msb_mask_##texturing(); \
4028 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4029 blend_blocks_subtract_set_stb_##texturing(); \
4030 blend_blocks_subtract_combine_##texturing(); \
4031 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4032 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4034 ldmia sp!, { r4, pc }; \
4037 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4038 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4039 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4040 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4041 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4042 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4043 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4047 blend_blocks_subtract_builder(textured, off)
4048 blend_blocks_subtract_builder(textured, on)
4049 blend_blocks_subtract_builder(untextured, off)
4050 blend_blocks_subtract_builder(untextured, on)
4053 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4056 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4057 stmdb sp!, { r4, r14 }; \
4058 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4059 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4061 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4062 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4064 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4067 vmov.u16 d128_0x7C1F, #0x7C00; \
4068 vmov.u16 d128_0x03E0, #0x0300; \
4069 vmov.u16 d128_0x1C07, #0x1C00; \
4070 vmov.u16 d128_0x00E0, #0x00E0; \
4071 vorr.u16 d128_0x7C1F, #0x001F; \
4072 vorr.u16 d128_0x03E0, #0x00E0; \
4073 vorr.u16 d128_0x1C07, #0x0007; \
4075 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4076 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4077 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4078 vclt.s16 blend_mask, pixels, #0; \
4079 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4080 blend_blocks_add_mask_set_##mask_evaluate(); \
4081 vshr.s16 pixels_fourth, pixels, #2; \
4082 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4084 blend_blocks_add_mask_copy_##mask_evaluate(); \
4085 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4086 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4087 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4088 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4089 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4090 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4091 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4093 subs num_blocks, num_blocks, #1; \
4097 mov fb_ptr, fb_ptr_next; \
4098 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4100 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4101 vbif.u16 blend_pixels, pixels, blend_mask; \
4103 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4104 vclt.s16 blend_mask, pixels, #0; \
4105 vshr.s16 pixels_fourth, pixels, #2; \
4106 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4107 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4109 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4110 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4112 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4113 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4114 cmp fb_ptr_cmp, #28; \
4117 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4118 blend_blocks_add_mask_set_##mask_evaluate(); \
4119 blend_blocks_add_mask_copy_##mask_evaluate(); \
4120 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4121 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4122 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4125 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4126 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4127 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4128 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4129 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4131 subs num_blocks, num_blocks, #1; \
4135 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4136 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4137 vbif.u16 blend_pixels, pixels, blend_mask; \
4138 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4139 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4141 ldmia sp!, { r4, pc }; \
4144 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4145 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4147 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4148 blend_blocks_add_mask_set_##mask_evaluate(); \
4149 blend_blocks_add_mask_copy_##mask_evaluate(); \
4150 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4155 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4158 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4159 stmdb sp!, { r4, r14 }; \
4160 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4161 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4163 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4164 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4166 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4169 vmov.u16 d128_0x7C1F, #0x7C00; \
4170 vmov.u16 d128_0x03E0, #0x0300; \
4171 vmov.u16 d128_0x1C07, #0x1C00; \
4172 vmov.u16 d128_0x00E0, #0x00E0; \
4173 vorr.u16 d128_0x7C1F, #0x001F; \
4174 vorr.u16 d128_0x03E0, #0x00E0; \
4175 vorr.u16 d128_0x1C07, #0x0007; \
4177 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4178 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4179 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4180 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4181 blend_blocks_add_mask_set_##mask_evaluate(); \
4182 vshr.s16 pixels_fourth, pixels, #2; \
4183 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4185 blend_blocks_add_mask_copy_##mask_evaluate(); \
4186 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4187 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4188 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4189 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4190 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4191 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4192 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4194 subs num_blocks, num_blocks, #1; \
4198 mov fb_ptr, fb_ptr_next; \
4199 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4201 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4203 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4204 vshr.s16 pixels_fourth, pixels, #2; \
4205 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4206 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4208 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4209 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4211 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4212 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4213 cmp fb_ptr_cmp, #28; \
4216 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4217 blend_blocks_add_mask_set_##mask_evaluate(); \
4218 blend_blocks_add_mask_copy_##mask_evaluate(); \
4219 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4220 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4221 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4224 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4225 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4226 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4227 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4228 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4230 subs num_blocks, num_blocks, #1; \
4234 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4235 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4236 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4237 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4239 ldmia sp!, { r4, pc }; \
4242 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4243 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4245 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4246 blend_blocks_add_mask_set_##mask_evaluate(); \
4247 blend_blocks_add_mask_copy_##mask_evaluate(); \
4248 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4252 blend_blocks_add_fourth_textured_builder(off)
4253 blend_blocks_add_fourth_textured_builder(on)
4254 blend_blocks_add_fourth_untextured_builder(off)
4255 blend_blocks_add_fourth_untextured_builder(on)
4257 // TODO: Optimize this more. Need a scene that actually uses it for
4262 function(blend_blocks_textured_unblended_on)
4263 stmdb sp!, { r4, r14 }
4264 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4265 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
4267 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4268 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
4270 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4273 ldr fb_ptr, [ pixel_ptr, #28 ]
4274 vld1.u16 { fb_pixels }, [ fb_ptr ]
4275 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4276 vclt.s16 write_mask, fb_pixels, #0
4277 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4279 subs num_blocks, num_blocks, #1
4283 vorr.u16 pixels, pixels, msb_mask
4284 vorr.u16 draw_mask, draw_mask, write_mask
4285 vbif.u16 fb_pixels, pixels, draw_mask
4286 vst1.u16 { fb_pixels }, [ fb_ptr ]
4288 ldr fb_ptr, [ pixel_ptr, #28 ]
4289 vld1.u16 { fb_pixels }, [ fb_ptr ]
4290 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4291 vclt.s16 write_mask, fb_pixels, #0
4292 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4294 subs num_blocks, num_blocks, #1
4298 vorr.u16 pixels, pixels, msb_mask
4299 vorr.u16 draw_mask, draw_mask, write_mask
4300 vbif.u16 fb_pixels, pixels, draw_mask
4301 vst1.u16 { fb_pixels }, [ fb_ptr ]
4303 ldmia sp!, { r4, pc }
4306 function(blend_blocks_textured_unblended_off)
4316 vld1.u32 { u_whole_8, v_whole_8 }, [ r1, :128 ], r3
4336 #define num_width r12
4346 function(render_block_fill_body)
4347 vdup.u16 colors_a, color
4350 vmov colors_b, colors_a
4351 sub pitch, pitch, width, lsl #1
4353 mov num_width, width
4356 vst1.u32 { colors_a, colors_b }, [ vram_ptr, :256 ]!
4358 subs num_width, num_width, #16
4361 add vram_ptr, vram_ptr, pitch
4362 mov num_width, width
4364 subs height, height, #1
4378 #undef dirty_textures_mask
4380 #undef current_texture_mask
4391 #define offset_u_right r10
4392 #define width_rounded r11
4393 #define height_rounded r12
4395 #define texture_offset_base r1
4396 #define tile_width r2
4397 #define tile_height r3
4398 #define num_blocks r4
4400 #define sub_tile_height r6
4402 #define texture_mask r8
4403 #define column_data r9
4404 #define texture_offset r10
4405 #define tiles_remaining r11
4406 #define fb_ptr_advance_column r12
4407 #define texture_block_ptr r14
4409 #define texture_page_ptr r3
4410 #define left_block_mask r4
4411 #define right_block_mask r5
4412 #define texture_mask_rev r10
4413 #define control_mask r11
4415 #define dirty_textures_mask r4
4417 #define current_texture_mask r6
4431 #define draw_masks_fb_ptrs q1
4433 #define draw_mask_fb_ptr_left d2
4434 #define draw_mask_fb_ptr_right d3
4436 #define draw_mask_fb_ptr_left_a d2
4437 #define draw_mask_fb_ptr_left_b d3
4438 #define draw_mask_fb_ptr_right_a d10
4439 #define draw_mask_fb_ptr_right_b d11
4440 #define draw_masks_fb_ptrs2 q5
4442 #define clut_low_a d4
4443 #define clut_low_b d5
4444 #define clut_high_a d6
4445 #define clut_high_b d7
4447 #define block_masks d8
4448 #define block_masks_shifted d9
4453 #define texels_low d12
4454 #define texels_high d13
4456 #define texels_wide_low d14
4457 #define texels_wide_high d15
4458 #define texels_wide q7
4461 setup_sprite_flush_blocks:
4464 stmdb sp!, { r0 - r3, r12, r14 }
4465 bl flush_render_block_buffer
4466 ldmia sp!, { r0 - r3, r12, r14 }
4470 add block, psx_gpu, #psx_gpu_blocks_offset
4474 setup_sprite_update_texture_4bpp_cache:
4475 stmdb sp!, { r0 - r3, r14 }
4476 bl update_texture_4bpp_cache
4477 ldmia sp!, { r0 - r3, pc }
4480 setup_sprite_update_texture_8bpp_cache:
4481 stmdb sp!, { r0 - r3, r14 }
4482 bl update_texture_8bpp_cache
4483 ldmia sp!, { r0 - r3, pc }
4486 #define setup_sprite_tiled_initialize_4bpp() \
4487 ldr dirty_textures_mask, \
4488 [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]; \
4489 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4491 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4492 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4494 tst current_texture_mask, dirty_textures_mask; \
4495 vuzp.u8 clut_a, clut_b; \
4497 blne setup_sprite_update_texture_4bpp_cache \
4499 #define setup_sprite_tiled_initialize_8bpp() \
4500 ldr dirty_textures_mask, \
4501 [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]; \
4502 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4504 tst current_texture_mask, dirty_textures_mask; \
4505 blne setup_sprite_update_texture_8bpp_cache \
4508 #define setup_sprite_block_count_single() \
4511 #define setup_sprite_block_count_double() \
4512 sub_tile_height, lsl #1 \
4514 #define setup_sprite_tile_add_blocks(type) \
4515 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4516 cmp num_blocks, #MAX_BLOCKS; \
4518 movgt num_blocks, setup_sprite_block_count_##type(); \
4519 blgt setup_sprite_flush_blocks \
4522 #define setup_sprite_tile_full_4bpp(edge) \
4523 setup_sprite_tile_add_blocks(double); \
4526 and texture_block_ptr, texture_offset, texture_mask; \
4527 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4530 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4531 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4533 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4534 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4536 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4537 add texture_block_ptr, texture_offset, #8; \
4539 and texture_block_ptr, texture_block_ptr, texture_mask; \
4540 add block, block, #40; \
4542 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4543 add fb_ptr, fb_ptr, #16; \
4545 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4546 add block, block, #24; \
4548 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4549 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4552 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4553 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4555 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4556 add block, block, #40; \
4558 add texture_offset, texture_offset, #0x10; \
4559 add fb_ptr, fb_ptr, #(2048 - 16); \
4561 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4562 add block, block, #24; \
4564 subs sub_tile_height, sub_tile_height, #1; \
4567 add texture_offset, texture_offset, #0xF00; \
4568 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4571 #define setup_sprite_tile_half_4bpp(edge) \
4572 setup_sprite_tile_add_blocks(single); \
4575 and texture_block_ptr, texture_offset, texture_mask; \
4576 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4579 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4580 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4582 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4583 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4585 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4586 add block, block, #40; \
4588 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4589 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4591 add block, block, #24; \
4592 add texture_offset, texture_offset, #0x10; \
4594 add fb_ptr, fb_ptr, #2048; \
4595 subs sub_tile_height, sub_tile_height, #1; \
4599 add texture_offset, texture_offset, #0xF00; \
4600 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4603 #define setup_sprite_tile_full_8bpp(edge) \
4604 setup_sprite_tile_add_blocks(double); \
4605 add block, block, #16; \
4608 and texture_block_ptr, texture_offset, texture_mask; \
4609 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4612 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4613 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4615 add texture_block_ptr, texture_offset, #8; \
4616 vst1.u32 { texels }, [ block, :64 ]; \
4618 and texture_block_ptr, texture_block_ptr, texture_mask; \
4619 add block, block, #24; \
4621 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4623 add fb_ptr, fb_ptr, #16; \
4624 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4626 add block, block, #40; \
4627 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4630 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4631 vst1.u32 { texels }, [ block, :64 ]; \
4632 add block, block, #24; \
4634 add texture_offset, texture_offset, #0x10; \
4635 add fb_ptr, fb_ptr, #(2048 - 16); \
4637 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4638 add block, block, #40; \
4640 subs sub_tile_height, sub_tile_height, #1; \
4643 sub block, block, #16; \
4644 add texture_offset, texture_offset, #0xF00; \
4645 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4648 #define setup_sprite_tile_half_8bpp(edge) \
4649 setup_sprite_tile_add_blocks(single); \
4650 add block, block, #16; \
4653 and texture_block_ptr, texture_offset, texture_mask; \
4654 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4657 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4658 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4660 vst1.u32 { texels }, [ block, :64 ]; \
4661 add block, block, #24; \
4663 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4664 add block, block, #40; \
4666 add texture_offset, texture_offset, #0x10; \
4667 add fb_ptr, fb_ptr, #2048; \
4669 subs sub_tile_height, sub_tile_height, #1; \
4672 sub block, block, #16; \
4673 add texture_offset, texture_offset, #0xF00; \
4674 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4677 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4678 add texture_offset, texture_offset_base, #8; \
4679 add fb_ptr, fb_ptr, #16 \
4681 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4682 mov texture_offset, texture_offset_base \
4684 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4685 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4687 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4688 mov texture_offset, texture_offset_base \
4690 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4691 sub fb_ptr, fb_ptr, #16 \
4693 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4695 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4696 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4698 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4701 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \
4703 mov sub_tile_height, column_data; \
4704 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4705 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4706 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4708 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \
4710 and sub_tile_height, column_data, #0xFF; \
4711 mov tiles_remaining, column_data, lsr #16; \
4712 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4713 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4715 subs tiles_remaining, tiles_remaining, #1; \
4719 mov sub_tile_height, #16; \
4720 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4721 subs tiles_remaining, tiles_remaining, #1; \
4725 uxtb sub_tile_height, column_data, ror #8; \
4726 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4727 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4730 #define setup_sprite_column_data_single() \
4731 mov column_data, height; \
4732 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] \
4734 #define setup_sprite_column_data_multi() \
4735 and height_rounded, height_rounded, #0xF; \
4736 rsb column_data, offset_v, #16; \
4738 add height_rounded, height_rounded, #1; \
4739 sub tile_height, tile_height, #1; \
4741 orr column_data, column_data, tile_height, lsl #16; \
4742 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]; \
4744 orr column_data, column_data, height_rounded, lsl #8 \
4746 #define setup_sprite_setup_left_draw_mask_fb_ptr() \
4747 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4748 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4750 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column() \
4751 mov fb_ptr_advance_column, #32; \
4752 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4754 sub fb_ptr_advance_column, height, lsl #11; \
4755 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4757 #define setup_sprite_setup_right_draw_mask_fb_ptr() \
4758 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4759 vdup.u8 draw_mask_fb_ptr_right, block_masks[5] \
4761 #define setup_sprite_tile_column_width_single(tm, multi_height, edge_mode, \
4763 setup_sprite_##tm##_single_##multi_height##_##edge_mode##_##edge##x4mode: \
4764 setup_sprite_column_data_##multi_height(); \
4765 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4766 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4767 setup_sprite_setup_left_draw_mask_fb_ptr##x4mode(); \
4769 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, tm, x4mode); \
4770 ldmia sp!, { r4 - r11, pc } \
4772 #define setup_sprite_tiled_advance_column() \
4773 add texture_offset_base, texture_offset_base, #0x100; \
4774 tst texture_offset_base, #0xF00; \
4775 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4777 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4778 right_mode, x4mode) \
4779 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode##x4mode:\
4780 setup_sprite_column_data_##multi_height(); \
4782 setup_sprite_setup_left_draw_mask_fb_ptr_advance_column##x4mode(); \
4784 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm, x4mode);\
4786 subs tile_width, tile_width, #2; \
4787 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4791 vmov.u8 draw_masks_fb_ptrs, #0; \
4792 vmov.u8 draw_masks_fb_ptrs2, #0; \
4795 setup_sprite_tiled_advance_column(); \
4796 setup_sprite_tile_column_height_##multi_height(full, none, tm, x4mode); \
4797 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4798 subs tile_width, tile_width, #1; \
4802 setup_sprite_setup_right_draw_mask_fb_ptr##x4mode(); \
4804 setup_sprite_tiled_advance_column(); \
4805 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm, x4mode);\
4806 ldmia sp!, { r4 - r11, pc } \
4809 #define setup_sprite_offset_u_adjust() \
4811 #define setup_sprite_get_left_block_mask() \
4812 and left_block_mask, left_block_mask, #0xFF \
4814 #define setup_sprite_compare_left_block_mask() \
4815 cmp left_block_mask, #0xFF \
4817 #define setup_sprite_get_right_block_mask() \
4818 uxtb right_block_mask, right_block_mask, ror #8 \
4820 #define setup_sprite_compare_right_block_mask() \
4821 cmp right_block_mask, #0xFF \
4826 #define fb_ptr2 column_data
4828 #define setup_sprite_offset_u_adjust_4x() \
4829 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4830 lsl offset_u_right, #1; \
4832 add offset_u_right, #1 \
4834 #define setup_sprite_get_left_block_mask_4x() \
4835 sxth left_block_mask, left_block_mask \
4837 #define setup_sprite_compare_left_block_mask_4x() \
4838 cmp left_block_mask, #0xFFFFFFFF \
4840 #define setup_sprite_get_right_block_mask_4x() \
4841 sxth right_block_mask, right_block_mask, ror #16 \
4843 #define setup_sprite_compare_right_block_mask_4x() \
4844 cmp right_block_mask, #0xFFFFFFFF \
4847 #define widen_texels_16bpp(texels_) \
4848 vmov texels_wide_low, texels_; \
4849 vmov texels_wide_high, texels_; \
4850 vzip.16 texels_wide_low, texels_wide_high \
4852 #define widen_texels_8bpp(texels_) \
4853 vmov texels_wide_low, texels_; \
4854 vmov texels_wide_high, texels_; \
4855 vzip.8 texels_wide_low, texels_wide_high \
4857 #define write_block_16bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4858 vst1.u32 { texels_ }, [ block_, :128 ]; \
4859 add block_, block_, #40; \
4861 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4862 vst1.u32 { draw_mask_fb_ptr_ }, [ block_, :64 ]; \
4863 add block_, block_, #24 \
4865 /* assumes 16-byte offset already added to block_ */
4866 #define write_block_8bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4867 vst1.u32 { texels_ }, [ block_, :64 ]; \
4868 add block_, block_, #24; \
4870 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4871 vst1.u32 { draw_mask_fb_ptr_ }, [ block_, :64 ]; \
4872 add block_, block_, #40 \
4874 #define do_texture_block_16bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4875 draw_mask_fb_ptr_b_) \
4876 widen_texels_16bpp(texels_low); \
4877 add fb_ptr_tmp, fb_ptr, #1024*2; \
4879 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr); \
4881 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4882 widen_texels_16bpp(texels_high); \
4884 add fb_ptr_tmp, fb_ptr, #8*2; \
4885 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4887 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4888 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4890 #define do_texture_block_8bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4891 draw_mask_fb_ptr_b_) \
4892 widen_texels_8bpp(texels); \
4893 add fb_ptr_tmp, fb_ptr, #1024*2; \
4895 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr); \
4896 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4898 add fb_ptr_tmp, fb_ptr, #8*2; \
4899 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4901 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4902 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4905 #define setup_sprite_tiled_initialize_4bpp_4x() \
4906 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4907 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4909 vuzp.u8 clut_a, clut_b \
4911 #define setup_sprite_tiled_initialize_8bpp_4x() \
4914 #define setup_sprite_block_count_single_4x() \
4915 sub_tile_height, lsl #2 \
4917 #define setup_sprite_block_count_double_4x() \
4918 sub_tile_height, lsl #(1+2) \
4920 #define setup_sprite_tile_full_4bpp_4x(edge) \
4921 setup_sprite_tile_add_blocks(double_4x); \
4922 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4925 and texture_block_ptr, texture_offset, texture_mask; \
4928 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4929 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4931 add texture_block_ptr, texture_offset, #8; \
4932 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4934 and texture_block_ptr, texture_block_ptr, texture_mask; \
4935 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4937 vzip.8 texels_low, texels_high; \
4938 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
4939 draw_mask_fb_ptr_left_b); \
4941 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4942 pld [ fb_ptr, #2048 ]; \
4944 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4945 add fb_ptr, fb_ptr, #16*2; \
4947 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4948 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4950 vzip.8 texels_low, texels_high; \
4951 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
4952 draw_mask_fb_ptr_right_b); \
4954 add texture_offset, texture_offset, #0x10; \
4955 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
4957 subs sub_tile_height, sub_tile_height, #1; \
4960 ldr column_data, [sp], #8; /* fb_ptr2 */ \
4961 add texture_offset, texture_offset, #0xF00; \
4962 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4965 #define setup_sprite_tile_half_4bpp_4x(edge) \
4966 setup_sprite_tile_add_blocks(single_4x); \
4967 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4970 and texture_block_ptr, texture_offset, texture_mask; \
4973 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4974 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4976 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4977 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4979 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4980 add texture_offset, texture_offset, #0x10; \
4982 vzip.8 texels_low, texels_high; \
4983 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
4984 draw_mask_fb_ptr_##edge##_b); \
4986 pld [ fb_ptr, #2048 ]; \
4987 add fb_ptr, fb_ptr, #2048 * 2; \
4989 subs sub_tile_height, sub_tile_height, #1; \
4992 ldr column_data, [sp], #8; /* fb_ptr2 */ \
4993 add texture_offset, texture_offset, #0xF00; \
4994 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4997 #define setup_sprite_tile_full_8bpp_4x(edge) \
4998 setup_sprite_tile_add_blocks(double_4x); \
4999 add block, block, #16; \
5000 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5003 and texture_block_ptr, texture_offset, texture_mask; \
5006 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5007 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
5009 add texture_block_ptr, texture_offset, #8; \
5010 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
5011 draw_mask_fb_ptr_left_b); \
5013 pld [ fb_ptr, #2048 ]; \
5014 and texture_block_ptr, texture_block_ptr, texture_mask; \
5016 add fb_ptr, fb_ptr, #16*2; \
5017 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5019 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
5021 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
5022 draw_mask_fb_ptr_right_b); \
5024 add texture_offset, texture_offset, #0x10; \
5025 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
5027 subs sub_tile_height, sub_tile_height, #1; \
5030 sub block, block, #16; \
5031 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5032 add texture_offset, texture_offset, #0xF00; \
5033 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
5036 #define setup_sprite_tile_half_8bpp_4x(edge) \
5037 setup_sprite_tile_add_blocks(single_4x); \
5038 add block, block, #16; \
5039 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5042 and texture_block_ptr, texture_offset, texture_mask; \
5045 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5046 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
5048 pld [ fb_ptr, #2048 ]; \
5049 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5050 draw_mask_fb_ptr_##edge##_b); \
5052 add texture_offset, texture_offset, #0x10; \
5053 add fb_ptr, fb_ptr, #2048 * 2; \
5055 subs sub_tile_height, sub_tile_height, #1; \
5058 sub block, block, #16; \
5059 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5060 add texture_offset, texture_offset, #0xF00; \
5061 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
5064 #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \
5065 add texture_offset, texture_offset_base, #8; \
5066 add fb_ptr, fb_ptr, #16 * 2 \
5068 #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \
5069 mov texture_offset, texture_offset_base \
5071 #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \
5072 setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \
5074 #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \
5075 mov texture_offset, texture_offset_base \
5077 #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \
5078 sub fb_ptr, fb_ptr, #16 * 2 \
5080 #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \
5082 #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \
5083 setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \
5085 #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \
5088 #define setup_sprite_setup_left_draw_mask_fb_ptr_4x() \
5089 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5090 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5091 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5092 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5094 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column_4x() \
5095 mov fb_ptr_advance_column, #32 * 2; \
5096 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5097 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5098 sub fb_ptr_advance_column, height, lsl #11 + 1; \
5099 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5100 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5102 #define setup_sprite_setup_right_draw_mask_fb_ptr_4x() \
5103 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[4]; \
5104 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[5]; \
5105 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[6]; \
5106 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[7] \
5114 // [ sp + 4 ]: width
5115 // [ sp + 8 ]: height
5116 // [ sp + 12 ]: color (unused)
5118 #define setup_sprite_tiled_builder(texture_mode, x4mode) \
5120 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \
5122 setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \
5124 setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \
5126 setup_sprite_tile_column_width_single(texture_mode, single, full, none, \
5128 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \
5130 setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \
5132 setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \
5134 setup_sprite_tile_column_width_single(texture_mode, single, half, right, \
5136 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \
5138 setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \
5140 setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \
5142 setup_sprite_tile_column_width_single(texture_mode, single, half, left, \
5144 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \
5146 setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \
5151 function(setup_sprite_##texture_mode##x4mode) \
5152 stmdb sp!, { r4 - r11, r14 }; \
5153 setup_sprite_tiled_initialize_##texture_mode##x4mode(); \
5155 ldr v, [ sp, #36 ]; \
5156 and offset_u, u, #0xF; \
5158 ldr width, [ sp, #40 ]; \
5159 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
5161 ldr height, [ sp, #44 ]; \
5162 add fb_ptr, fb_ptr, y, lsl #11; \
5164 add fb_ptr, fb_ptr, x, lsl #1; \
5165 and offset_v, v, #0xF; \
5167 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
5168 add width_rounded, offset_u, width; \
5170 add height_rounded, offset_v, height; \
5171 add width_rounded, width_rounded, #15; \
5173 add height_rounded, height_rounded, #15; \
5174 mov tile_width, width_rounded, lsr #4; \
5176 /* texture_offset_base = VH-VL-00-00 */\
5177 mov texture_offset_base, v, lsl #8; \
5178 and offset_u_right, width_rounded, #0xF; \
5180 /* texture_offset_base = VH-UH-UL-00 */\
5181 bfi texture_offset_base, u, #4, #8; \
5182 mov right_block_mask, #0xFFFFFFFE; \
5184 setup_sprite_offset_u_adjust##x4mode(); \
5186 /* texture_offset_base = VH-UH-VL-00 */\
5187 bfi texture_offset_base, v, #4, #4; \
5188 mov left_block_mask, #0xFFFFFFFF; \
5190 mov tile_height, height_rounded, lsr #4; \
5191 mvn left_block_mask, left_block_mask, lsl offset_u; \
5193 /* texture_mask = HH-HL-WH-WL */\
5194 ldrh texture_mask, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]; \
5195 mov right_block_mask, right_block_mask, lsl offset_u_right; \
5197 /* texture_mask_rev = WH-WL-HH-HL */\
5198 rev16 texture_mask_rev, texture_mask; \
5199 vmov block_masks, left_block_mask, right_block_mask; \
5201 /* texture_mask = HH-HL-HL-WL */\
5202 bfi texture_mask, texture_mask_rev, #4, #4; \
5203 /* texture_mask_rev = 00-00-00-WH */\
5204 mov texture_mask_rev, texture_mask_rev, lsr #12; \
5206 /* texture_mask = HH-WH-HL-WL */\
5207 bfi texture_mask, texture_mask_rev, #8, #4; \
5208 setup_sprite_get_left_block_mask##x4mode(); \
5210 mov control_mask, #0; \
5211 setup_sprite_compare_left_block_mask##x4mode(); \
5213 setup_sprite_get_right_block_mask##x4mode(); \
5214 orreq control_mask, control_mask, #0x4; \
5216 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
5217 setup_sprite_compare_right_block_mask##x4mode(); \
5219 orreq control_mask, control_mask, #0x8; \
5220 cmp tile_width, #1; \
5222 add block, psx_gpu, #psx_gpu_blocks_offset; \
5223 orreq control_mask, control_mask, #0x1; \
5225 cmp tile_height, #1; \
5226 add block, block, num_blocks, lsl #6; \
5228 orreq control_mask, control_mask, #0x2; \
5229 ldr pc, [ pc, control_mask, lsl #2 ]; \
5232 .word setup_sprite_##texture_mode##_multi_multi_full_full##x4mode; \
5233 .word setup_sprite_##texture_mode##_single_multi_full_none##x4mode; \
5234 .word setup_sprite_##texture_mode##_multi_single_full_full##x4mode; \
5235 .word setup_sprite_##texture_mode##_single_single_full_none##x4mode; \
5236 .word setup_sprite_##texture_mode##_multi_multi_half_full##x4mode; \
5237 .word setup_sprite_##texture_mode##_single_multi_half_right##x4mode; \
5238 .word setup_sprite_##texture_mode##_multi_single_half_full##x4mode; \
5239 .word setup_sprite_##texture_mode##_single_single_half_right##x4mode; \
5240 .word setup_sprite_##texture_mode##_multi_multi_full_half##x4mode; \
5241 .word setup_sprite_##texture_mode##_single_multi_half_left##x4mode; \
5242 .word setup_sprite_##texture_mode##_multi_single_full_half##x4mode; \
5243 .word setup_sprite_##texture_mode##_single_single_half_left##x4mode; \
5244 .word setup_sprite_##texture_mode##_multi_multi_half_half##x4mode; \
5246 .word setup_sprite_##texture_mode##_multi_single_half_half##x4mode; \
5249 setup_sprite_tiled_builder(4bpp,);
5250 setup_sprite_tiled_builder(8bpp,);
5252 #undef draw_mask_fb_ptr_left
5253 #undef draw_mask_fb_ptr_right
5255 setup_sprite_tiled_builder(4bpp, _4x);
5256 setup_sprite_tiled_builder(8bpp, _4x);
5264 #define block_ptr r0
5265 #define num_blocks r1
5267 #define texel_shift_mask r3
5268 #define block_pixels_a r4
5269 #define block_pixels_b r5
5278 #define texels_01 r6
5279 #define texels_23 r7
5280 #define texels_45 r8
5281 #define texels_67 r9
5283 function(texture_sprite_blocks_8bpp)
5284 stmdb sp!, { r4 - r11, r14 }
5285 movw texel_shift_mask, #(0xFF << 1)
5287 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5288 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
5290 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
5291 ldr block_pixels_a, [ block_ptr, #16 ]
5294 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
5295 ldr block_pixels_b, [ block_ptr, #20 ]
5297 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
5298 ldrh texel_0, [ clut_ptr, texel_0 ]
5300 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
5301 ldrh texel_1, [ clut_ptr, texel_1 ]
5303 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
5304 ldr block_pixels_a, [ block_ptr, #(64 + 16) ]
5306 ldrh texel_2, [ clut_ptr, texel_2 ]
5307 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
5309 ldrh texel_3, [ clut_ptr, texel_3 ]
5310 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
5312 ldrh texel_4, [ clut_ptr, texel_4 ]
5313 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
5315 ldrh texel_5, [ clut_ptr, texel_5 ]
5316 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
5318 ldrh texel_6, [ clut_ptr, texel_6 ]
5319 orr texels_01, texel_0, texel_1, lsl #16
5321 ldrh texel_7, [ clut_ptr, texel_7 ]
5322 orr texels_23, texel_2, texel_3, lsl #16
5324 orr texels_45, texel_4, texel_5, lsl #16
5325 str texels_01, [ block_ptr, #0 ]
5327 orr texels_67, texel_6, texel_7, lsl #16
5328 str texels_23, [ block_ptr, #4 ]
5330 subs num_blocks, num_blocks, #1
5331 str texels_45, [ block_ptr, #8 ]
5333 str texels_67, [ block_ptr, #12 ]
5334 add block_ptr, block_ptr, #64
5338 ldmia sp!, { r4 - r11, pc }
5341 #undef width_rounded
5344 #undef texture_offset
5347 #undef texels_wide_low
5348 #undef texels_wide_high
5359 #define left_offset r8
5360 #define width_rounded r9
5361 #define right_width r10
5363 #define block_width r11
5365 #define texture_offset_base r1
5366 #define texture_mask r2
5367 #define texture_page_ptr r3
5368 #define num_blocks r4
5371 #define texture_offset r8
5372 #define blocks_remaining r9
5374 #define fb_ptr_pitch r12
5375 #define texture_block_ptr r14
5377 #define texture_mask_width r2
5378 #define texture_mask_height r3
5379 #define left_mask_bits r4
5380 #define right_mask_bits r5
5384 #undef block_masks_shifted
5387 #define block_masks d0
5388 #define block_masks_shifted d1
5389 #define draw_mask_fb_ptr d2
5392 #define draw_mask_fb_ptr_a d2
5393 #define draw_mask_fb_ptr_b d3
5394 #define texels_low d4
5395 #define texels_high d5
5396 #define texels_wide_low d6
5397 #define texels_wide_high d7
5398 #define texels_wide q3
5401 setup_sprites_16bpp_flush:
5404 stmdb sp!, { r0 - r3, r12, r14 }
5405 bl flush_render_block_buffer
5406 ldmia sp!, { r0 - r3, r12, r14 }
5410 add block, psx_gpu, #psx_gpu_blocks_offset
5411 mov num_blocks, block_width
5415 function(setup_sprite_16bpp)
5416 stmdb sp!, { r4 - r11, r14 }
5417 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5420 add fb_ptr, fb_ptr, y, lsl #11
5422 ldr width, [ sp, #40 ]
5423 add fb_ptr, fb_ptr, x, lsl #1
5425 ldr height, [ sp, #44 ]
5426 and left_offset, u, #0x7
5428 add texture_offset_base, u, u
5429 add width_rounded, width, #7
5431 add texture_offset_base, v, lsl #11
5432 mov left_mask_bits, #0xFF
5434 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5435 add width_rounded, width_rounded, left_offset
5437 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5438 sub fb_ptr, fb_ptr, left_offset, lsl #1
5440 add texture_mask, texture_mask_width, texture_mask_width
5441 mov right_mask_bits, #0xFE
5443 and right_width, width_rounded, #0x7
5444 mvn left_mask_bits, left_mask_bits, lsl left_offset
5446 add texture_mask, texture_mask_height, lsl #11
5447 mov block_width, width_rounded, lsr #3
5449 mov right_mask_bits, right_mask_bits, lsl right_width
5450 movw fb_ptr_pitch, #(2048 + 16)
5452 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5453 vmov block_masks, left_mask_bits, right_mask_bits
5455 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5456 add block, psx_gpu, #psx_gpu_blocks_offset
5458 bic texture_offset_base, texture_offset_base, #0xF
5461 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5462 add block, block, num_blocks, lsl #6
5466 vext.32 block_masks_shifted, block_masks, block_masks, #1
5467 vorr.u32 block_masks, block_masks, block_masks_shifted
5468 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5471 add num_blocks, num_blocks, #1
5472 cmp num_blocks, #MAX_BLOCKS
5473 blgt setup_sprites_16bpp_flush
5475 and texture_block_ptr, texture_offset_base, texture_mask
5476 subs height, height, #1
5478 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5479 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5481 vst1.u32 { texels }, [ block, :128 ]
5482 add block, block, #40
5484 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5487 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5489 add block, block, #24
5490 add texture_offset_base, texture_offset_base, #2048
5491 add fb_ptr, fb_ptr, #2048
5492 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5495 ldmia sp!, { r4 - r11, pc }
5498 add num_blocks, num_blocks, block_width
5499 mov texture_offset, texture_offset_base
5501 cmp num_blocks, #MAX_BLOCKS
5502 blgt setup_sprites_16bpp_flush
5504 add texture_offset_base, texture_offset_base, #2048
5505 and texture_block_ptr, texture_offset, texture_mask
5507 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5508 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5510 vst1.u32 { texels }, [ block, :128 ]
5511 add block, block, #40
5513 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5514 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5517 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5518 subs blocks_remaining, block_width, #2
5520 add texture_offset, texture_offset, #16
5521 add fb_ptr, fb_ptr, #16
5523 vmov.u8 draw_mask_fb_ptr, #0
5525 add block, block, #24
5529 and texture_block_ptr, texture_offset, texture_mask
5530 subs blocks_remaining, blocks_remaining, #1
5532 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5533 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5535 vst1.u32 { texels }, [ block, :128 ]
5536 add block, block, #40
5538 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5541 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5543 add texture_offset, texture_offset, #16
5544 add fb_ptr, fb_ptr, #16
5546 add block, block, #24
5550 and texture_block_ptr, texture_offset, texture_mask
5551 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5553 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5554 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5556 vst1.u32 { texels }, [ block, :128 ]
5557 add block, block, #40
5559 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5560 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5562 add block, block, #24
5563 subs height, height, #1
5565 add fb_ptr, fb_ptr, fb_ptr_pitch
5566 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5570 ldmia sp!, { r4 - r11, pc }
5574 // FIXME: duplicate code with normal version :(
5575 #undef draw_mask_fb_ptr
5577 function(setup_sprite_16bpp_4x)
5578 stmdb sp!, { r4 - r11, r14 }
5579 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5582 add fb_ptr, fb_ptr, y, lsl #11
5584 ldr width, [ sp, #40 ]
5585 add fb_ptr, fb_ptr, x, lsl #1
5587 ldr height, [ sp, #44 ]
5588 and left_offset, u, #0x7
5590 add texture_offset_base, u, u
5591 add width_rounded, width, #7
5593 add texture_offset_base, v, lsl #11
5594 movw left_mask_bits, #0xFFFF
5596 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5597 add width_rounded, width_rounded, left_offset
5601 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5602 sub fb_ptr, fb_ptr, left_offset, lsl #1
5604 add texture_mask, texture_mask_width, texture_mask_width
5605 movw right_mask_bits, #0xFFFC
5607 and right_width, width_rounded, #0x7
5608 mvn left_mask_bits, left_mask_bits, lsl left_offset
5612 add texture_mask, texture_mask_height, lsl #11
5613 mov block_width, width_rounded, lsr #3
5615 mov right_mask_bits, right_mask_bits, lsl right_width
5616 movw fb_ptr_pitch, #(2048 + 16) * 2
5618 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4+1
5619 vmov block_masks, left_mask_bits, right_mask_bits
5621 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5622 add block, psx_gpu, #psx_gpu_blocks_offset
5624 bic texture_offset_base, texture_offset_base, #0xF
5627 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5628 add block, block, num_blocks, lsl #6
5633 vext.32 block_masks_shifted, block_masks, block_masks, #1
5634 vorr.u32 block_masks, block_masks, block_masks_shifted
5635 vdup.u8 draw_mask_fb_ptr_a, block_masks[0]
5636 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5639 add num_blocks, num_blocks, block_width
5640 cmp num_blocks, #MAX_BLOCKS
5641 blgt setup_sprites_16bpp_flush
5643 and texture_block_ptr, texture_offset_base, texture_mask
5644 subs height, height, #1
5646 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5647 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5649 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5651 add texture_offset_base, texture_offset_base, #2048
5652 add fb_ptr, fb_ptr, #2048*2
5653 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5656 ldmia sp!, { r4 - r11, pc }
5659 add num_blocks, num_blocks, block_width
5660 mov texture_offset, texture_offset_base
5662 vdup.u8 draw_mask_fb_ptr_a, block_masks[0] // left_mask_bits
5663 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5665 cmp num_blocks, #MAX_BLOCKS
5666 blgt setup_sprites_16bpp_flush
5668 add texture_offset_base, texture_offset_base, #2048
5669 and texture_block_ptr, texture_offset, texture_mask
5671 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5672 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5674 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5676 subs blocks_remaining, block_width, #2*4
5677 add texture_offset, texture_offset, #16
5679 vmov.u8 draw_mask_fb_ptr_a, #0
5680 vmov.u8 draw_mask_fb_ptr_b, #0
5682 add fb_ptr, fb_ptr, #16*2
5686 and texture_block_ptr, texture_offset, texture_mask
5687 subs blocks_remaining, blocks_remaining, #4
5689 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5690 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5692 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5693 add texture_offset, texture_offset, #16
5695 add fb_ptr, fb_ptr, #16*2
5699 vdup.u8 draw_mask_fb_ptr_a, block_masks[4] // right_mask_bits
5700 vdup.u8 draw_mask_fb_ptr_b, block_masks[5]
5702 and texture_block_ptr, texture_offset, texture_mask
5703 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5705 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5707 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5708 subs height, height, #1
5710 add fb_ptr, fb_ptr, fb_ptr_pitch
5711 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5715 ldmia sp!, { r4 - r11, pc }
5720 #undef right_mask_bits
5723 #undef blocks_remaining
5733 #define right_width r5
5734 #define right_mask_bits r6
5738 #define fb_ptr_pitch r12
5740 // referenced by setup_sprites_16bpp_flush
5741 #define num_blocks r4
5743 #define block_width r11
5748 #define blocks_remaining r6
5751 #define right_mask q1
5752 #define test_mask q2
5753 #define draw_mask q2
5754 #define draw_mask_bits_fb_ptr d6
5759 function(setup_sprite_untextured)
5760 ldrh r12, [ psx_gpu, #psx_gpu_render_state_offset ]
5761 tst r12, #(RENDER_STATE_MASK_EVALUATE | RENDER_FLAGS_MODULATE_TEXELS \
5762 | RENDER_FLAGS_BLEND)
5763 ldreqb r12, [ psx_gpu, #psx_gpu_render_mode_offset ]
5764 tsteq r12, #RENDER_INTERLACE_ENABLED
5765 beq setup_sprite_untextured_simple
5767 stmdb sp!, { r4 - r11, r14 }
5769 ldr width, [ sp, #40 ]
5770 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5772 ldr height, [ sp, #44 ]
5773 add fb_ptr, fb_ptr, y, lsl #11
5775 add fb_ptr, fb_ptr, x, lsl #1
5776 sub right_width, width, #1
5778 ldr color, [ sp, #48 ]
5781 add block_width, width, #7
5785 mov right_mask_bits, #0xff
5787 sub fb_ptr_pitch, block_width, #1
5788 lsl right_mask_bits, right_width
5790 lsl fb_ptr_pitch, #3+1
5791 ubfx color_r, color, #3, #5
5793 rsb fb_ptr_pitch, #1024*2
5794 ubfx color_g, color, #11, #5
5796 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
5797 ubfx color_b, color, #19, #5
5799 vdup.u16 right_mask, right_mask_bits
5800 orr color, color_r, color_b, lsl #10
5802 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5803 orr color, color, color_g, lsl #5
5805 vtst.u16 right_mask, right_mask, test_mask
5806 add block, psx_gpu, #psx_gpu_blocks_offset
5808 vdup.u16 colors, color
5809 add block, block, num_blocks, lsl #6
5812 setup_sprite_untextured_height_loop:
5813 add num_blocks, block_width
5814 sub blocks_remaining, block_width, #1
5816 cmp num_blocks, #MAX_BLOCKS
5817 blgt setup_sprites_16bpp_flush
5819 cmp blocks_remaining, #0
5822 vmov.u8 draw_mask, #0 /* zero_mask */
5823 vmov.u8 draw_mask_bits_fb_ptr, #0
5826 vst1.u32 { draw_mask }, [ block, :128 ]!
5827 subs blocks_remaining, #1
5829 vst1.u32 { colors }, [ block, :128 ]
5830 add block, block, #24
5832 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5833 vst1.u32 { draw_mask_bits_fb_ptr }, [ block, :64 ]
5835 add block, block, #24
5840 vst1.u32 { right_mask }, [ block, :128 ]!
5843 vst1.u32 { colors }, [ block, :128 ]
5844 add block, block, #24
5846 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5847 vst1.u32 { draw_mask_bits_fb_ptr }, [ block, :64 ]
5849 add block, block, #24
5850 add fb_ptr, fb_ptr_pitch
5852 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5853 bgt setup_sprite_untextured_height_loop
5855 ldmia sp!, { r4 - r11, pc }
5859 #undef texture_page_ptr
5861 #undef dirty_textures_mask
5862 #undef current_texture_mask
5865 #define current_texture_page r1
5866 #define texture_page_ptr r2
5867 #define vram_ptr_a r3
5868 #define current_texture_page_x r12
5869 #define current_texture_page_y r4
5870 #define dirty_textures_mask r5
5874 #define current_texture_mask r9
5876 #define vram_ptr_b r11
5878 #define texel_block_a d0
5879 #define texel_block_b d1
5880 #define texel_block_expanded_a q1
5881 #define texel_block_expanded_b q2
5882 #define texel_block_expanded_ab q2
5883 #define texel_block_expanded_c q3
5884 #define texel_block_expanded_d q4
5885 #define texel_block_expanded_cd q3
5887 function(update_texture_4bpp_cache)
5888 stmdb sp!, { r4 - r11, r14 }
5891 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5893 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5894 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5896 and current_texture_page_x, current_texture_page, #0xF
5897 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
5899 mov current_texture_page_y, current_texture_page, lsr #4
5900 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5902 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5905 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5906 bic dirty_textures_mask, current_texture_mask
5909 str dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5914 add vram_ptr_b, vram_ptr_a, #2048
5917 vld1.u32 { texel_block_a }, [ vram_ptr_a, :64 ], c_4096
5918 vld1.u32 { texel_block_b }, [ vram_ptr_b, :64 ], c_4096
5920 vmovl.u8 texel_block_expanded_a, texel_block_a
5921 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5922 vmovl.u8 texel_block_expanded_c, texel_block_b
5923 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5925 vbic.u16 texel_block_expanded_a, #0x00F0
5926 vbic.u16 texel_block_expanded_b, #0x00F0
5927 vbic.u16 texel_block_expanded_c, #0x00F0
5928 vbic.u16 texel_block_expanded_d, #0x00F0
5930 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
5931 texel_block_expanded_b
5932 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
5933 texel_block_expanded_d
5935 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
5936 [ texture_page_ptr, :256 ]!
5938 subs sub_y, sub_y, #1
5942 add vram_ptr_a, vram_ptr_a, #8
5943 add vram_ptr_b, vram_ptr_b, #8
5945 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5946 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5948 subs tile_x, tile_x, #1
5952 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5953 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5955 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5956 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5958 subs tile_y, tile_y, #1
5962 ldmia sp!, { r4 - r11, pc }
5965 #undef current_texture_page
5968 #define texture_page r1
5969 #define texture_page_ptr r2
5970 #define vram_ptr_a r3
5971 #define texture_page_x r12
5972 #define texture_page_y r4
5973 #define current_texture_page r5
5978 #define vram_ptr_b r11
5990 function(update_texture_8bpp_cache_slice)
5991 stmdb sp!, { r4 - r11, r14 }
5994 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5995 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5997 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
6000 and texture_page_x, texture_page, #0xF
6001 mov texture_page_y, texture_page, lsr #4
6003 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
6006 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
6007 eor current_texture_page, current_texture_page, texture_page
6009 ands current_texture_page, current_texture_page, #0x1
6012 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6015 add vram_ptr_b, vram_ptr_a, #2048
6018 vld1.u32 { texels_a }, [ vram_ptr_a, :128 ], c_4096
6019 vld1.u32 { texels_b }, [ vram_ptr_b, :128 ], c_4096
6020 vld1.u32 { texels_c }, [ vram_ptr_a, :128 ], c_4096
6021 vld1.u32 { texels_d }, [ vram_ptr_b, :128 ], c_4096
6023 vst1.u32 { texels_a, texels_b }, [ texture_page_ptr, :256 ]!
6024 vst1.u32 { texels_c, texels_d }, [ texture_page_ptr, :256 ]!
6026 subs sub_y, sub_y, #1
6031 add vram_ptr_a, vram_ptr_a, #16
6032 add vram_ptr_b, vram_ptr_b, #16
6034 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
6035 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
6037 subs tile_x, tile_x, #1
6042 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
6043 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
6045 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
6046 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
6048 subs tile_y, tile_y, #1
6049 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6054 ldmia sp!, { r4 - r11, pc }
6057 /* void scale2x_tiles8(void *dst, const void *src, int w8, int h) */
6058 function(scale2x_tiles8)
6062 add r12, r0, #1024*2
6066 vld1.u16 { q0 }, [ r1, :128 ]!
6067 vld1.u16 { q2 }, [ r1, :128 ]!
6073 vst1.u16 { q0, q1 }, [ r0, :128 ]!
6074 vst1.u16 { q0, q1 }, [ r12, :128 ]!
6076 vst1.u16 { q2, q3 }, [ r0, :128 ]!
6077 vst1.u16 { q2, q3 }, [ r12, :128 ]!
6084 sub r0, r2, lsl #4+1
6086 add r12, r0, #1024*2
6092 // vim:filetype=armasm