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
24 #include "psx_gpu_offsets.h"
26 #define psx_gpu_b_dx_offset (psx_gpu_b_block_span_offset + 4)
28 #define edge_data_left_x_offset 0
29 #define edge_data_num_blocks_offset 2
30 #define edge_data_right_mask_offset 4
31 #define edge_data_y_offset 6
75 #define gw_by_l gw_bx_l
100 #define uvrg_xxxx1 q4
105 #define uvrg_xxxx2 q5
110 #define yyyy_uvrg0 q6
115 #define yyyy_uvrg1 q7
120 #define yyyy_uvrg2 q8
147 #define uvrg_base q11
190 #define uvrgb_phase q13
194 /* FIXME: users of this should be in psx_gpu instead */
196 #define load_pointer(register, pointer) \
197 movw register, :lower16:pointer; \
198 movt register, :upper16:pointer; \
201 #define load_pointer(register, pointer) \
202 ldr register, =pointer \
206 #define function(name) \
215 function(compute_all_gradients)
216 // First compute the triangle area reciprocal and shift. The division will
217 // happen concurrently with much of the work which follows.
218 @ r12 = psx_gpu->triangle_area
219 ldr r12, [ psx_gpu, #psx_gpu_triangle_area_offset ]
220 stmdb sp!, { r4 - r11, lr }
222 @ load exponent of 62 into upper half of double
224 clz r14, r12 @ r14 = shift
226 movt r4, #((62 + 1023) << 4)
227 mov r12, r12, lsl r14 @ r12 = triangle_area_normalized
229 @ load area normalized into lower half of double
231 vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n
233 movt r4, #((1022 + 31) << 4)
236 add r4, r4, r12, lsr #11
239 vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n
241 // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) =
242 // ( d0 * d1 ) - ( d2 * d3 ) =
243 // ( m0 ) - ( m1 ) = gradient
245 // This is split to do 12 elements at a time over three sets: a, b, and c.
246 // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so
247 // two of the slots are unused.
249 // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as
252 // First type is: uvrg bxxx xxxx
253 // Second type is: yyyy ybyy uvrg
254 // Since x_a and y_c are the same the same variable is used for both.
256 vld1.u32 { v0 }, [ v_a, : 128 ] @ v0 = { uvrg0, b0, x0, y0 }
257 ldrsh x0, [ v_a, #8 ] @ load x0
259 vld1.u32 { v1 }, [ v_b, : 128 ] @ v1 = { uvrg1, b1, x1, y1}
260 ldrh x1, [ v_b, #8 ] @ load x1
262 vld1.u32 { v2 }, [ v_c, : 128 ] @ v2 = { uvrg2, b2, x2, y2 }
263 ldrh x2, [ v_c, #8 ] @ load x2
265 vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- }
266 ldrh y0, [ v_a, #10 ] @ load y0
268 vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- }
269 ldrh y1, [ v_b, #10 ] @ load y1
271 vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- }
272 ldrh y2, [ v_c, #10 ] @ load y2
274 vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 }
275 vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 }
277 orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 }
278 pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 }
280 vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 }
281 vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 }
283 vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 }
284 vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 }
286 ldrb b2, [ v_c, #4 ] @ load b2
287 orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 }
289 ldrb b1, [ v_b, #4 ] @ load b1
290 orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 }
292 vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 }
293 vsub.s16 d0_ab, x1_ab, x0_ab
295 ldrb b0, [ v_a, #4 ] @ load b0
296 orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 }
298 vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 }
299 vsub.s16 d2_ab, x2_ab, x1_ab
301 vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 }
302 vsub.s16 d1_ab, y2_ab, y1_ab
304 orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 }
305 ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 }
307 ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 }
308 ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 }
310 vsub.s16 d3_ab, y1_ab, y0_ab
311 smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) -
312 @ ((x2 - X1) * (b1 - b0))
313 vmull.s16 ga_uvrg_x, d0_a, d1_a
314 smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) -
315 @ ((b2 - b1) * (y1 - y0))
316 vmlsl.s16 ga_uvrg_x, d2_a, d3_a
317 movs gs_bx, ga_bx, asr #31
319 vmull.s16 ga_uvrg_y, d0_b, d1_b
320 rsbmi ga_bx, ga_bx, #0
322 @ r12 = psx_gpu->uvrgb_phase
323 ldr r12, [ psx_gpu, #psx_gpu_uvrgb_phase_offset ]
325 vmlsl.s16 ga_uvrg_y, d2_b, d3_b
326 movs gs_by, ga_by, asr #31
328 vshr.u64 d0, d30, #22
329 add b_base, r12, b0, lsl #16
331 vdup.u32 uvrgb_phase, r12
333 rsbmi ga_by, ga_by, #0
334 vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0
336 @ r12 = psx_gpu->triangle_winding_offset
337 ldrb r12, [ psx_gpu, #psx_gpu_triangle_winding_offset ]
338 vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0
340 rsb r12, r12, #0 @ r12 = -(triangle->winding)
342 vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w }
343 sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS)
345 vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16
346 vdup.u32 r_shift, r14 @ r_shift = { shift, shift, shift, shift }
348 vadd.u32 uvrg_base, uvrgb_phase
349 vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x)
351 vmov area_r_s, s0 @ area_r_s = triangle_reciprocal
352 vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y)
354 vmull.u32 gw_rg_x, ga_rg_x, d0[0]
355 vmull.u32 gw_uv_x, ga_uv_x, d0[0]
356 vmull.u32 gw_rg_y, ga_rg_y, d0[0]
357 vmull.u32 gw_uv_y, ga_uv_y, d0[0]
359 vshl.u64 gw_rg_x, gw_rg_x, r_shift
360 vshl.u64 gw_uv_x, gw_uv_x, r_shift
361 vshl.u64 gw_rg_y, gw_rg_y, r_shift
362 vshl.u64 gw_uv_y, gw_uv_y, r_shift
364 veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask
365 vmovn.u64 g_uv_x, gw_uv_x
367 veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask
368 vmovn.u64 g_rg_x, gw_rg_x
370 veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
371 vmovn.u64 g_uv_y, gw_uv_y
373 vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
374 vmovn.u64 g_rg_y, gw_rg_y
376 veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
377 mov ga_bx, ga_bx, lsl #13
379 vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
380 mov ga_by, ga_by, lsl #13
383 umull gw_bx_l, gw_bx_h, ga_bx, area_r_s
385 vshl.u32 g_uvrg_x, g_uvrg_x, #4
386 vshl.u32 g_uvrg_y, g_uvrg_y, #4
388 umull gw_by_l, gw_by_h, ga_by, area_r_s
389 vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0]
391 eor gs_bx, gs_bx, r12
392 vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1
394 veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0
395 eor gs_by, gs_by, r12
397 rsb r11, r14, #0 @ r11 = negative shift for scalar lsr
398 add store_a, psx_gpu, #psx_gpu_uvrg_offset
400 sub r11, r11, #(32 - 13)
402 add store_b, store_a, #16
405 vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1
406 vst1.u32 { uvrg_base }, [ store_a, : 128 ], store_inc
408 vst1.u32 { uvrg_dx1 }, [ store_b, : 128 ], store_inc
409 mov g_bx, gw_bx_h, lsr r11
411 vst1.u32 { g_uvrg_y }, [ store_a, : 128 ], store_inc
412 mov g_by, gw_by_h, lsr r11
414 vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \
415 [ store_b, : 128 ], store_inc
416 eor g_bx, g_bx, gs_bx
418 vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \
419 [ store_b, : 128 ], store_inc
420 sub g_bx, g_bx, gs_bx
423 eor g_by, g_by, gs_by
425 mls b_base, g_bx, x0, b_base
426 sub g_by, g_by, gs_by
431 add g_bx2, g_bx, g_bx
432 add g_bx3, g_bx, g_bx2
434 stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by }
436 ldmia sp!, { r4 - r11, pc }
453 #define height_minor_a r7
454 #define height_minor_b r8
455 #define height_major r9
458 #define reciprocal_table_ptr r10
460 #define edge_alt_low r4
461 #define edge_alt_high r5
462 #define edge_dx_dy_alt r6
463 #define edge_shift_alt r10
465 #define edge_dx_dy_alt_low r4
466 #define edge_dx_dy_alt_high r5
468 #define span_edge_data r4
469 #define span_uvrg_offset r5
470 #define span_b_offset r6
478 #define alternate_x q0
479 #define alternate_dx_dy q1
480 #define alternate_x_32 q2
482 #define alternate_x_low d0
483 #define alternate_x_high d1
484 #define alternate_dx_dy_low d2
485 #define alternate_dx_dy_high d3
486 #define alternate_x_32_low d4
487 #define alternate_x_32_high d5
491 #define left_dx_dy q5
492 #define right_dx_dy q6
494 #define right_edge q8
496 #define left_x_low d6
497 #define left_x_high d7
498 #define right_x_low d8
499 #define right_x_high d9
500 #define left_dx_dy_low d10
501 #define left_dx_dy_high d11
502 #define right_dx_dy_low d12
503 #define right_dx_dy_high d13
504 #define left_edge_low d14
505 #define left_edge_high d15
506 #define right_edge_low d16
507 #define right_edge_high d17
509 #define y_mid_point d18
512 #define left_right_x_16 q11
513 #define span_shifts_y q12
516 #define span_shifts d24
521 #define left_right_x_16_low d22
522 #define left_right_x_16_high d23
527 #define alternate_x_16 d4
530 #define v_clip_low d6
532 #define right_x_32 q10
533 #define left_x_32 q11
534 #define alternate_select d24
536 #define right_x_32_low d20
537 #define right_x_32_high d21
538 #define left_x_32_low d22
539 #define left_x_32_high d23
542 #define edges_dx_dy d2
543 #define edge_shifts d3
544 #define edge_shifts_64 q2
546 #define edges_xy_left d0
547 #define edges_xy_right d1
549 #define height_reciprocals d6
557 #define heights_b d12
558 #define edges_dx_dy_64 q10
560 #define edges_dx_dy_64_left d20
561 #define edges_dx_dy_64_right d21
564 #define setup_spans_prologue() \
565 stmdb sp!, { r4 - r11, lr }; \
567 ldrsh x_a, [ v_a, #8 ]; \
568 ldrsh x_b, [ v_b, #8 ]; \
569 ldrsh x_c, [ v_c, #8 ]; \
570 ldrsh y_a, [ v_a, #10 ]; \
571 ldrsh y_b, [ v_b, #10 ]; \
572 ldrsh y_c, [ v_c, #10 ]; \
574 add temp, psx_gpu, #psx_gpu_uvrg_offset; \
575 vld1.32 { uvrg }, [ temp ]; \
576 add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \
577 vld1.32 { uvrg_dy }, [ temp ]; \
578 load_pointer(reciprocal_table_ptr, reciprocal_table); \
580 vmov.u32 c_0x01, #0x01 \
582 #define setup_spans_load_b() \
583 ldr b, [ psx_gpu, #psx_gpu_b_offset ]; \
584 ldr b_dy, [ psx_gpu, #psx_gpu_b_dy_offset ] \
586 #define setup_spans_prologue_b() \
587 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
588 add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \
590 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
591 vmov.u16 c_0x0004, #0x0004; \
593 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
594 vmov.u16 c_0x0001, #0x0001; \
596 vld1.u16 { left_edge_low[], left_edge_high[] }, [ temp ]; \
597 add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \
599 vld1.u16 { right_edge_low[], right_edge_high[] }, [ temp ]; \
600 vadd.u16 right_edge, right_edge, c_0x0001; \
602 vmov.u16 c_0x0007, #0x0007; \
603 vmvn.u16 c_0xFFFE, #0x0001 \
606 #define compute_edge_delta_x2() \
607 ldr temp, [ reciprocal_table_ptr, height, lsl #2 ]; \
609 vdup.u32 heights, height; \
610 vsub.u32 widths, x_ends, x_starts; \
612 vdup.u32 edge_shifts, temp; \
613 vsub.u32 heights_b, heights, c_0x01; \
614 vshr.u32 height_reciprocals, edge_shifts, #10; \
616 vmla.s32 heights_b, x_starts, heights; \
617 vbic.u16 edge_shifts, #0xE0; \
618 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
619 vmull.s32 edges_xy, heights_b, height_reciprocals \
622 #define height_reciprocal_alt r11
623 #define height_b_alt r12
625 #define compute_edge_delta_x3(start_c, height_a, height_b) \
626 vmov.u32 heights, height_a, height_b; \
627 ldr temp, [ reciprocal_table_ptr, height_a, lsl #2 ]; \
628 vmov.u32 edge_shifts[0], temp; \
629 ldr temp, [ reciprocal_table_ptr, height_b, lsl #2 ]; \
630 vmov.u32 edge_shifts[1], temp; \
631 ldr edge_shift_alt, [ reciprocal_table_ptr, height_minor_b, lsl #2 ]; \
633 vsub.u32 widths, x_ends, x_starts; \
634 sub width_alt, x_c, start_c; \
636 vsub.u32 heights_b, heights, c_0x01; \
637 sub height_b_alt, height_minor_b, #1; \
639 vshr.u32 height_reciprocals, edge_shifts, #10; \
640 lsr height_reciprocal_alt, edge_shift_alt, #10; \
642 vmla.s32 heights_b, x_starts, heights; \
643 mla height_b_alt, height_minor_b, start_c, height_b_alt; \
645 vbic.u16 edge_shifts, #0xE0; \
646 and edge_shift_alt, edge_shift_alt, #0x1F; \
648 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
649 mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \
651 vmull.s32 edges_xy, heights_b, height_reciprocals; \
652 smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \
655 #define setup_spans_adjust_y_up() \
656 vsub.u32 y_x4, y_x4, c_0x0004 \
658 #define setup_spans_adjust_y_down() \
659 vadd.u32 y_x4, y_x4, c_0x0004 \
661 #define setup_spans_adjust_interpolants_up() \
662 vsub.u32 uvrg, uvrg, uvrg_dy; \
665 #define setup_spans_adjust_interpolants_down() \
666 vadd.u32 uvrg, uvrg, uvrg_dy; \
670 #define setup_spans_clip_interpolants_increment() \
671 mla b, b_dy, clip, b; \
672 vmla.s32 uvrg, uvrg_dy, v_clip \
674 #define setup_spans_clip_interpolants_decrement() \
675 mls b, b_dy, clip, b; \
676 vmls.s32 uvrg, uvrg_dy, v_clip \
678 #define setup_spans_clip_alternate_yes() \
679 smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \
681 #define setup_spans_clip_alternate_no() \
683 #define setup_spans_clip(direction, alternate_active) \
684 vdup.u32 v_clip, clip; \
685 setup_spans_clip_alternate_##alternate_active(); \
686 setup_spans_clip_interpolants_##direction(); \
687 vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \
690 #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \
691 vmovl.s32 edge_shifts_64, edge_shifts; \
692 vmovl.s32 edges_dx_dy_64, edges_dx_dy; \
694 vshl.s64 edges_xy, edges_xy, edge_shifts_64; \
695 vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \
697 vmov left_x_low, edges_xy_##left_index; \
698 vmov right_x_low, edges_xy_##right_index; \
700 vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \
701 vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \
702 vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \
703 vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \
705 vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \
706 vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \
708 vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \
709 vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \
712 #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \
713 setup_spans_adjust_edges_alternate_no(left_index, right_index); \
715 vdup.u16 y_mid_point, y_b; \
716 rsb temp, edge_shift_alt, #32; \
718 lsl edge_alt_high, edge_alt_high, edge_shift_alt; \
719 orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \
720 lsl edge_alt_low, edge_alt_low, edge_shift_alt; \
721 vmov alternate_x_low, edge_alt_low, edge_alt_high; \
723 asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \
724 lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \
725 vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \
726 vmov alternate_dx_dy_high, alternate_dx_dy_low; \
728 vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \
729 vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \
732 #define setup_spans_y_select_up() \
733 vclt.s16 alternate_select, y_x4, y_mid_point \
735 #define setup_spans_y_select_down() \
736 vcgt.s16 alternate_select, y_x4, y_mid_point \
739 #define setup_spans_alternate_select_left() \
740 vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \
742 #define setup_spans_alternate_select_right() \
743 vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \
746 #define setup_spans_set_x4_alternate_yes(alternate, direction) \
747 vshrn.s64 alternate_x_32_low, alternate_x, #32; \
748 vshrn.s64 left_x_32_low, left_x, #32; \
749 vshrn.s64 right_x_32_low, right_x, #32; \
751 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
752 vadd.u64 left_x, left_x, left_dx_dy; \
753 vadd.u64 right_x, right_x, right_dx_dy; \
755 vshrn.s64 alternate_x_32_high, alternate_x, #32; \
756 vshrn.s64 left_x_32_high, left_x, #32; \
757 vshrn.s64 right_x_32_high, right_x, #32; \
759 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
760 vadd.u64 left_x, left_x, left_dx_dy; \
761 vadd.u64 right_x, right_x, right_dx_dy; \
763 vmovn.u32 alternate_x_16, alternate_x_32; \
764 setup_spans_y_select_##direction(); \
765 vmovn.u32 left_right_x_16_low, left_x_32; \
767 vmovn.u32 left_right_x_16_high, right_x_32; \
768 setup_spans_alternate_select_##alternate(); \
770 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
771 str b, [ span_b_offset ], #4; \
772 setup_spans_adjust_interpolants_##direction(); \
774 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
776 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
777 str b, [ span_b_offset ], #4; \
778 setup_spans_adjust_interpolants_##direction(); \
780 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
782 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
783 str b, [ span_b_offset ], #4; \
784 setup_spans_adjust_interpolants_##direction(); \
786 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
787 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
788 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
790 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
791 str b, [ span_b_offset ], #4; \
792 setup_spans_adjust_interpolants_##direction(); \
794 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
795 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
797 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
799 setup_spans_adjust_y_##direction() \
802 #define setup_spans_set_x4_alternate_no(alternate, direction) \
803 vshrn.s64 left_x_32_low, left_x, #32; \
804 vshrn.s64 right_x_32_low, right_x, #32; \
806 vadd.u64 left_x, left_x, left_dx_dy; \
807 vadd.u64 right_x, right_x, right_dx_dy; \
809 vshrn.s64 left_x_32_high, left_x, #32; \
810 vshrn.s64 right_x_32_high, right_x, #32; \
812 vadd.u64 left_x, left_x, left_dx_dy; \
813 vadd.u64 right_x, right_x, right_dx_dy; \
815 vmovn.u32 left_right_x_16_low, left_x_32; \
816 vmovn.u32 left_right_x_16_high, right_x_32; \
818 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
819 str b, [ span_b_offset ], #4; \
820 setup_spans_adjust_interpolants_##direction(); \
822 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
824 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
825 str b, [ span_b_offset ], #4; \
826 setup_spans_adjust_interpolants_##direction(); \
828 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
830 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
831 str b, [ span_b_offset ], #4; \
832 setup_spans_adjust_interpolants_##direction(); \
834 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
835 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
836 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
838 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
839 str b, [ span_b_offset ], #4; \
840 setup_spans_adjust_interpolants_##direction(); \
842 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
843 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
845 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
847 setup_spans_adjust_y_##direction() \
850 #define edge_adjust_low r11
851 #define edge_adjust_high r12
853 #define setup_spans_alternate_adjust_yes() \
854 smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \
855 subs edge_alt_low, edge_alt_low, edge_adjust_low; \
856 sbc edge_alt_high, edge_alt_high, edge_adjust_high \
858 #define setup_spans_alternate_adjust_no() \
861 #define setup_spans_down(left_index, right_index, alternate, alternate_active) \
862 setup_spans_alternate_adjust_##alternate_active(); \
863 setup_spans_load_b(); \
865 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
866 subs y_c, y_c, temp; \
867 subgt height, height, y_c; \
868 addgt height, height, #1; \
870 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
871 subs clip, temp, y_a; \
874 sub height, height, clip; \
875 add y_a, y_a, clip; \
876 setup_spans_clip(increment, alternate_active); \
882 orr temp, y_a, y_a, lsl #16; \
883 add temp, temp, #(1 << 16); \
885 add y_a, y_a, #(2 << 16); \
886 vmov.u32 y_x4, temp, y_a; \
888 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
890 setup_spans_prologue_b(); \
892 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
895 setup_spans_set_x4_alternate_##alternate_active(alternate, down); \
896 subs height, height, #4; \
902 #define setup_spans_alternate_pre_increment_yes() \
903 adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \
904 adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \
906 #define setup_spans_alternate_pre_increment_no() \
909 #define setup_spans_up_decrement_yes() \
910 suble height, height, #1 \
912 #define setup_spans_up_decrement_no() \
915 #define setup_spans_up(left_index, right_index, alternate, alternate_active) \
916 setup_spans_alternate_adjust_##alternate_active(); \
917 setup_spans_load_b(); \
920 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
921 subs temp, temp, y_c; \
922 subgt height, height, temp; \
923 setup_spans_up_decrement_##alternate_active(); \
925 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
926 subs clip, y_a, temp; \
929 sub height, height, clip; \
930 sub y_a, y_a, clip; \
931 setup_spans_clip(decrement, alternate_active); \
937 orr temp, y_a, y_a, lsl #16; \
938 sub temp, temp, #(1 << 16); \
940 sub y_a, y_a, #(2 << 16); \
941 vmov.u32 y_x4, temp, y_a; \
943 vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \
945 setup_spans_alternate_pre_increment_##alternate_active(); \
946 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
948 setup_spans_adjust_interpolants_up(); \
949 setup_spans_prologue_b(); \
951 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
954 setup_spans_set_x4_alternate_##alternate_active(alternate, up); \
955 subs height, height, #4; \
961 #define setup_spans_epilogue() \
962 ldmia sp!, { r4 - r11, pc } \
965 #define setup_spans_up_up(minor, major) \
966 setup_spans_prologue(); \
967 sub height_minor_a, y_a, y_b; \
968 sub height_minor_b, y_b, y_c; \
969 sub height, y_a, y_c; \
971 vdup.u32 x_starts, x_a; \
972 vmov.u32 x_ends, x_c, x_b; \
974 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
975 setup_spans_up(major, minor, minor, yes); \
976 setup_spans_epilogue() \
978 function(setup_spans_up_left)
979 setup_spans_up_up(left, right)
981 function(setup_spans_up_right)
982 setup_spans_up_up(right, left)
986 #define setup_spans_down_down(minor, major) \
987 setup_spans_prologue(); \
988 sub height_minor_a, y_b, y_a; \
989 sub height_minor_b, y_c, y_b; \
990 sub height, y_c, y_a; \
992 vdup.u32 x_starts, x_a; \
993 vmov.u32 x_ends, x_c, x_b; \
995 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
996 setup_spans_down(major, minor, minor, yes); \
997 setup_spans_epilogue() \
999 function(setup_spans_down_left)
1000 setup_spans_down_down(left, right)
1002 function(setup_spans_down_right)
1003 setup_spans_down_down(right, left)
1006 #define setup_spans_up_flat() \
1007 sub height, y_a, y_c; \
1009 compute_edge_delta_x2(); \
1010 setup_spans_up(left, right, none, no); \
1011 setup_spans_epilogue() \
1013 function(setup_spans_up_a)
1014 setup_spans_prologue()
1016 vmov.u32 x_starts, x_a, x_b
1017 vdup.u32 x_ends, x_c
1019 setup_spans_up_flat()
1021 function(setup_spans_up_b)
1022 setup_spans_prologue()
1024 vdup.u32 x_starts, x_a
1025 vmov.u32 x_ends, x_b, x_c
1027 setup_spans_up_flat()
1029 #define setup_spans_down_flat() \
1030 sub height, y_c, y_a; \
1032 compute_edge_delta_x2(); \
1033 setup_spans_down(left, right, none, no); \
1034 setup_spans_epilogue() \
1036 function(setup_spans_down_a)
1037 setup_spans_prologue()
1039 vmov.u32 x_starts, x_a, x_b
1040 vdup.u32 x_ends, x_c
1042 setup_spans_down_flat()
1044 function(setup_spans_down_b)
1045 setup_spans_prologue()
1047 vdup.u32 x_starts, x_a
1048 vmov.u32 x_ends, x_b, x_c
1050 setup_spans_down_flat()
1055 #define edges_xy_b q11
1056 #define edges_dx_dy_b d26
1057 #define edge_shifts_b d27
1058 #define edges_dx_dy_and_shifts_b q13
1059 #define height_increment d20
1061 #define edges_dx_dy_and_shifts q1
1063 #define edges_xy_b_left d22
1064 #define edges_xy_b_right d23
1066 #define setup_spans_up_down_load_edge_set_b() \
1067 vmov edges_xy, edges_xy_b; \
1068 vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \
1071 function(setup_spans_up_down)
1072 setup_spans_prologue()
1074 // s32 middle_y = y_a;
1075 sub height_minor_a, y_a, y_b
1076 sub height_minor_b, y_c, y_a
1077 sub height_major, y_c, y_b
1079 vmov.u32 x_starts, x_a, x_c
1080 vdup.u32 x_ends, x_b
1082 compute_edge_delta_x3(x_a, height_minor_a, height_major)
1085 vmov.u32 height_increment, temp, height_minor_b
1086 vmlal.s32 edges_xy, edges_dx_dy, height_increment
1088 vmov edges_xy_b_left, edge_alt_low, edge_alt_high
1089 vmov edges_xy_b_right, edges_xy_right
1091 vmov edge_shifts_b, edge_shifts
1092 vmov.u32 edge_shifts_b[0], edge_shift_alt
1094 vneg.s32 edges_dx_dy_b, edges_dx_dy
1095 vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt
1099 setup_spans_load_b()
1102 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1103 subs temp, temp, y_b
1104 subgt height_minor_a, height_minor_a, temp
1106 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1107 subs clip, y_a, temp
1110 sub height_minor_a, height_minor_a, clip
1112 setup_spans_clip(decrement, no)
1115 cmp height_minor_a, #0
1118 orr temp, y_a, y_a, lsl #16
1119 sub temp, temp, #(1 << 16)
1121 sub y_a, y_a, #(2 << 16)
1122 vmov.u32 y_x4, temp, y_a
1124 vaddw.s32 edges_xy, edges_xy, edges_dx_dy
1126 strh height_minor_a, [ psx_gpu, #psx_gpu_num_spans_offset ]
1128 setup_spans_adjust_edges_alternate_no(left, right);
1129 setup_spans_adjust_interpolants_up()
1130 setup_spans_up_down_load_edge_set_b()
1132 setup_spans_prologue_b()
1136 setup_spans_set_x4_alternate_no(none, up)
1137 subs height_minor_a, height_minor_a, #4
1140 add span_edge_data, span_edge_data, height_minor_a, lsl #3
1141 add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4
1142 add span_b_offset, span_b_offset, height_minor_a, lsl #2
1145 add temp, psx_gpu, #psx_gpu_uvrg_offset
1146 vld1.32 { uvrg }, [ temp ]
1149 setup_spans_load_b()
1151 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1153 subgt height_minor_b, height_minor_b, y_c
1154 addgt height_minor_b, height_minor_b, #1
1156 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1157 subs clip, temp, y_a
1160 sub height_minor_b, height_minor_b, clip
1162 setup_spans_clip(increment, no)
1165 cmp height_minor_b, #0
1168 orr temp, y_a, y_a, lsl #16
1169 add temp, temp, #(1 << 16)
1171 add y_a, y_a, #(2 << 16)
1172 vmov.u32 y_x4, temp, y_a
1174 setup_spans_adjust_edges_alternate_no(left, right)
1176 ldrh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1177 add temp, temp, height_minor_b
1179 cmp temp, #MAX_SPANS
1182 strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1185 setup_spans_set_x4_alternate_no(none, down)
1186 subs height_minor_b, height_minor_b, #4
1190 setup_spans_epilogue()
1193 setup_spans_up_down_load_edge_set_b()
1194 setup_spans_prologue_b()
1198 // FIXME: overflow corner case
1199 sub temp, temp, height_minor_b
1200 bics height_minor_b, #3
1201 add temp, temp, height_minor_b
1202 strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1208 #undef span_uvrg_offset
1209 #undef span_edge_data
1210 #undef span_b_offset
1215 #define num_spans r1
1216 #define span_uvrg_offset r2
1217 #define span_edge_data r3
1218 #define span_b_offset r4
1220 #define span_num_blocks r6
1224 #define dither_offset_ptr r10
1225 #define block_ptr_a r11
1227 #define num_blocks r14
1229 #define uvrg_dx_ptr r2
1230 #define texture_mask_ptr r3
1231 #define dither_shift r8
1232 #define dither_row r10
1237 #define block_ptr_b r10
1239 #define block_span_ptr r10
1240 #define right_mask r8
1259 #define b_whole_8 d14
1260 #define fb_mask_ptrs d15
1273 #define u_whole_low d16
1274 #define u_whole_high d17
1275 #define v_whole_low d18
1276 #define v_whole_high d19
1277 #define r_whole_low d20
1278 #define r_whole_high d21
1279 #define g_whole_low d22
1280 #define g_whole_high d23
1281 #define b_whole_low d24
1282 #define b_whole_high d25
1287 #define u_whole_8 d26
1288 #define v_whole_8 d27
1289 #define u_whole_8b d24
1290 #define r_whole_8 d24
1291 #define g_whole_8 d25
1293 #define uv_whole_8 q13
1294 #define uv_whole_8b q14
1296 #define dither_offsets q14
1297 #define texture_mask q15
1298 #define texture_mask_u d30
1299 #define texture_mask_v d31
1301 #define dither_offsets_short d28
1305 #define block_span q10
1310 #define draw_mask q1
1311 #define draw_mask_edge q13
1312 #define test_mask q0
1318 #define setup_blocks_texture_swizzled() \
1319 vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \
1320 vsli.u8 u_whole_8, v_whole_8, #4; \
1321 vsri.u8 v_whole_8, u_whole_8b, #4 \
1323 #define setup_blocks_texture_unswizzled() \
1326 #define setup_blocks_shaded_textured_builder(swizzling) \
1329 function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \
1330 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1331 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1333 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1334 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1336 cmp num_spans, #0; \
1339 stmdb sp!, { r4 - r11, r14 }; \
1340 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1342 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
1343 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1345 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1346 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1348 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1349 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1351 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
1352 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1354 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1357 vmov.u8 fb_mask_ptrs, #0; \
1359 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1360 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1362 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1363 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
1365 cmp span_num_blocks, #0; \
1368 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1369 add num_blocks, span_num_blocks, num_blocks; \
1371 cmp num_blocks, #MAX_BLOCKS; \
1375 ldr b, [ span_b_offset ]; \
1376 add fb_ptr, fb_ptr, y, lsl #11; \
1378 vdup.u32 v_left_x, left_x; \
1381 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1382 add fb_ptr, fb_ptr, left_x, lsl #1; \
1384 mla b, b_dx, left_x, b; \
1385 and dither_shift, left_x, #0x03; \
1387 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1388 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1390 mov dither_shift, dither_shift, lsl #3; \
1391 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1394 subs span_num_blocks, span_num_blocks, #1; \
1396 mov dither_row, dither_row, ror dither_shift; \
1397 mov b_dx4, b_dx, lsl #2; \
1399 vdup.u32 dither_offsets_short, dither_row; \
1400 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1402 vdup.u32 b_block, b; \
1403 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1405 vdup.u32 u_block, uv[0]; \
1406 mov b_dx8, b_dx, lsl #3; \
1408 vdup.u32 v_block, uv[1]; \
1409 vdup.u32 r_block, rg[0]; \
1410 vdup.u32 g_block, rg[1]; \
1412 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1414 vadd.u32 u_block, u_block, block_span; \
1415 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1417 vadd.u32 v_block, v_block, block_span; \
1418 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1420 vadd.u32 r_block, r_block, block_span; \
1421 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1423 vadd.u32 g_block, g_block, block_span; \
1424 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
1426 vadd.u32 b_block, b_block, block_span; \
1427 add block_ptr_b, block_ptr_a, #16; \
1429 vshrn.u32 u_whole_low, u_block, #16; \
1430 vshrn.u32 v_whole_low, v_block, #16; \
1431 vshrn.u32 r_whole_low, r_block, #16; \
1432 vshrn.u32 g_whole_low, g_block, #16; \
1434 vdup.u32 dx4, uv_dx4[0]; \
1435 vshrn.u32 b_whole_low, b_block, #16; \
1437 vaddhn.u32 u_whole_high, u_block, dx4; \
1438 vdup.u32 dx4, uv_dx4[1]; \
1440 vaddhn.u32 v_whole_high, v_block, dx4; \
1441 vdup.u32 dx4, rg_dx4[0]; \
1443 vaddhn.u32 r_whole_high, r_block, dx4; \
1444 vdup.u32 dx4, rg_dx4[1]; \
1446 vaddhn.u32 g_whole_high, g_block, dx4; \
1447 vdup.u32 dx4, b_dx4; \
1449 vaddhn.u32 b_whole_high, b_block, dx4; \
1450 vdup.u32 dx8, uv_dx8[0]; \
1452 vadd.u32 u_block, u_block, dx8; \
1453 vdup.u32 dx8, uv_dx8[1]; \
1455 vadd.u32 v_block, v_block, dx8; \
1456 vdup.u32 dx8, rg_dx8[0]; \
1458 vadd.u32 r_block, r_block, dx8; \
1459 vdup.u32 dx8, rg_dx8[1]; \
1461 vadd.u32 g_block, g_block, dx8; \
1462 vdup.u32 dx8, b_dx8; \
1464 vadd.u32 b_block, b_block, dx8; \
1465 vmovn.u16 u_whole_8, u_whole; \
1467 vmovn.u16 v_whole_8, v_whole; \
1469 vmovn.u16 b_whole_8, b_whole; \
1471 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1473 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1474 setup_blocks_texture_##swizzling(); \
1476 vmovn.u16 r_whole_8, r_whole; \
1480 vmovn.u16 g_whole_8, g_whole; \
1481 vshrn.u32 u_whole_low, u_block, #16; \
1483 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1484 vshrn.u32 v_whole_low, v_block, #16; \
1486 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1487 vshrn.u32 r_whole_low, r_block, #16; \
1489 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1490 vshrn.u32 g_whole_low, g_block, #16; \
1492 vdup.u32 dx4, uv_dx4[0]; \
1493 vshrn.u32 b_whole_low, b_block, #16; \
1495 vaddhn.u32 u_whole_high, u_block, dx4; \
1496 vdup.u32 dx4, uv_dx4[1]; \
1498 vaddhn.u32 v_whole_high, v_block, dx4; \
1499 vdup.u32 dx4, rg_dx4[0]; \
1501 vaddhn.u32 r_whole_high, r_block, dx4; \
1502 vdup.u32 dx4, rg_dx4[1]; \
1504 vaddhn.u32 g_whole_high, g_block, dx4; \
1505 vdup.u32 dx4, b_dx4; \
1507 vaddhn.u32 b_whole_high, b_block, dx4; \
1508 vdup.u32 dx8, uv_dx8[0]; \
1510 vadd.u32 u_block, u_block, dx8; \
1511 vdup.u32 dx8, uv_dx8[1]; \
1513 vadd.u32 v_block, v_block, dx8; \
1514 vdup.u32 dx8, rg_dx8[0]; \
1516 vadd.u32 r_block, r_block, dx8; \
1517 vdup.u32 dx8, rg_dx8[1]; \
1519 vadd.u32 g_block, g_block, dx8; \
1520 vdup.u32 dx8, b_dx8; \
1522 vadd.u32 b_block, b_block, dx8; \
1523 vmovn.u16 u_whole_8, u_whole; \
1525 add fb_ptr, fb_ptr, #16; \
1526 vmovn.u16 v_whole_8, v_whole; \
1528 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1529 vmovn.u16 b_whole_8, b_whole; \
1533 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1534 subs span_num_blocks, span_num_blocks, #1; \
1536 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1537 setup_blocks_texture_##swizzling(); \
1539 vmovn.u16 r_whole_8, r_whole; \
1543 vmovn.u16 g_whole_8, g_whole; \
1544 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1546 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1547 vdup.u8 draw_mask, right_mask; \
1549 vmov.u32 fb_mask_ptrs[0], right_mask; \
1550 vtst.u16 draw_mask, draw_mask, test_mask; \
1551 vzip.u8 u_whole_8, v_whole_8; \
1553 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1554 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1555 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1556 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1557 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1560 add span_uvrg_offset, span_uvrg_offset, #16; \
1561 add span_b_offset, span_b_offset, #4; \
1563 add span_edge_data, span_edge_data, #8; \
1564 subs num_spans, num_spans, #1; \
1566 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1569 ldmia sp!, { r4 - r11, pc }; \
1572 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1573 vpush { texture_mask }; \
1574 vpush { uvrg_dx4 }; \
1576 stmdb sp!, { r0 - r3, r12, r14 }; \
1577 bl flush_render_block_buffer; \
1578 ldmia sp!, { r0 - r3, r12, r14 }; \
1580 vpop { uvrg_dx4 }; \
1581 vpop { texture_mask }; \
1583 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1584 vmov.u8 fb_mask_ptrs, #0; \
1586 mov num_blocks, span_num_blocks; \
1587 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1591 setup_blocks_shaded_textured_builder(swizzled)
1592 setup_blocks_shaded_textured_builder(unswizzled)
1595 #define setup_blocks_unshaded_textured_builder(swizzling) \
1598 function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \
1599 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1600 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1602 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1603 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1605 cmp num_spans, #0; \
1608 stmdb sp!, { r4 - r11, r14 }; \
1609 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1611 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1613 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1614 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1616 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1617 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1619 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1621 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1624 vmov.u8 fb_mask_ptrs, #0; \
1626 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1627 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1629 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1630 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
1632 cmp span_num_blocks, #0; \
1635 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1636 add num_blocks, span_num_blocks, num_blocks; \
1638 cmp num_blocks, #MAX_BLOCKS; \
1642 add fb_ptr, fb_ptr, y, lsl #11; \
1644 vdup.u32 v_left_x, left_x; \
1647 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1648 add fb_ptr, fb_ptr, left_x, lsl #1; \
1650 and dither_shift, left_x, #0x03; \
1652 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1653 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1655 mov dither_shift, dither_shift, lsl #3; \
1656 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1659 subs span_num_blocks, span_num_blocks, #1; \
1661 mov dither_row, dither_row, ror dither_shift; \
1663 vdup.u32 dither_offsets_short, dither_row; \
1664 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1666 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1668 vdup.u32 u_block, uv[0]; \
1670 vdup.u32 v_block, uv[1]; \
1671 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1673 vadd.u32 u_block, u_block, block_span; \
1674 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1676 vadd.u32 v_block, v_block, block_span; \
1677 add block_ptr_b, block_ptr_a, #16; \
1679 vshrn.u32 u_whole_low, u_block, #16; \
1680 vshrn.u32 v_whole_low, v_block, #16; \
1682 vdup.u32 dx4, uv_dx4[0]; \
1684 vaddhn.u32 u_whole_high, u_block, dx4; \
1685 vdup.u32 dx4, uv_dx4[1]; \
1687 vaddhn.u32 v_whole_high, v_block, dx4; \
1688 vdup.u32 dx8, uv_dx8[0]; \
1690 vadd.u32 u_block, u_block, dx8; \
1691 vdup.u32 dx8, uv_dx8[1]; \
1693 vadd.u32 v_block, v_block, dx8; \
1694 vmovn.u16 u_whole_8, u_whole; \
1696 vmovn.u16 v_whole_8, v_whole; \
1699 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1701 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1702 setup_blocks_texture_##swizzling(); \
1707 vshrn.u32 u_whole_low, u_block, #16; \
1709 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1710 vshrn.u32 v_whole_low, v_block, #16; \
1712 add block_ptr_b, block_ptr_b, #32; \
1713 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1715 vdup.u32 dx4, uv_dx4[0]; \
1716 vaddhn.u32 u_whole_high, u_block, dx4; \
1717 vdup.u32 dx4, uv_dx4[1]; \
1719 vaddhn.u32 v_whole_high, v_block, dx4; \
1720 vdup.u32 dx8, uv_dx8[0]; \
1722 vadd.u32 u_block, u_block, dx8; \
1723 vdup.u32 dx8, uv_dx8[1]; \
1725 vadd.u32 v_block, v_block, dx8; \
1726 vmovn.u16 u_whole_8, u_whole; \
1728 add fb_ptr, fb_ptr, #16; \
1729 vmovn.u16 v_whole_8, v_whole; \
1731 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1734 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1735 subs span_num_blocks, span_num_blocks, #1; \
1737 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1738 setup_blocks_texture_##swizzling(); \
1743 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1745 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1746 vdup.u8 draw_mask, right_mask; \
1748 vmov.u32 fb_mask_ptrs[0], right_mask; \
1749 vtst.u16 draw_mask, draw_mask, test_mask; \
1750 vzip.u8 u_whole_8, v_whole_8; \
1752 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1753 add block_ptr_b, block_ptr_b, #32; \
1754 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1755 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1756 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1759 add span_uvrg_offset, span_uvrg_offset, #16; \
1760 add span_edge_data, span_edge_data, #8; \
1761 subs num_spans, num_spans, #1; \
1763 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1766 ldmia sp!, { r4 - r11, pc }; \
1769 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1770 vpush { texture_mask }; \
1771 vpush { uvrg_dx4 }; \
1773 stmdb sp!, { r0 - r3, r12, r14 }; \
1774 bl flush_render_block_buffer; \
1775 ldmia sp!, { r0 - r3, r12, r14 }; \
1777 vpop { uvrg_dx4 }; \
1778 vpop { texture_mask }; \
1780 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1781 vmov.u8 fb_mask_ptrs, #0; \
1783 mov num_blocks, span_num_blocks; \
1784 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1788 setup_blocks_unshaded_textured_builder(swizzled)
1789 setup_blocks_unshaded_textured_builder(unswizzled)
1794 function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect)
1795 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1796 veor.u32 draw_mask, draw_mask, draw_mask
1801 stmdb sp!, { r4 - r11, r14 }
1802 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1804 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1806 ubfx color_r, color, #3, #5
1807 ubfx color_g, color, #11, #5
1808 ubfx color_b, color, #19, #5
1810 orr color, color_r, color_b, lsl #10
1811 orr color, color, color_g, lsl #5
1813 vdup.u16 colors, color
1815 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1816 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1818 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1819 add block_ptr_a, block_ptr_a, num_blocks, lsl #6
1822 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1823 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1825 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
1827 cmp span_num_blocks, #0
1830 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1831 add num_blocks, span_num_blocks, num_blocks
1833 cmp num_blocks, #MAX_BLOCKS
1837 add fb_ptr, fb_ptr, y, lsl #11
1840 add fb_ptr, fb_ptr, left_x, lsl #1
1843 subs span_num_blocks, span_num_blocks, #1
1845 add block_ptr_b, block_ptr_a, #16
1848 vmov.u32 fb_mask_ptrs[1], fb_ptr
1852 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_32
1853 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1854 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1856 add fb_ptr, fb_ptr, #16
1857 add block_ptr_b, block_ptr_b, #32
1861 vmov.u32 fb_mask_ptrs[1], fb_ptr
1862 subs span_num_blocks, span_num_blocks, #1
1867 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1869 vdup.u8 draw_mask_edge, right_mask
1870 vtst.u16 draw_mask_edge, draw_mask_edge, test_mask
1872 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1873 vst1.u32 { draw_mask_edge }, [ block_ptr_a, :128 ], c_32
1874 add block_ptr_b, block_ptr_b, #32
1875 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1878 add span_edge_data, span_edge_data, #8
1879 subs num_spans, num_spans, #1
1881 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1884 ldmia sp!, { r4 - r11, pc }
1889 stmdb sp!, { r0 - r3, r12, r14 }
1890 bl flush_render_block_buffer
1891 ldmia sp!, { r0 - r3, r12, r14 }
1895 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1896 veor.u32 draw_mask, draw_mask, draw_mask
1898 mov num_blocks, span_num_blocks
1899 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1903 #define mask_msb_scalar r14
1905 #define msb_mask q15
1907 #define pixels_low d16
1909 #define msb_mask_low d30
1910 #define msb_mask_high d31
1915 function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct)
1916 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1921 stmdb sp!, { r4 - r11, r14 }
1923 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1925 ubfx color_r, color, #3, #5
1926 ubfx color_g, color, #11, #5
1928 ldrh mask_msb_scalar, [ psx_gpu, #psx_gpu_mask_msb_offset ]
1929 ubfx color_b, color, #19, #5
1931 orr color, color_r, color_b, lsl #10
1932 orr color, color, color_g, lsl #5
1933 orr color, color, mask_msb_scalar
1935 vdup.u16 colors, color
1937 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1938 orr color, color, lsl #16
1942 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1943 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1945 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
1947 cmp span_num_blocks, #0
1950 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1952 add fb_ptr, fb_ptr, y, lsl #11
1953 subs span_num_blocks, span_num_blocks, #1
1955 add fb_ptr, fb_ptr, left_x, lsl #1
1959 vst1.u32 { colors }, [ fb_ptr ]!
1960 subs span_num_blocks, span_num_blocks, #1
1965 ldrb right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1967 cmp right_mask, #0x0
1970 tst right_mask, #0xF
1971 streq color, [ fb_ptr ], #4
1972 moveq right_mask, right_mask, lsr #4
1973 streq color, [ fb_ptr ], #4
1975 tst right_mask, #0x3
1976 streq color, [ fb_ptr ], #4
1977 moveq right_mask, right_mask, lsr #2
1979 tst right_mask, #0x1
1980 streqh color, [ fb_ptr ]
1983 add span_edge_data, span_edge_data, #8
1984 subs num_spans, num_spans, #1
1987 ldmia sp!, { r4 - r11, pc }
1990 vst1.u32 { colors }, [ fb_ptr ]
1997 #define rg_dx_ptr r2
2015 #undef dither_offsets
2035 #define r_whole_low d6
2036 #define r_whole_high d7
2037 #define g_whole_low d8
2038 #define g_whole_high d9
2039 #define b_whole_low d10
2040 #define b_whole_high d11
2042 #define gb_whole_8 q6
2044 #define g_whole_8 d12
2045 #define b_whole_8 d13
2047 #define r_whole_8 d14
2059 #define block_span q5
2067 #define d128_0x7 q13
2071 #define dither_offsets q14
2072 #define draw_mask q15
2074 #define dither_offsets_low d28
2077 #define test_mask q10
2080 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2081 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2082 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2084 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2085 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2086 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2088 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2090 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2093 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2096 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2097 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2098 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2100 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2102 cmp num_spans, #0; \
2105 stmdb sp!, { r4 - r11, r14 }; \
2106 vshl.u32 rg_dx4, rg_dx, #2; \
2108 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2109 vshl.u32 rg_dx8, rg_dx, #3; \
2111 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2113 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2114 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2116 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2117 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2119 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2120 vmov.u8 d64_1, #1; \
2122 vmov.u8 d128_4, #4; \
2123 vmov.u8 d64_128, #128; \
2125 vmov.u8 d128_0x7, #0x7; \
2128 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2129 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2131 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2132 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
2134 cmp span_num_blocks, #0; \
2137 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2138 add num_blocks, span_num_blocks, num_blocks; \
2140 cmp num_blocks, #MAX_BLOCKS; \
2144 ldr b, [ span_b_offset ]; \
2145 add fb_ptr, fb_ptr, y, lsl #11; \
2147 vdup.u32 v_left_x, left_x; \
2150 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2151 add fb_ptr, fb_ptr, left_x, lsl #1; \
2153 mla b, b_dx, left_x, b; \
2154 and dither_shift, left_x, #0x03; \
2156 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2157 vshr.u32 rg_dx, rg_dx4, #2; \
2159 mov dither_shift, dither_shift, lsl #3; \
2160 vmla.u32 rg, rg_dx, v_left_x; \
2163 subs span_num_blocks, span_num_blocks, #1; \
2165 mov dither_row, dither_row, ror dither_shift; \
2166 mov b_dx4, b_dx, lsl #2; \
2168 vdup.u32 dither_offsets, dither_row; \
2169 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2171 vdup.u32 b_block, b; \
2172 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2174 mov b_dx8, b_dx, lsl #3; \
2175 vdup.u32 r_block, rg[0]; \
2176 vdup.u32 g_block, rg[1]; \
2178 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2180 vadd.u32 r_block, r_block, block_span; \
2181 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2183 vadd.u32 g_block, g_block, block_span; \
2184 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2186 vadd.u32 b_block, b_block, block_span; \
2187 add block_ptr_b, block_ptr_a, #16; \
2189 vshrn.u32 r_whole_low, r_block, #16; \
2190 vshrn.u32 g_whole_low, g_block, #16; \
2191 vshrn.u32 b_whole_low, b_block, #16; \
2192 vdup.u32 dx4, rg_dx4[0]; \
2194 vaddhn.u32 r_whole_high, r_block, dx4; \
2195 vdup.u32 dx4, rg_dx4[1]; \
2197 vaddhn.u32 g_whole_high, g_block, dx4; \
2198 vdup.u32 dx4, b_dx4; \
2200 vaddhn.u32 b_whole_high, b_block, dx4; \
2201 vdup.u32 dx8, rg_dx8[0]; \
2203 vadd.u32 r_block, r_block, dx8; \
2204 vdup.u32 dx8, rg_dx8[1]; \
2206 vadd.u32 g_block, g_block, dx8; \
2207 vdup.u32 dx8, b_dx8; \
2209 vadd.u32 b_block, b_block, dx8; \
2211 vmovn.u16 r_whole_8, r_whole; \
2212 vmovn.u16 g_whole_8, g_whole; \
2213 vmovn.u16 b_whole_8, b_whole; \
2216 veor.u32 draw_mask, draw_mask, draw_mask; \
2219 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2220 vshrn.u32 r_whole_low, r_block, #16; \
2222 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2223 vshrn.u32 g_whole_low, g_block, #16; \
2225 vshrn.u32 b_whole_low, b_block, #16; \
2226 str fb_ptr, [ block_ptr_a, #44 ]; \
2228 vdup.u32 dx4, rg_dx4[0]; \
2229 vshr.u8 r_whole_8, r_whole_8, #3; \
2230 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2232 vaddhn.u32 r_whole_high, r_block, dx4; \
2233 vdup.u32 dx4, rg_dx4[1]; \
2235 vaddhn.u32 g_whole_high, g_block, dx4; \
2236 vdup.u32 dx4, b_dx4; \
2238 vaddhn.u32 b_whole_high, b_block, dx4; \
2239 vdup.u32 dx8, rg_dx8[0]; \
2241 vmull.u8 pixels, r_whole_8, d64_1; \
2242 vmlal.u8 pixels, g_whole_8, d64_4; \
2243 vmlal.u8 pixels, b_whole_8, d64_128; \
2245 vadd.u32 r_block, r_block, dx8; \
2246 vdup.u32 dx8, rg_dx8[1]; \
2248 vadd.u32 g_block, g_block, dx8; \
2249 vdup.u32 dx8, b_dx8; \
2251 vadd.u32 b_block, b_block, dx8; \
2252 add fb_ptr, fb_ptr, #16; \
2254 vmovn.u16 r_whole_8, r_whole; \
2255 vmovn.u16 g_whole_8, g_whole; \
2256 vmovn.u16 b_whole_8, b_whole; \
2258 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2259 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2263 subs span_num_blocks, span_num_blocks, #1; \
2267 str fb_ptr, [ block_ptr_a, #44 ]; \
2268 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2270 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2271 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2273 vshr.u8 r_whole_8, r_whole_8, #3; \
2274 vdup.u8 draw_mask, right_mask; \
2276 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2277 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
2279 vtst.u16 draw_mask, draw_mask, test_mask; \
2281 vmull.u8 pixels, r_whole_8, d64_1; \
2282 vmlal.u8 pixels, g_whole_8, d64_4; \
2283 vmlal.u8 pixels, b_whole_8, d64_128; \
2285 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2286 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2289 add span_uvrg_offset, span_uvrg_offset, #16; \
2290 add span_b_offset, span_b_offset, #4; \
2292 add span_edge_data, span_edge_data, #8; \
2293 subs num_spans, num_spans, #1; \
2295 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2298 ldmia sp!, { r4 - r11, pc }; \
2301 /* TODO: Load from psx_gpu instead of saving/restoring these */\
2304 stmdb sp!, { r0 - r3, r12, r14 }; \
2305 bl flush_render_block_buffer; \
2306 ldmia sp!, { r0 - r3, r12, r14 }; \
2310 vmov.u8 d64_1, #1; \
2311 vmov.u8 d128_4, #4; \
2312 vmov.u8 d64_128, #128; \
2313 vmov.u8 d128_0x7, #0x7; \
2315 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2317 mov num_blocks, span_num_blocks; \
2318 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2322 setup_blocks_shaded_untextured_indirect_builder(undithered)
2323 setup_blocks_shaded_untextured_indirect_builder(dithered)
2328 #define mask_msb_ptr r14
2330 #define draw_mask q0
2331 #define pixels_low d16
2332 #define pixels_high d17
2336 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2339 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2340 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2341 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2343 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2345 cmp num_spans, #0; \
2348 stmdb sp!, { r4 - r11, r14 }; \
2349 vshl.u32 rg_dx4, rg_dx, #2; \
2351 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2352 vshl.u32 rg_dx8, rg_dx, #3; \
2354 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2355 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2357 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2358 vmov.u8 d64_1, #1; \
2360 vmov.u8 d128_4, #4; \
2361 vmov.u8 d64_128, #128; \
2363 vmov.u8 d128_0x7, #0x7; \
2364 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2365 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
2368 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2369 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2371 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2372 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
2374 cmp span_num_blocks, #0; \
2377 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2378 add fb_ptr, fb_ptr, y, lsl #11; \
2380 ldr b, [ span_b_offset ]; \
2381 vdup.u32 v_left_x, left_x; \
2384 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2385 add fb_ptr, fb_ptr, left_x, lsl #1; \
2387 mla b, b_dx, left_x, b; \
2388 and dither_shift, left_x, #0x03; \
2390 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2391 vshr.u32 rg_dx, rg_dx4, #2; \
2393 mov dither_shift, dither_shift, lsl #3; \
2394 vmla.u32 rg, rg_dx, v_left_x; \
2396 subs span_num_blocks, span_num_blocks, #1; \
2398 mov dither_row, dither_row, ror dither_shift; \
2399 mov b_dx4, b_dx, lsl #2; \
2401 vdup.u32 dither_offsets, dither_row; \
2402 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2404 vdup.u32 b_block, b; \
2405 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2407 mov b_dx8, b_dx, lsl #3; \
2408 vdup.u32 r_block, rg[0]; \
2409 vdup.u32 g_block, rg[1]; \
2411 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2413 vadd.u32 r_block, r_block, block_span; \
2414 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2416 vadd.u32 g_block, g_block, block_span; \
2417 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2419 vadd.u32 b_block, b_block, block_span; \
2420 add block_ptr_b, block_ptr_a, #16; \
2422 vshrn.u32 r_whole_low, r_block, #16; \
2423 vshrn.u32 g_whole_low, g_block, #16; \
2424 vshrn.u32 b_whole_low, b_block, #16; \
2425 vdup.u32 dx4, rg_dx4[0]; \
2427 vaddhn.u32 r_whole_high, r_block, dx4; \
2428 vdup.u32 dx4, rg_dx4[1]; \
2430 vaddhn.u32 g_whole_high, g_block, dx4; \
2431 vdup.u32 dx4, b_dx4; \
2433 vaddhn.u32 b_whole_high, b_block, dx4; \
2434 vdup.u32 dx8, rg_dx8[0]; \
2436 vadd.u32 r_block, r_block, dx8; \
2437 vdup.u32 dx8, rg_dx8[1]; \
2439 vadd.u32 g_block, g_block, dx8; \
2440 vdup.u32 dx8, b_dx8; \
2442 vadd.u32 b_block, b_block, dx8; \
2444 vmovn.u16 r_whole_8, r_whole; \
2445 vmovn.u16 g_whole_8, g_whole; \
2446 vmovn.u16 b_whole_8, b_whole; \
2451 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2452 vshrn.u32 r_whole_low, r_block, #16; \
2454 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2455 vshrn.u32 g_whole_low, g_block, #16; \
2457 vshrn.u32 b_whole_low, b_block, #16; \
2459 vdup.u32 dx4, rg_dx4[0]; \
2460 vshr.u8 r_whole_8, r_whole_8, #3; \
2461 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2463 vaddhn.u32 r_whole_high, r_block, dx4; \
2464 vdup.u32 dx4, rg_dx4[1]; \
2466 vmov pixels, msb_mask; \
2467 vaddhn.u32 g_whole_high, g_block, dx4; \
2468 vdup.u32 dx4, b_dx4; \
2470 vaddhn.u32 b_whole_high, b_block, dx4; \
2471 vdup.u32 dx8, rg_dx8[0]; \
2473 vmlal.u8 pixels, r_whole_8, d64_1; \
2474 vmlal.u8 pixels, g_whole_8, d64_4; \
2475 vmlal.u8 pixels, b_whole_8, d64_128; \
2477 vadd.u32 r_block, r_block, dx8; \
2478 vdup.u32 dx8, rg_dx8[1]; \
2480 vadd.u32 g_block, g_block, dx8; \
2481 vdup.u32 dx8, b_dx8; \
2483 vadd.u32 b_block, b_block, dx8; \
2485 vmovn.u16 r_whole_8, r_whole; \
2486 vmovn.u16 g_whole_8, g_whole; \
2487 vmovn.u16 b_whole_8, b_whole; \
2489 vst1.u32 { pixels }, [ fb_ptr ]!; \
2490 subs span_num_blocks, span_num_blocks, #1; \
2494 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2496 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2497 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2499 vshr.u8 r_whole_8, r_whole_8, #3; \
2500 rbit right_mask, right_mask; \
2501 vmov pixels, msb_mask; \
2502 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2503 clz right_mask, right_mask; \
2505 vmlal.u8 pixels, r_whole_8, d64_1; \
2506 vmlal.u8 pixels, g_whole_8, d64_4; \
2507 vmlal.u8 pixels, b_whole_8, d64_128; \
2509 ldr pc, [ pc, right_mask, lsl #2 ]; \
2522 vst1.u16 { pixels_low[0] }, [ fb_ptr ]; \
2526 vst1.u32 { pixels_low[0] }, [ fb_ptr ]; \
2530 vst1.u32 { pixels_low[0] }, [ fb_ptr ]!; \
2531 vst1.u16 { pixels_low[2] }, [ fb_ptr ]; \
2535 vst1.u32 { pixels_low }, [ fb_ptr ]; \
2539 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2540 vst1.u16 { pixels_high[0] }, [ fb_ptr ]; \
2544 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2545 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2549 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2550 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2551 vst1.u16 { pixels_high[2] }, [ fb_ptr ]; \
2555 vst1.u32 { pixels }, [ fb_ptr ]; \
2559 add span_uvrg_offset, span_uvrg_offset, #16; \
2560 add span_b_offset, span_b_offset, #4; \
2562 add span_edge_data, span_edge_data, #8; \
2563 subs num_spans, num_spans, #1; \
2567 ldmia sp!, { r4 - r11, pc } \
2569 setup_blocks_shaded_untextured_direct_builder(undithered)
2570 setup_blocks_shaded_untextured_direct_builder(dithered)
2579 #define block_ptr r1
2580 #define num_blocks r2
2593 #define texture_ptr r11
2607 #define pixels_d r10
2611 #define clut_ptr r12
2612 #define current_texture_mask r5
2613 #define dirty_textures_mask r6
2617 #define clut_low_a d2
2618 #define clut_low_b d3
2619 #define clut_high_a d4
2620 #define clut_high_b d5
2625 #define texels_low d6
2626 #define texels_high d7
2630 function(texture_blocks_untextured)
2636 function(texture_blocks_4bpp)
2637 stmdb sp!, { r3 - r11, r14 }
2638 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2640 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2641 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2643 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2644 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]
2646 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2647 vuzp.u8 clut_a, clut_b
2649 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
2650 tst dirty_textures_mask, current_texture_mask
2656 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2658 uxtah uv_0, texture_ptr, uv_01
2659 uxtah uv_1, texture_ptr, uv_01, ror #16
2661 uxtah uv_2, texture_ptr, uv_23
2662 uxtah uv_3, texture_ptr, uv_23, ror #16
2664 uxtah uv_4, texture_ptr, uv_45
2665 ldrb pixel_0, [ uv_0 ]
2667 uxtah uv_5, texture_ptr, uv_45, ror #16
2668 ldrb pixel_1, [ uv_1 ]
2670 uxtah uv_6, texture_ptr, uv_67
2671 ldrb pixel_2, [ uv_2 ]
2673 uxtah uv_7, texture_ptr, uv_67, ror #16
2674 ldrb pixel_3, [ uv_3 ]
2676 ldrb pixel_4, [ uv_4 ]
2677 subs num_blocks, num_blocks, #1
2679 ldrb pixel_5, [ uv_5 ]
2680 orr pixels_a, pixel_0, pixel_1, lsl #8
2682 ldrb pixel_6, [ uv_6 ]
2683 orr pixels_b, pixel_4, pixel_5, lsl #8
2685 ldrb pixel_7, [ uv_7 ]
2686 orr pixels_a, pixels_a, pixel_2, lsl #16
2688 orr pixels_b, pixels_b, pixel_6, lsl #16
2689 orr pixels_a, pixels_a, pixel_3, lsl #24
2691 orr pixels_b, pixels_b, pixel_7, lsl #24
2692 vmov.u32 texels, pixels_a, pixels_b
2694 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2695 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2697 vst2.u8 { texels_low, texels_high }, [ block_ptr, :128 ], c_64
2700 ldmia sp!, { r3 - r11, pc }
2703 stmdb sp!, { r1 - r2 }
2704 bl update_texture_4bpp_cache
2707 ldmia sp!, { r1 - r2 }
2713 function(texture_blocks_8bpp)
2714 stmdb sp!, { r3 - r11, r14 }
2715 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2717 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2718 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2720 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2721 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2723 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]
2724 tst dirty_textures_mask, current_texture_mask
2730 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2732 uxtah uv_0, texture_ptr, uv_01
2733 uxtah uv_1, texture_ptr, uv_01, ror #16
2735 uxtah uv_2, texture_ptr, uv_23
2736 uxtah uv_3, texture_ptr, uv_23, ror #16
2738 uxtah uv_4, texture_ptr, uv_45
2739 ldrb pixel_0, [ uv_0 ]
2741 uxtah uv_5, texture_ptr, uv_45, ror #16
2742 ldrb pixel_1, [ uv_1 ]
2744 uxtah uv_6, texture_ptr, uv_67
2745 ldrb pixel_2, [ uv_2 ]
2747 uxtah uv_7, texture_ptr, uv_67, ror #16
2748 ldrb pixel_3, [ uv_3 ]
2750 ldrb pixel_4, [ uv_4 ]
2751 add pixel_0, pixel_0, pixel_0
2753 ldrb pixel_5, [ uv_5 ]
2754 add pixel_1, pixel_1, pixel_1
2756 ldrb pixel_6, [ uv_6 ]
2757 add pixel_2, pixel_2, pixel_2
2759 ldrb pixel_7, [ uv_7 ]
2760 add pixel_3, pixel_3, pixel_3
2762 ldrh pixel_0, [ clut_ptr, pixel_0 ]
2763 add pixel_4, pixel_4, pixel_4
2765 ldrh pixel_1, [ clut_ptr, pixel_1 ]
2766 add pixel_5, pixel_5, pixel_5
2768 ldrh pixel_2, [ clut_ptr, pixel_2 ]
2769 add pixel_6, pixel_6, pixel_6
2771 ldrh pixel_3, [ clut_ptr, pixel_3 ]
2772 add pixel_7, pixel_7, pixel_7
2774 ldrh pixel_4, [ clut_ptr, pixel_4 ]
2775 orr pixels_a, pixel_0, pixel_1, lsl #16
2777 ldrh pixel_5, [ clut_ptr, pixel_5 ]
2778 orr pixels_c, pixel_2, pixel_3, lsl #16
2780 ldrh pixel_6, [ clut_ptr, pixel_6 ]
2781 subs num_blocks, num_blocks, #1
2783 ldrh pixel_7, [ clut_ptr, pixel_7 ]
2784 orr pixels_b, pixel_4, pixel_5, lsl #16
2786 orr pixels_d, pixel_6, pixel_7, lsl #16
2787 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2789 add block_ptr, block_ptr, #64
2792 ldmia sp!, { r3 - r11, pc }
2795 stmdb sp!, { r1 - r2, r12 }
2797 bl update_texture_8bpp_cache
2799 ldmia sp!, { r1 - r2, r12 }
2829 #define block_ptr r1
2830 #define num_blocks r2
2874 #define texture_ptr r12
2879 function(texture_blocks_16bpp)
2880 stmdb sp!, { r3 - r11, r14 }
2881 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2883 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2884 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2887 ldrh uv_0, [ block_ptr ]
2888 subs num_blocks, num_blocks, #1
2890 ldrh uv_1, [ block_ptr, #2 ]
2892 and v_0, uv_0, #0xFF00
2893 and v_1, uv_1, #0xFF00
2895 and u_0, uv_0, #0xFF
2896 and u_1, uv_1, #0xFF
2898 add uv_0, u_0, v_0, lsl #2
2899 ldrh uv_2, [ block_ptr, #4 ]
2901 add uv_1, u_1, v_1, lsl #2
2902 ldrh uv_3, [ block_ptr, #6 ]
2904 add uv_0, uv_0, uv_0
2905 add uv_1, uv_1, uv_1
2907 and v_2, uv_2, #0xFF00
2908 and v_3, uv_3, #0xFF00
2910 and u_2, uv_2, #0xFF
2911 and u_3, uv_3, #0xFF
2913 add uv_2, u_2, v_2, lsl #2
2914 ldrh uv_4, [ block_ptr, #8 ]
2916 add uv_3, u_3, v_3, lsl #2
2917 ldrh uv_5, [ block_ptr, #10 ]
2919 add uv_2, uv_2, uv_2
2920 add uv_3, uv_3, uv_3
2922 and v_4, uv_4, #0xFF00
2923 and v_5, uv_5, #0xFF00
2925 and u_4, uv_4, #0xFF
2926 and u_5, uv_5, #0xFF
2928 add uv_4, u_4, v_4, lsl #2
2929 ldrh uv_6, [ block_ptr, #12 ]
2931 add uv_5, u_5, v_5, lsl #2
2932 ldrh uv_7, [ block_ptr, #14 ]
2934 add uv_4, uv_4, uv_4
2935 ldrh pixel_0, [ texture_ptr, uv_0 ]
2937 add uv_5, uv_5, uv_5
2938 ldrh pixel_1, [ texture_ptr, uv_1 ]
2940 and v_6, uv_6, #0xFF00
2941 ldrh pixel_2, [ texture_ptr, uv_2 ]
2943 and v_7, uv_7, #0xFF00
2944 ldrh pixel_3, [ texture_ptr, uv_3 ]
2946 and u_6, uv_6, #0xFF
2947 ldrh pixel_4, [ texture_ptr, uv_4 ]
2949 and u_7, uv_7, #0xFF
2950 ldrh pixel_5, [ texture_ptr, uv_5 ]
2952 add uv_6, u_6, v_6, lsl #2
2953 add uv_7, u_7, v_7, lsl #2
2955 add uv_6, uv_6, uv_6
2956 add uv_7, uv_7, uv_7
2958 orr pixels_a, pixel_0, pixel_1, lsl #16
2959 orr pixels_b, pixel_2, pixel_3, lsl #16
2961 ldrh pixel_6, [ texture_ptr, uv_6 ]
2962 orr pixels_c, pixel_4, pixel_5, lsl #16
2964 ldrh pixel_7, [ texture_ptr, uv_7 ]
2965 orr pixels_d, pixel_6, pixel_7, lsl #16
2967 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
2968 add block_ptr, block_ptr, #64
2972 ldmia sp!, { r3 - r11, pc }
2987 #undef msb_mask_high
2995 #define num_blocks r1
2996 #define color_ptr r2
2997 #define colors_scalar r2
2998 #define colors_scalar_compare r3
2999 #define mask_msb_ptr r2
3001 #define block_ptr_load_a r0
3002 #define block_ptr_store r3
3003 #define block_ptr_load_b r12
3008 #define draw_mask_bits_scalar r5
3010 #define d128_0x07 q0
3011 #define d128_0x1F q1
3012 #define d128_0x8000 q2
3013 #define test_mask q3
3015 #define colors_rg q5
3016 #define colors_b_dm_bits q6
3017 #define texels_rg q7
3020 #define pixels_b q10
3022 #define zero_mask q4
3023 #define draw_mask q12
3024 #define msb_mask q13
3026 #define fb_pixels q8
3028 #define pixels_gb_low q9
3030 #define colors_r d10
3031 #define colors_g d11
3032 #define colors_b d12
3033 #define draw_mask_bits d13
3034 #define texels_r d14
3035 #define texels_g d15
3036 #define pixels_r_low d16
3037 #define pixels_g_low d18
3038 #define pixels_b_low d19
3039 #define msb_mask_low d26
3040 #define msb_mask_high d27
3045 #define texels_b d31
3047 #define shade_blocks_textured_modulated_prologue_indirect() \
3049 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3051 #define shade_blocks_textured_modulated_prologue_direct() \
3052 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3053 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ] \
3056 #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \
3058 #define shade_blocks_textured_false_modulation_check_undithered(target) \
3059 ldr colors_scalar, [ psx_gpu, #psx_gpu_triangle_color_offset ]; \
3060 movw colors_scalar_compare, #0x8080; \
3062 movt colors_scalar_compare, #0x80; \
3063 cmp colors_scalar, colors_scalar_compare; \
3064 beq shade_blocks_textured_unmodulated_##target \
3066 #define shade_blocks_textured_false_modulation_check_dithered(target) \
3068 #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \
3069 shade_blocks_textured_false_modulation_check_##dithering(target); \
3070 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3071 vld1.u32 { colors_r[] }, [ color_ptr, :32 ]; \
3072 vdup.u8 colors_g, colors_r[1]; \
3073 vdup.u8 colors_b, colors_r[2]; \
3074 vdup.u8 colors_r, colors_r[0] \
3077 #define shade_blocks_textured_modulated_load_dithered(target) \
3078 vld1.u32 { target }, [ block_ptr_load_b, :128 ] \
3080 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3081 vld1.u32 { target }, [ block_ptr_load_b, :128 ], c_32 \
3083 #define shade_blocks_textured_modulated_load_undithered(target) \
3085 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3086 add block_ptr_load_b, block_ptr_load_b, #32 \
3088 #define shade_blocks_textured_modulate_dithered(channel) \
3089 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3091 #define shade_blocks_textured_modulate_undithered(channel) \
3092 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3095 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3096 vst1.u32 { draw_mask }, [ block_ptr_store, :128 ]! \
3098 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3099 ldr fb_ptr, [ block_ptr_load_b, #(offset - 64) ]; \
3100 vld1.u32 { fb_pixels }, [ fb_ptr ]; \
3101 vbit.u16 pixels, fb_pixels, draw_mask \
3103 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3104 vst1.u32 { pixels }, [ block_ptr_store, :128 ], c_48 \
3106 #define shade_blocks_textured_modulated_store_pixels_direct() \
3107 vst1.u32 { pixels }, [ fb_ptr ] \
3110 #define shade_blocks_textured_modulated_load_rg_shaded() \
3111 vld1.u32 { colors_r, colors_g }, [ block_ptr_load_b, :128 ], c_32 \
3113 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3114 add block_ptr_load_b, block_ptr_load_b, #32 \
3116 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3117 vld1.u32 { colors_b, draw_mask_bits }, [ block_ptr_load_a, :128 ], c_32 \
3119 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3120 ldr draw_mask_bits_scalar, [ block_ptr_load_a, #8 ]; \
3121 add block_ptr_load_a, block_ptr_load_a, #32 \
3123 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3124 vdup.u16 draw_mask, draw_mask_bits[0] \
3126 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3127 vdup.u16 draw_mask, draw_mask_bits_scalar \
3130 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3132 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3133 vorr.u16 pixels, pixels, msb_mask \
3136 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3139 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3140 shade_blocks_textured_modulated_prologue_##shading(dithering, target); \
3141 stmdb sp!, { r4 - r5, lr }; \
3142 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3144 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
3146 shade_blocks_textured_modulated_prologue_##target(); \
3148 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3151 add block_ptr_load_b, block_ptr_load_a, #16; \
3152 vmov.u8 d64_1, #1; \
3153 vmov.u8 d64_4, #4; \
3154 vmov.u8 d64_128, #128; \
3156 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3157 vmov.u8 d128_0x07, #0x07; \
3159 shade_blocks_textured_modulated_load_rg_##shading(); \
3160 vmov.u8 d128_0x1F, #0x1F; \
3162 shade_blocks_textured_modulated_load_bdm_##shading(); \
3163 vmov.u16 d128_0x8000, #0x8000; \
3165 vmovn.u16 texels_r, texels; \
3166 vshrn.u16 texels_g, texels, #5; \
3168 vshrn.u16 texels_b, texels, #7; \
3169 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3171 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3172 vtst.u16 draw_mask, draw_mask, test_mask; \
3174 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3175 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3177 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3178 vshr.u8 texels_b, texels_b, #3; \
3180 shade_blocks_textured_modulate_##dithering(r); \
3181 shade_blocks_textured_modulate_##dithering(g); \
3182 shade_blocks_textured_modulate_##dithering(b); \
3184 vand.u16 pixels, texels, d128_0x8000; \
3185 vceq.u16 zero_mask, texels, #0; \
3187 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3188 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3189 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3191 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3192 vorr.u16 draw_mask, draw_mask, zero_mask; \
3193 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3194 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3196 subs num_blocks, num_blocks, #1; \
3202 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3203 shade_blocks_textured_modulated_load_rg_##shading(); \
3204 vshrn.u16 texels_g, texels, #5; \
3206 shade_blocks_textured_modulated_load_bdm_##shading(); \
3207 vshrn.u16 texels_b, texels, #7; \
3209 pld [ block_ptr_load_a ]; \
3210 vmovn.u16 texels_r, texels; \
3211 vmlal.u8 pixels, pixels_r_low, d64_1; \
3213 vmlal.u8 pixels, pixels_g_low, d64_4; \
3214 vmlal.u8 pixels, pixels_b_low, d64_128; \
3215 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3217 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3218 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3220 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3221 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3223 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3224 vtst.u16 draw_mask, draw_mask, test_mask; \
3226 shade_blocks_textured_modulated_store_pixels_##target(); \
3227 vshr.u8 texels_b, texels_b, #3; \
3229 shade_blocks_textured_modulate_##dithering(r); \
3230 shade_blocks_textured_modulate_##dithering(g); \
3231 shade_blocks_textured_modulate_##dithering(b); \
3233 vand.u16 pixels, texels, d128_0x8000; \
3234 vceq.u16 zero_mask, texels, #0; \
3236 subs num_blocks, num_blocks, #1; \
3238 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3239 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3240 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3242 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3243 vorr.u16 draw_mask, draw_mask, zero_mask; \
3244 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3245 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3250 vmlal.u8 pixels, pixels_r_low, d64_1; \
3251 vmlal.u8 pixels, pixels_g_low, d64_4; \
3252 vmlal.u8 pixels, pixels_b_low, d64_128; \
3254 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3255 shade_blocks_textured_modulated_store_pixels_##target(); \
3257 ldmia sp!, { r4 - r5, pc } \
3260 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3261 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3262 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3263 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3265 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3266 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3267 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3268 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3286 #undef msb_mask_high
3289 #define num_blocks r1
3290 #define mask_msb_ptr r2
3291 #define color_ptr r3
3293 #define block_ptr_load r0
3294 #define draw_mask_store_ptr r3
3295 #define draw_mask_bits_ptr r12
3296 #define draw_mask_ptr r12
3297 #define pixel_store_ptr r14
3299 #define fb_ptr_cmp r4
3302 #define fb_ptr_next r14
3306 #define test_mask q0
3308 #define draw_mask q2
3309 #define zero_mask q3
3310 #define draw_mask_combined q4
3311 #define fb_pixels q5
3312 #define fb_pixels_next q6
3315 #define draw_mask_low d4
3316 #define draw_mask_high d5
3317 #define msb_mask_low d14
3318 #define msb_mask_high d15
3321 function(shade_blocks_textured_unmodulated_indirect)
3322 str r14, [ sp, #-4 ]
3323 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3325 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3326 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3328 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3329 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3332 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3334 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3335 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3336 [ draw_mask_bits_ptr, :16 ], c_64
3337 vceq.u16 zero_mask, pixels, #0
3339 vtst.u16 draw_mask, draw_mask, test_mask
3340 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3342 subs num_blocks, num_blocks, #1
3346 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3347 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3349 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3350 [ draw_mask_bits_ptr, :16 ], c_64
3351 vceq.u16 zero_mask, pixels, #0
3353 vtst.u16 draw_mask, draw_mask, test_mask
3354 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3356 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3357 subs num_blocks, num_blocks, #1
3362 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3363 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3370 function(shade_blocks_textured_unmodulated_direct)
3371 stmdb sp!, { r4, r14 }
3372 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3374 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3375 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3377 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3380 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3381 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3383 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3384 [ draw_mask_bits_ptr, :16 ], c_64
3385 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3387 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3388 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3389 vceq.u16 zero_mask, pixels, #0
3390 vtst.u16 draw_mask, draw_mask, test_mask
3392 subs num_blocks, num_blocks, #1
3396 mov fb_ptr, fb_ptr_next
3397 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3399 vorr.u16 pixels, pixels, msb_mask
3401 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3402 vmov fb_pixels, fb_pixels_next
3404 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3405 [ draw_mask_bits_ptr, :16 ], c_64
3406 vbif.u16 fb_pixels, pixels, draw_mask_combined
3408 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3409 pld [ fb_ptr_next, #64 ]
3411 add fb_ptr_cmp, fb_ptr_cmp, #14
3412 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3417 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3418 vceq.u16 zero_mask, pixels, #0
3420 vst1.u16 { fb_pixels }, [ fb_ptr ]
3421 vtst.u16 draw_mask, draw_mask, test_mask
3424 subs num_blocks, num_blocks, #1
3428 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3429 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3431 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3433 ldmia sp!, { r4, pc }
3436 vst1.u16 { fb_pixels }, [ fb_ptr ]
3437 vceq.u16 zero_mask, pixels, #0
3439 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3440 vtst.u16 draw_mask, draw_mask, test_mask
3445 function(shade_blocks_unshaded_untextured_indirect)
3450 function(shade_blocks_unshaded_untextured_direct)
3451 stmdb sp!, { r4, r14 }
3452 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3454 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3455 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3457 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3458 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3460 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3461 vld1.u16 { pixels }, [ color_ptr, :128 ]
3464 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3466 vorr.u16 pixels, pixels, msb_mask
3467 subs num_blocks, num_blocks, #1
3469 ldr fb_ptr_next, [ block_ptr_load ], #64
3471 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3475 vmov fb_pixels, fb_pixels_next
3476 mov fb_ptr, fb_ptr_next
3477 ldr fb_ptr_next, [ block_ptr_load ], #64
3479 vbif.u16 fb_pixels, pixels, draw_mask
3480 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3482 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3483 add fb_ptr_cmp, fb_ptr_cmp, #14
3487 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3488 vst1.u16 { fb_pixels }, [ fb_ptr ]
3491 subs num_blocks, num_blocks, #1
3495 vbif.u16 fb_pixels_next, pixels, draw_mask
3496 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3498 ldmia sp!, { r4, pc }
3501 vst1.u16 { fb_pixels }, [ fb_ptr ]
3502 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3506 #undef draw_mask_ptr
3513 #define num_blocks r1
3514 #define msb_mask_ptr r2
3515 #define pixel_ptr r3
3516 #define draw_mask_ptr r0
3519 #define fb_ptr_next r14
3520 #define fb_ptr_cmp r4
3528 #undef msb_mask_high
3529 #undef draw_mask_next
3532 #undef fb_pixels_next
3535 #define draw_mask q1
3537 #define fb_pixels q3
3538 #define blend_pixels q4
3539 #define pixels_no_msb q5
3540 #define blend_mask q6
3541 #define fb_pixels_no_msb q7
3542 #define d128_0x8000 q8
3543 #define d128_0x0421 q9
3544 #define fb_pixels_next q10
3545 #define blend_pixels_next q11
3546 #define pixels_next q12
3547 #define draw_mask_next q13
3548 #define write_mask q14
3550 #define pixels_rb q5
3551 #define pixels_mg q7
3553 #define d128_0x7C1F q8
3554 #define d128_0x03E0 q9
3555 #define fb_pixels_rb q10
3556 #define fb_pixels_g q11
3557 #define fb_pixels_masked q11
3558 #define d128_0x83E0 q15
3559 #define pixels_fourth q7
3560 #define d128_0x1C07 q12
3561 #define d128_0x00E0 q13
3562 #define d128_0x80E0 q13
3564 #define msb_mask_low d0
3565 #define msb_mask_high d1
3567 #define blend_blocks_average_set_blend_mask_textured(source) \
3568 vclt.s16 blend_mask, source, #0 \
3570 #define blend_blocks_average_set_stp_bit_textured() \
3571 vorr.u16 blend_pixels, #0x8000 \
3573 #define blend_blocks_average_combine_textured(source) \
3574 vbif.u16 blend_pixels, source, blend_mask \
3576 #define blend_blocks_average_set_blend_mask_untextured(source) \
3578 #define blend_blocks_average_set_stp_bit_untextured() \
3580 #define blend_blocks_average_combine_untextured(source) \
3582 #define blend_blocks_average_mask_set_on() \
3583 vclt.s16 write_mask, fb_pixels_next, #0 \
3585 #define blend_blocks_average_mask_copy_on() \
3586 vorr.u16 draw_mask, draw_mask_next, write_mask \
3588 #define blend_blocks_average_mask_copy_b_on() \
3589 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3591 #define blend_blocks_average_mask_set_off() \
3593 #define blend_blocks_average_mask_copy_off() \
3594 vmov draw_mask, draw_mask_next \
3596 #define blend_blocks_average_mask_copy_b_off() \
3598 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3601 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3602 stmdb sp!, { r4, r14 }; \
3603 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3604 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3606 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3607 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3609 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3612 vmov.u16 d128_0x8000, #0x8000; \
3613 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3614 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3616 vmov.u16 d128_0x0421, #0x0400; \
3617 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3619 vorr.u16 d128_0x0421, #0x0021; \
3620 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3622 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3623 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3624 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3625 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3626 blend_blocks_average_mask_set_##mask_evaluate(); \
3627 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3629 subs num_blocks, num_blocks, #1; \
3633 mov fb_ptr, fb_ptr_next; \
3634 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3636 vmov pixels, pixels_next; \
3637 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3639 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3641 blend_blocks_average_mask_copy_##mask_evaluate(); \
3642 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3644 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3645 blend_blocks_average_set_stp_bit_##texturing(); \
3646 vmov fb_pixels, fb_pixels_next; \
3647 blend_blocks_average_combine_##texturing(pixels); \
3649 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3650 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3651 cmp fb_ptr_cmp, #28; \
3654 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3655 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3657 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3658 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3660 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3661 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3663 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3664 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3665 blend_blocks_average_mask_set_##mask_evaluate(); \
3666 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3669 subs num_blocks, num_blocks, #1; \
3673 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3674 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3676 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3677 blend_blocks_average_set_stp_bit_##texturing(); \
3678 blend_blocks_average_combine_##texturing(pixels_next); \
3680 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3681 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3682 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3684 ldmia sp!, { r4, pc }; \
3687 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3688 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3689 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3691 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3692 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3693 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3694 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3695 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3696 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3700 blend_blocks_average_builder(textured, off)
3701 blend_blocks_average_builder(untextured, off)
3702 blend_blocks_average_builder(textured, on)
3703 blend_blocks_average_builder(untextured, on)
3706 #define blend_blocks_add_mask_set_on() \
3707 vclt.s16 write_mask, fb_pixels, #0 \
3709 #define blend_blocks_add_mask_copy_on() \
3710 vorr.u16 draw_mask, draw_mask, write_mask \
3712 #define blend_blocks_add_mask_set_off() \
3714 #define blend_blocks_add_mask_copy_off() \
3717 #define blend_blocks_add_textured_builder(mask_evaluate) \
3720 function(blend_blocks_textured_add_##mask_evaluate) \
3721 stmdb sp!, { r4, r14 }; \
3722 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3723 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3725 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3726 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3728 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3731 vmov.u16 d128_0x7C1F, #0x7C00; \
3732 vmov.u16 d128_0x03E0, #0x0300; \
3733 vmov.u16 d128_0x83E0, #0x8000; \
3734 vorr.u16 d128_0x03E0, #0x00E0; \
3735 vorr.u16 d128_0x7C1F, #0x001F; \
3736 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3738 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3739 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3740 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3741 vclt.s16 blend_mask, pixels, #0; \
3742 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3743 blend_blocks_add_mask_set_##mask_evaluate(); \
3744 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3746 blend_blocks_add_mask_copy_##mask_evaluate(); \
3747 vorr.u16 pixels, pixels, msb_mask; \
3748 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3749 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3750 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3751 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3752 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3753 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3754 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3755 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3757 subs num_blocks, num_blocks, #1; \
3761 mov fb_ptr, fb_ptr_next; \
3763 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3765 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3766 vclt.s16 blend_mask, pixels, #0; \
3768 vorr.u16 pixels, pixels, msb_mask; \
3769 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3770 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3772 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3773 pld [ fb_ptr_next, #64 ]; \
3775 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3776 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3778 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3779 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3781 cmp fb_ptr_cmp, #28; \
3784 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3785 blend_blocks_add_mask_set_##mask_evaluate(); \
3786 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3787 blend_blocks_add_mask_copy_##mask_evaluate(); \
3788 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3789 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3790 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3793 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3794 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3795 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3796 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3797 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3799 subs num_blocks, num_blocks, #1; \
3803 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3804 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3805 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3807 ldmia sp!, { r4, pc }; \
3810 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3811 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3813 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3814 blend_blocks_add_mask_set_##mask_evaluate(); \
3815 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3816 blend_blocks_add_mask_copy_##mask_evaluate(); \
3817 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3821 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3824 function(blend_blocks_untextured_add_##mask_evaluate) \
3825 stmdb sp!, { r4, r14 }; \
3826 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3827 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3829 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3830 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3832 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3835 vmov.u16 d128_0x7C1F, #0x7C00; \
3836 vmov.u16 d128_0x03E0, #0x0300; \
3837 vorr.u16 d128_0x7C1F, #0x001F; \
3838 vorr.u16 d128_0x03E0, #0x00E0; \
3840 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3841 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3842 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3843 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3844 blend_blocks_add_mask_set_##mask_evaluate(); \
3845 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3847 blend_blocks_add_mask_copy_##mask_evaluate(); \
3848 vand.u16 pixels_g, pixels, d128_0x03E0; \
3849 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3850 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3851 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3852 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3853 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3854 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3856 subs num_blocks, num_blocks, #1; \
3860 mov fb_ptr, fb_ptr_next; \
3862 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3864 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3866 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3867 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3868 vand.u16 pixels_g, pixels, d128_0x03E0; \
3870 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3871 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3873 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3874 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3875 cmp fb_ptr_cmp, #28; \
3878 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3879 blend_blocks_add_mask_set_##mask_evaluate(); \
3880 blend_blocks_add_mask_copy_##mask_evaluate(); \
3881 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3882 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3883 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3886 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3887 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3888 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3889 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3890 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3892 subs num_blocks, num_blocks, #1; \
3896 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3897 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3898 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3899 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3901 ldmia sp!, { r4, pc }; \
3904 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3905 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3907 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3908 blend_blocks_add_mask_set_##mask_evaluate(); \
3909 blend_blocks_add_mask_copy_##mask_evaluate(); \
3910 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3914 blend_blocks_add_textured_builder(off)
3915 blend_blocks_add_textured_builder(on)
3916 blend_blocks_add_untextured_builder(off)
3917 blend_blocks_add_untextured_builder(on)
3919 #define blend_blocks_subtract_set_blend_mask_textured() \
3920 vclt.s16 blend_mask, pixels_next, #0 \
3922 #define blend_blocks_subtract_combine_textured() \
3923 vbif.u16 blend_pixels, pixels, blend_mask \
3925 #define blend_blocks_subtract_set_stb_textured() \
3926 vorr.u16 blend_pixels, #0x8000 \
3928 #define blend_blocks_subtract_msb_mask_textured() \
3929 vorr.u16 pixels, pixels_next, msb_mask \
3931 #define blend_blocks_subtract_set_blend_mask_untextured() \
3933 #define blend_blocks_subtract_combine_untextured() \
3935 #define blend_blocks_subtract_set_stb_untextured() \
3936 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3938 #define blend_blocks_subtract_msb_mask_untextured() \
3941 #define blend_blocks_subtract_mask_set_on() \
3942 vclt.s16 write_mask, fb_pixels, #0 \
3944 #define blend_blocks_subtract_mask_copy_on() \
3945 vorr.u16 draw_mask, draw_mask_next, write_mask \
3947 #define blend_blocks_subtract_mask_set_off() \
3949 #define blend_blocks_subtract_mask_copy_off() \
3950 vmov draw_mask, draw_mask_next \
3953 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
3956 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
3957 stmdb sp!, { r4, r14 }; \
3958 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3959 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3961 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3962 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3964 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3967 vmov.u16 d128_0x7C1F, #0x7C00; \
3968 vmov.u16 d128_0x03E0, #0x0300; \
3969 vorr.u16 d128_0x7C1F, #0x001F; \
3970 vorr.u16 d128_0x03E0, #0x00E0; \
3972 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3973 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3974 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3975 blend_blocks_subtract_set_blend_mask_##texturing(); \
3976 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3977 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3978 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3980 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3981 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3982 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3983 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3984 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3986 subs num_blocks, num_blocks, #1; \
3990 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
3991 mov fb_ptr, fb_ptr_next; \
3992 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3994 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3995 blend_blocks_subtract_msb_mask_##texturing(); \
3997 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3998 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3999 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4000 blend_blocks_subtract_set_stb_##texturing(); \
4001 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4002 blend_blocks_subtract_combine_##texturing(); \
4003 blend_blocks_subtract_set_blend_mask_##texturing(); \
4004 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4006 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4007 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4008 cmp fb_ptr_cmp, #28; \
4011 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4012 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4013 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4014 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4015 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4016 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4017 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4020 subs num_blocks, num_blocks, #1; \
4024 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4026 blend_blocks_subtract_msb_mask_##texturing(); \
4027 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4028 blend_blocks_subtract_set_stb_##texturing(); \
4029 blend_blocks_subtract_combine_##texturing(); \
4030 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4031 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4033 ldmia sp!, { r4, pc }; \
4036 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4037 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4038 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4039 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4040 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4041 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4042 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4046 blend_blocks_subtract_builder(textured, off)
4047 blend_blocks_subtract_builder(textured, on)
4048 blend_blocks_subtract_builder(untextured, off)
4049 blend_blocks_subtract_builder(untextured, on)
4052 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4055 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4056 stmdb sp!, { r4, r14 }; \
4057 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4058 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4060 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4061 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4063 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4066 vmov.u16 d128_0x7C1F, #0x7C00; \
4067 vmov.u16 d128_0x03E0, #0x0300; \
4068 vmov.u16 d128_0x1C07, #0x1C00; \
4069 vmov.u16 d128_0x00E0, #0x00E0; \
4070 vorr.u16 d128_0x7C1F, #0x001F; \
4071 vorr.u16 d128_0x03E0, #0x00E0; \
4072 vorr.u16 d128_0x1C07, #0x0007; \
4074 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4075 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4076 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4077 vclt.s16 blend_mask, pixels, #0; \
4078 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4079 blend_blocks_add_mask_set_##mask_evaluate(); \
4080 vshr.s16 pixels_fourth, pixels, #2; \
4081 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4083 blend_blocks_add_mask_copy_##mask_evaluate(); \
4084 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4085 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4086 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4087 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4088 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4089 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4090 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4092 subs num_blocks, num_blocks, #1; \
4096 mov fb_ptr, fb_ptr_next; \
4097 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4099 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4100 vbif.u16 blend_pixels, pixels, blend_mask; \
4102 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4103 vclt.s16 blend_mask, pixels, #0; \
4104 vshr.s16 pixels_fourth, pixels, #2; \
4105 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4106 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4108 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4109 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4111 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4112 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4113 cmp fb_ptr_cmp, #28; \
4116 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4117 blend_blocks_add_mask_set_##mask_evaluate(); \
4118 blend_blocks_add_mask_copy_##mask_evaluate(); \
4119 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4120 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4121 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4124 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4125 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4126 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4127 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4128 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4130 subs num_blocks, num_blocks, #1; \
4134 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4135 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4136 vbif.u16 blend_pixels, pixels, blend_mask; \
4137 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4138 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4140 ldmia sp!, { r4, pc }; \
4143 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4144 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4146 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4147 blend_blocks_add_mask_set_##mask_evaluate(); \
4148 blend_blocks_add_mask_copy_##mask_evaluate(); \
4149 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4154 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4157 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4158 stmdb sp!, { r4, r14 }; \
4159 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4160 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4162 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4163 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4165 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4168 vmov.u16 d128_0x7C1F, #0x7C00; \
4169 vmov.u16 d128_0x03E0, #0x0300; \
4170 vmov.u16 d128_0x1C07, #0x1C00; \
4171 vmov.u16 d128_0x00E0, #0x00E0; \
4172 vorr.u16 d128_0x7C1F, #0x001F; \
4173 vorr.u16 d128_0x03E0, #0x00E0; \
4174 vorr.u16 d128_0x1C07, #0x0007; \
4176 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4177 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4178 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4179 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4180 blend_blocks_add_mask_set_##mask_evaluate(); \
4181 vshr.s16 pixels_fourth, pixels, #2; \
4182 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4184 blend_blocks_add_mask_copy_##mask_evaluate(); \
4185 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4186 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4187 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4188 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4189 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4190 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4191 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4193 subs num_blocks, num_blocks, #1; \
4197 mov fb_ptr, fb_ptr_next; \
4198 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4200 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4202 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4203 vshr.s16 pixels_fourth, pixels, #2; \
4204 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4205 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4207 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4208 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4210 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4211 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4212 cmp fb_ptr_cmp, #28; \
4215 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4216 blend_blocks_add_mask_set_##mask_evaluate(); \
4217 blend_blocks_add_mask_copy_##mask_evaluate(); \
4218 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4219 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4220 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4223 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4224 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4225 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4226 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4227 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4229 subs num_blocks, num_blocks, #1; \
4233 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4234 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4235 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4236 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4238 ldmia sp!, { r4, pc }; \
4241 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4242 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4244 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4245 blend_blocks_add_mask_set_##mask_evaluate(); \
4246 blend_blocks_add_mask_copy_##mask_evaluate(); \
4247 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4251 blend_blocks_add_fourth_textured_builder(off)
4252 blend_blocks_add_fourth_textured_builder(on)
4253 blend_blocks_add_fourth_untextured_builder(off)
4254 blend_blocks_add_fourth_untextured_builder(on)
4256 // TODO: Optimize this more. Need a scene that actually uses it for
4261 function(blend_blocks_textured_unblended_on)
4262 stmdb sp!, { r4, r14 }
4263 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4264 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
4266 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4267 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
4269 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4272 ldr fb_ptr, [ pixel_ptr, #28 ]
4273 vld1.u16 { fb_pixels }, [ fb_ptr ]
4274 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4275 vclt.s16 write_mask, fb_pixels, #0
4276 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4278 subs num_blocks, num_blocks, #1
4282 vorr.u16 draw_mask, draw_mask, write_mask
4283 vbif.u16 fb_pixels, pixels, draw_mask
4284 vst1.u16 { fb_pixels }, [ fb_ptr ]
4286 ldr fb_ptr, [ pixel_ptr, #28 ]
4287 vld1.u16 { fb_pixels }, [ fb_ptr ]
4288 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4289 vclt.s16 write_mask, fb_pixels, #0
4290 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4292 subs num_blocks, num_blocks, #1
4296 vorr.u16 draw_mask, draw_mask, write_mask
4297 vbif.u16 fb_pixels, pixels, draw_mask
4298 vst1.u16 { fb_pixels }, [ fb_ptr ]
4300 ldmia sp!, { r4, pc }
4303 function(blend_blocks_textured_unblended_off)
4313 vld1.u32 { u_whole_8, v_whole_8 }, [ r1, :128 ], r3
4333 #define num_width r12
4343 function(render_block_fill_body)
4344 vdup.u16 colors_a, color
4347 vmov colors_b, colors_a
4348 sub pitch, pitch, width, lsl #1
4350 mov num_width, width
4353 vst1.u32 { colors_a, colors_b }, [ vram_ptr, :256 ]!
4355 subs num_width, num_width, #16
4358 add vram_ptr, vram_ptr, pitch
4359 mov num_width, width
4361 subs height, height, #1
4375 #undef dirty_textures_mask
4377 #undef current_texture_mask
4388 #define offset_u_right r10
4389 #define width_rounded r11
4390 #define height_rounded r12
4392 #define texture_offset_base r1
4393 #define tile_width r2
4394 #define tile_height r3
4395 #define num_blocks r4
4397 #define sub_tile_height r6
4399 #define texture_mask r8
4400 #define column_data r9
4401 #define texture_offset r10
4402 #define tiles_remaining r11
4403 #define fb_ptr_advance_column r12
4404 #define texture_block_ptr r14
4406 #define texture_page_ptr r3
4407 #define left_block_mask r4
4408 #define right_block_mask r5
4409 #define texture_mask_rev r10
4410 #define control_mask r11
4412 #define dirty_textures_mask r4
4414 #define current_texture_mask r6
4428 #define draw_masks_fb_ptrs q1
4430 #define draw_mask_fb_ptr_left d2
4431 #define draw_mask_fb_ptr_right d3
4433 #define draw_mask_fb_ptr_left_a d2
4434 #define draw_mask_fb_ptr_left_b d3
4435 #define draw_mask_fb_ptr_right_a d10
4436 #define draw_mask_fb_ptr_right_b d11
4437 #define draw_masks_fb_ptrs2 q5
4439 #define clut_low_a d4
4440 #define clut_low_b d5
4441 #define clut_high_a d6
4442 #define clut_high_b d7
4444 #define block_masks d8
4445 #define block_masks_shifted d9
4450 #define texels_low d12
4451 #define texels_high d13
4453 #define texels_wide_low d14
4454 #define texels_wide_high d15
4455 #define texels_wide q7
4458 setup_sprite_flush_blocks:
4461 stmdb sp!, { r0 - r3, r12, r14 }
4462 bl flush_render_block_buffer
4463 ldmia sp!, { r0 - r3, r12, r14 }
4467 add block, psx_gpu, #psx_gpu_blocks_offset
4471 setup_sprite_update_texture_4bpp_cache:
4472 stmdb sp!, { r0 - r3, r14 }
4473 bl update_texture_4bpp_cache
4474 ldmia sp!, { r0 - r3, pc }
4477 setup_sprite_update_texture_8bpp_cache:
4478 stmdb sp!, { r0 - r3, r14 }
4479 bl update_texture_8bpp_cache
4480 ldmia sp!, { r0 - r3, pc }
4483 #define setup_sprite_tiled_initialize_4bpp() \
4484 ldr dirty_textures_mask, \
4485 [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]; \
4486 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4488 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4489 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4491 tst current_texture_mask, dirty_textures_mask; \
4492 vuzp.u8 clut_a, clut_b; \
4494 blne setup_sprite_update_texture_4bpp_cache \
4496 #define setup_sprite_tiled_initialize_8bpp() \
4497 ldr dirty_textures_mask, \
4498 [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]; \
4499 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4501 tst current_texture_mask, dirty_textures_mask; \
4502 blne setup_sprite_update_texture_8bpp_cache \
4505 #define setup_sprite_block_count_single() \
4508 #define setup_sprite_block_count_double() \
4509 sub_tile_height, lsl #1 \
4511 #define setup_sprite_tile_add_blocks(type) \
4512 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4513 cmp num_blocks, #MAX_BLOCKS; \
4515 movgt num_blocks, setup_sprite_block_count_##type(); \
4516 blgt setup_sprite_flush_blocks \
4519 #define setup_sprite_tile_full_4bpp(edge) \
4520 setup_sprite_tile_add_blocks(double); \
4523 and texture_block_ptr, texture_offset, texture_mask; \
4524 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4527 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4528 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4530 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4531 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4533 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4534 add texture_block_ptr, texture_offset, #8; \
4536 and texture_block_ptr, texture_block_ptr, texture_mask; \
4537 add block, block, #40; \
4539 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4540 add fb_ptr, fb_ptr, #16; \
4542 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4543 add block, block, #24; \
4545 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4546 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4549 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4550 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4552 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4553 add block, block, #40; \
4555 add texture_offset, texture_offset, #0x10; \
4556 add fb_ptr, fb_ptr, #(2048 - 16); \
4558 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4559 add block, block, #24; \
4561 subs sub_tile_height, sub_tile_height, #1; \
4564 add texture_offset, texture_offset, #0xF00; \
4565 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4568 #define setup_sprite_tile_half_4bpp(edge) \
4569 setup_sprite_tile_add_blocks(single); \
4572 and texture_block_ptr, texture_offset, texture_mask; \
4573 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4576 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4577 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4579 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4580 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4582 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4583 add block, block, #40; \
4585 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4586 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4588 add block, block, #24; \
4589 add texture_offset, texture_offset, #0x10; \
4591 add fb_ptr, fb_ptr, #2048; \
4592 subs sub_tile_height, sub_tile_height, #1; \
4596 add texture_offset, texture_offset, #0xF00; \
4597 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4600 #define setup_sprite_tile_full_8bpp(edge) \
4601 setup_sprite_tile_add_blocks(double); \
4602 add block, block, #16; \
4605 and texture_block_ptr, texture_offset, texture_mask; \
4606 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4609 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4610 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4612 add texture_block_ptr, texture_offset, #8; \
4613 vst1.u32 { texels }, [ block, :64 ]; \
4615 and texture_block_ptr, texture_block_ptr, texture_mask; \
4616 add block, block, #24; \
4618 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4620 add fb_ptr, fb_ptr, #16; \
4621 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4623 add block, block, #40; \
4624 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4627 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4628 vst1.u32 { texels }, [ block, :64 ]; \
4629 add block, block, #24; \
4631 add texture_offset, texture_offset, #0x10; \
4632 add fb_ptr, fb_ptr, #(2048 - 16); \
4634 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4635 add block, block, #40; \
4637 subs sub_tile_height, sub_tile_height, #1; \
4640 sub block, block, #16; \
4641 add texture_offset, texture_offset, #0xF00; \
4642 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4645 #define setup_sprite_tile_half_8bpp(edge) \
4646 setup_sprite_tile_add_blocks(single); \
4647 add block, block, #16; \
4650 and texture_block_ptr, texture_offset, texture_mask; \
4651 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4654 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4655 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4657 vst1.u32 { texels }, [ block, :64 ]; \
4658 add block, block, #24; \
4660 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4661 add block, block, #40; \
4663 add texture_offset, texture_offset, #0x10; \
4664 add fb_ptr, fb_ptr, #2048; \
4666 subs sub_tile_height, sub_tile_height, #1; \
4669 sub block, block, #16; \
4670 add texture_offset, texture_offset, #0xF00; \
4671 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4674 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4675 add texture_offset, texture_offset_base, #8; \
4676 add fb_ptr, fb_ptr, #16 \
4678 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4679 mov texture_offset, texture_offset_base \
4681 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4682 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4684 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4685 mov texture_offset, texture_offset_base \
4687 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4688 sub fb_ptr, fb_ptr, #16 \
4690 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4692 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4693 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4695 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4698 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \
4700 mov sub_tile_height, column_data; \
4701 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4702 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4703 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4705 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \
4707 and sub_tile_height, column_data, #0xFF; \
4708 mov tiles_remaining, column_data, lsr #16; \
4709 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4710 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4712 subs tiles_remaining, tiles_remaining, #1; \
4716 mov sub_tile_height, #16; \
4717 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4718 subs tiles_remaining, tiles_remaining, #1; \
4722 uxtb sub_tile_height, column_data, ror #8; \
4723 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4724 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4727 #define setup_sprite_column_data_single() \
4728 mov column_data, height; \
4729 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] \
4731 #define setup_sprite_column_data_multi() \
4732 and height_rounded, height_rounded, #0xF; \
4733 rsb column_data, offset_v, #16; \
4735 add height_rounded, height_rounded, #1; \
4736 sub tile_height, tile_height, #1; \
4738 orr column_data, column_data, tile_height, lsl #16; \
4739 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]; \
4741 orr column_data, column_data, height_rounded, lsl #8 \
4743 #define setup_sprite_setup_left_draw_mask_fb_ptr() \
4744 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4745 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4747 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column() \
4748 mov fb_ptr_advance_column, #32; \
4749 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4751 sub fb_ptr_advance_column, height, lsl #11; \
4752 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4754 #define setup_sprite_setup_right_draw_mask_fb_ptr() \
4755 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4756 vdup.u8 draw_mask_fb_ptr_right, block_masks[5] \
4758 #define setup_sprite_tile_column_width_single(tm, multi_height, edge_mode, \
4760 setup_sprite_##tm##_single_##multi_height##_##edge_mode##_##edge##x4mode: \
4761 setup_sprite_column_data_##multi_height(); \
4762 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4763 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4764 setup_sprite_setup_left_draw_mask_fb_ptr##x4mode(); \
4766 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, tm, x4mode); \
4767 ldmia sp!, { r4 - r11, pc } \
4769 #define setup_sprite_tiled_advance_column() \
4770 add texture_offset_base, texture_offset_base, #0x100; \
4771 tst texture_offset_base, #0xF00; \
4772 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4774 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4775 right_mode, x4mode) \
4776 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode##x4mode:\
4777 setup_sprite_column_data_##multi_height(); \
4779 setup_sprite_setup_left_draw_mask_fb_ptr_advance_column##x4mode(); \
4781 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm, x4mode);\
4783 subs tile_width, tile_width, #2; \
4784 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4788 vmov.u8 draw_masks_fb_ptrs, #0; \
4789 vmov.u8 draw_masks_fb_ptrs2, #0; \
4792 setup_sprite_tiled_advance_column(); \
4793 setup_sprite_tile_column_height_##multi_height(full, none, tm, x4mode); \
4794 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4795 subs tile_width, tile_width, #1; \
4799 setup_sprite_setup_right_draw_mask_fb_ptr##x4mode(); \
4801 setup_sprite_tiled_advance_column(); \
4802 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm, x4mode);\
4803 ldmia sp!, { r4 - r11, pc } \
4806 #define setup_sprite_offset_u_adjust() \
4808 #define setup_sprite_get_left_block_mask() \
4809 and left_block_mask, left_block_mask, #0xFF \
4811 #define setup_sprite_compare_left_block_mask() \
4812 cmp left_block_mask, #0xFF \
4814 #define setup_sprite_get_right_block_mask() \
4815 uxtb right_block_mask, right_block_mask, ror #8 \
4817 #define setup_sprite_compare_right_block_mask() \
4818 cmp right_block_mask, #0xFF \
4823 #define fb_ptr2 column_data
4825 #define setup_sprite_offset_u_adjust_4x() \
4826 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4827 lsl offset_u_right, #1; \
4829 add offset_u_right, #1 \
4831 #define setup_sprite_get_left_block_mask_4x() \
4832 sxth left_block_mask, left_block_mask \
4834 #define setup_sprite_compare_left_block_mask_4x() \
4835 cmp left_block_mask, #0xFFFFFFFF \
4837 #define setup_sprite_get_right_block_mask_4x() \
4838 sxth right_block_mask, right_block_mask, ror #16 \
4840 #define setup_sprite_compare_right_block_mask_4x() \
4841 cmp right_block_mask, #0xFFFFFFFF \
4844 #define widen_texels_16bpp(texels_) \
4845 vmov texels_wide_low, texels_; \
4846 vmov texels_wide_high, texels_; \
4847 vzip.16 texels_wide_low, texels_wide_high \
4849 #define widen_texels_8bpp(texels_) \
4850 vmov texels_wide_low, texels_; \
4851 vmov texels_wide_high, texels_; \
4852 vzip.8 texels_wide_low, texels_wide_high \
4854 #define write_block_16bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4855 vst1.u32 { texels_ }, [ block_, :128 ]; \
4856 add block_, block_, #40; \
4858 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4859 vst1.u32 { draw_mask_fb_ptr_ }, [ block_, :64 ]; \
4860 add block_, block_, #24 \
4862 /* assumes 16-byte offset already added to block_ */
4863 #define write_block_8bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4864 vst1.u32 { texels_ }, [ block_, :64 ]; \
4865 add block_, block_, #24; \
4867 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4868 vst1.u32 { draw_mask_fb_ptr_ }, [ block_, :64 ]; \
4869 add block_, block_, #40 \
4871 #define do_texture_block_16bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4872 draw_mask_fb_ptr_b_) \
4873 widen_texels_16bpp(texels_low); \
4874 add fb_ptr_tmp, fb_ptr, #1024*2; \
4876 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr); \
4878 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4879 widen_texels_16bpp(texels_high); \
4881 add fb_ptr_tmp, fb_ptr, #8*2; \
4882 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4884 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4885 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4887 #define do_texture_block_8bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4888 draw_mask_fb_ptr_b_) \
4889 widen_texels_8bpp(texels); \
4890 add fb_ptr_tmp, fb_ptr, #1024*2; \
4892 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr); \
4893 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4895 add fb_ptr_tmp, fb_ptr, #8*2; \
4896 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4898 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4899 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4902 #define setup_sprite_tiled_initialize_4bpp_4x() \
4903 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4904 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4906 vuzp.u8 clut_a, clut_b \
4908 #define setup_sprite_tiled_initialize_8bpp_4x() \
4911 #define setup_sprite_block_count_single_4x() \
4912 sub_tile_height, lsl #2 \
4914 #define setup_sprite_block_count_double_4x() \
4915 sub_tile_height, lsl #(1+2) \
4917 #define setup_sprite_tile_full_4bpp_4x(edge) \
4918 setup_sprite_tile_add_blocks(double_4x); \
4919 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4922 and texture_block_ptr, texture_offset, texture_mask; \
4925 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4926 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4928 add texture_block_ptr, texture_offset, #8; \
4929 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4931 and texture_block_ptr, texture_block_ptr, texture_mask; \
4932 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4934 vzip.8 texels_low, texels_high; \
4935 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
4936 draw_mask_fb_ptr_left_b); \
4938 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4939 pld [ fb_ptr, #2048 ]; \
4941 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4942 add fb_ptr, fb_ptr, #16*2; \
4944 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4945 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4947 vzip.8 texels_low, texels_high; \
4948 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
4949 draw_mask_fb_ptr_right_b); \
4951 add texture_offset, texture_offset, #0x10; \
4952 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
4954 subs sub_tile_height, sub_tile_height, #1; \
4957 ldr column_data, [sp], #8; /* fb_ptr2 */ \
4958 add texture_offset, texture_offset, #0xF00; \
4959 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4962 #define setup_sprite_tile_half_4bpp_4x(edge) \
4963 setup_sprite_tile_add_blocks(single_4x); \
4964 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4967 and texture_block_ptr, texture_offset, texture_mask; \
4970 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4971 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4973 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4974 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4976 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4977 add texture_offset, texture_offset, #0x10; \
4979 vzip.8 texels_low, texels_high; \
4980 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
4981 draw_mask_fb_ptr_##edge##_b); \
4983 pld [ fb_ptr, #2048 ]; \
4984 add fb_ptr, fb_ptr, #2048 * 2; \
4986 subs sub_tile_height, sub_tile_height, #1; \
4989 ldr column_data, [sp], #8; /* fb_ptr2 */ \
4990 add texture_offset, texture_offset, #0xF00; \
4991 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4994 #define setup_sprite_tile_full_8bpp_4x(edge) \
4995 setup_sprite_tile_add_blocks(double_4x); \
4996 add block, block, #16; \
4997 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5000 and texture_block_ptr, texture_offset, texture_mask; \
5003 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5004 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
5006 add texture_block_ptr, texture_offset, #8; \
5007 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
5008 draw_mask_fb_ptr_left_b); \
5010 pld [ fb_ptr, #2048 ]; \
5011 and texture_block_ptr, texture_block_ptr, texture_mask; \
5013 add fb_ptr, fb_ptr, #16*2; \
5014 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5016 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
5018 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
5019 draw_mask_fb_ptr_right_b); \
5021 add texture_offset, texture_offset, #0x10; \
5022 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
5024 subs sub_tile_height, sub_tile_height, #1; \
5027 sub block, block, #16; \
5028 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5029 add texture_offset, texture_offset, #0xF00; \
5030 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
5033 #define setup_sprite_tile_half_8bpp_4x(edge) \
5034 setup_sprite_tile_add_blocks(single_4x); \
5035 add block, block, #16; \
5036 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5039 and texture_block_ptr, texture_offset, texture_mask; \
5042 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5043 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
5045 pld [ fb_ptr, #2048 ]; \
5046 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5047 draw_mask_fb_ptr_##edge##_b); \
5049 add texture_offset, texture_offset, #0x10; \
5050 add fb_ptr, fb_ptr, #2048 * 2; \
5052 subs sub_tile_height, sub_tile_height, #1; \
5055 sub block, block, #16; \
5056 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5057 add texture_offset, texture_offset, #0xF00; \
5058 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
5061 #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \
5062 add texture_offset, texture_offset_base, #8; \
5063 add fb_ptr, fb_ptr, #16 * 2 \
5065 #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \
5066 mov texture_offset, texture_offset_base \
5068 #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \
5069 setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \
5071 #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \
5072 mov texture_offset, texture_offset_base \
5074 #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \
5075 sub fb_ptr, fb_ptr, #16 * 2 \
5077 #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \
5079 #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \
5080 setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \
5082 #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \
5085 #define setup_sprite_setup_left_draw_mask_fb_ptr_4x() \
5086 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5087 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5088 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5089 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5091 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column_4x() \
5092 mov fb_ptr_advance_column, #32 * 2; \
5093 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5094 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5095 sub fb_ptr_advance_column, height, lsl #11 + 1; \
5096 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5097 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5099 #define setup_sprite_setup_right_draw_mask_fb_ptr_4x() \
5100 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[4]; \
5101 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[5]; \
5102 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[6]; \
5103 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[7] \
5111 // [ sp + 4 ]: width
5112 // [ sp + 8 ]: height
5113 // [ sp + 12 ]: color (unused)
5115 #define setup_sprite_tiled_builder(texture_mode, x4mode) \
5117 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \
5119 setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \
5121 setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \
5123 setup_sprite_tile_column_width_single(texture_mode, single, full, none, \
5125 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \
5127 setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \
5129 setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \
5131 setup_sprite_tile_column_width_single(texture_mode, single, half, right, \
5133 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \
5135 setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \
5137 setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \
5139 setup_sprite_tile_column_width_single(texture_mode, single, half, left, \
5141 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \
5143 setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \
5148 function(setup_sprite_##texture_mode##x4mode) \
5149 stmdb sp!, { r4 - r11, r14 }; \
5150 setup_sprite_tiled_initialize_##texture_mode##x4mode(); \
5152 ldr v, [ sp, #36 ]; \
5153 and offset_u, u, #0xF; \
5155 ldr width, [ sp, #40 ]; \
5156 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
5158 ldr height, [ sp, #44 ]; \
5159 add fb_ptr, fb_ptr, y, lsl #11; \
5161 add fb_ptr, fb_ptr, x, lsl #1; \
5162 and offset_v, v, #0xF; \
5164 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
5165 add width_rounded, offset_u, width; \
5167 add height_rounded, offset_v, height; \
5168 add width_rounded, width_rounded, #15; \
5170 add height_rounded, height_rounded, #15; \
5171 mov tile_width, width_rounded, lsr #4; \
5173 /* texture_offset_base = VH-VL-00-00 */\
5174 mov texture_offset_base, v, lsl #8; \
5175 and offset_u_right, width_rounded, #0xF; \
5177 /* texture_offset_base = VH-UH-UL-00 */\
5178 bfi texture_offset_base, u, #4, #8; \
5179 mov right_block_mask, #0xFFFFFFFE; \
5181 setup_sprite_offset_u_adjust##x4mode(); \
5183 /* texture_offset_base = VH-UH-VL-00 */\
5184 bfi texture_offset_base, v, #4, #4; \
5185 mov left_block_mask, #0xFFFFFFFF; \
5187 mov tile_height, height_rounded, lsr #4; \
5188 mvn left_block_mask, left_block_mask, lsl offset_u; \
5190 /* texture_mask = HH-HL-WH-WL */\
5191 ldrh texture_mask, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]; \
5192 mov right_block_mask, right_block_mask, lsl offset_u_right; \
5194 /* texture_mask_rev = WH-WL-HH-HL */\
5195 rev16 texture_mask_rev, texture_mask; \
5196 vmov block_masks, left_block_mask, right_block_mask; \
5198 /* texture_mask = HH-HL-HL-WL */\
5199 bfi texture_mask, texture_mask_rev, #4, #4; \
5200 /* texture_mask_rev = 00-00-00-WH */\
5201 mov texture_mask_rev, texture_mask_rev, lsr #12; \
5203 /* texture_mask = HH-WH-HL-WL */\
5204 bfi texture_mask, texture_mask_rev, #8, #4; \
5205 setup_sprite_get_left_block_mask##x4mode(); \
5207 mov control_mask, #0; \
5208 setup_sprite_compare_left_block_mask##x4mode(); \
5210 setup_sprite_get_right_block_mask##x4mode(); \
5211 orreq control_mask, control_mask, #0x4; \
5213 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
5214 setup_sprite_compare_right_block_mask##x4mode(); \
5216 orreq control_mask, control_mask, #0x8; \
5217 cmp tile_width, #1; \
5219 add block, psx_gpu, #psx_gpu_blocks_offset; \
5220 orreq control_mask, control_mask, #0x1; \
5222 cmp tile_height, #1; \
5223 add block, block, num_blocks, lsl #6; \
5225 orreq control_mask, control_mask, #0x2; \
5226 ldr pc, [ pc, control_mask, lsl #2 ]; \
5229 .word setup_sprite_##texture_mode##_multi_multi_full_full##x4mode; \
5230 .word setup_sprite_##texture_mode##_single_multi_full_none##x4mode; \
5231 .word setup_sprite_##texture_mode##_multi_single_full_full##x4mode; \
5232 .word setup_sprite_##texture_mode##_single_single_full_none##x4mode; \
5233 .word setup_sprite_##texture_mode##_multi_multi_half_full##x4mode; \
5234 .word setup_sprite_##texture_mode##_single_multi_half_right##x4mode; \
5235 .word setup_sprite_##texture_mode##_multi_single_half_full##x4mode; \
5236 .word setup_sprite_##texture_mode##_single_single_half_right##x4mode; \
5237 .word setup_sprite_##texture_mode##_multi_multi_full_half##x4mode; \
5238 .word setup_sprite_##texture_mode##_single_multi_half_left##x4mode; \
5239 .word setup_sprite_##texture_mode##_multi_single_full_half##x4mode; \
5240 .word setup_sprite_##texture_mode##_single_single_half_left##x4mode; \
5241 .word setup_sprite_##texture_mode##_multi_multi_half_half##x4mode; \
5243 .word setup_sprite_##texture_mode##_multi_single_half_half##x4mode; \
5246 setup_sprite_tiled_builder(4bpp,);
5247 setup_sprite_tiled_builder(8bpp,);
5249 #undef draw_mask_fb_ptr_left
5250 #undef draw_mask_fb_ptr_right
5252 setup_sprite_tiled_builder(4bpp, _4x);
5253 setup_sprite_tiled_builder(8bpp, _4x);
5261 #define block_ptr r0
5262 #define num_blocks r1
5264 #define texel_shift_mask r3
5265 #define block_pixels_a r4
5266 #define block_pixels_b r5
5275 #define texels_01 r6
5276 #define texels_23 r7
5277 #define texels_45 r8
5278 #define texels_67 r9
5280 function(texture_sprite_blocks_8bpp)
5281 stmdb sp!, { r4 - r11, r14 }
5282 movw texel_shift_mask, #(0xFF << 1)
5284 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5285 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
5287 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
5288 ldr block_pixels_a, [ block_ptr, #16 ]
5291 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
5292 ldr block_pixels_b, [ block_ptr, #20 ]
5294 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
5295 ldrh texel_0, [ clut_ptr, texel_0 ]
5297 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
5298 ldrh texel_1, [ clut_ptr, texel_1 ]
5300 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
5301 ldr block_pixels_a, [ block_ptr, #(64 + 16) ]
5303 ldrh texel_2, [ clut_ptr, texel_2 ]
5304 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
5306 ldrh texel_3, [ clut_ptr, texel_3 ]
5307 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
5309 ldrh texel_4, [ clut_ptr, texel_4 ]
5310 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
5312 ldrh texel_5, [ clut_ptr, texel_5 ]
5313 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
5315 ldrh texel_6, [ clut_ptr, texel_6 ]
5316 orr texels_01, texel_0, texel_1, lsl #16
5318 ldrh texel_7, [ clut_ptr, texel_7 ]
5319 orr texels_23, texel_2, texel_3, lsl #16
5321 orr texels_45, texel_4, texel_5, lsl #16
5322 str texels_01, [ block_ptr, #0 ]
5324 orr texels_67, texel_6, texel_7, lsl #16
5325 str texels_23, [ block_ptr, #4 ]
5327 subs num_blocks, num_blocks, #1
5328 str texels_45, [ block_ptr, #8 ]
5330 str texels_67, [ block_ptr, #12 ]
5331 add block_ptr, block_ptr, #64
5335 ldmia sp!, { r4 - r11, pc }
5338 #undef width_rounded
5341 #undef texture_offset
5344 #undef texels_wide_low
5345 #undef texels_wide_high
5356 #define left_offset r8
5357 #define width_rounded r9
5358 #define right_width r10
5360 #define block_width r11
5362 #define texture_offset_base r1
5363 #define texture_mask r2
5364 #define texture_page_ptr r3
5365 #define num_blocks r4
5368 #define texture_offset r8
5369 #define blocks_remaining r9
5371 #define fb_ptr_pitch r12
5372 #define texture_block_ptr r14
5374 #define texture_mask_width r2
5375 #define texture_mask_height r3
5376 #define left_mask_bits r4
5377 #define right_mask_bits r5
5381 #undef block_masks_shifted
5384 #define block_masks d0
5385 #define block_masks_shifted d1
5386 #define draw_mask_fb_ptr d2
5389 #define draw_mask_fb_ptr_a d2
5390 #define draw_mask_fb_ptr_b d3
5391 #define texels_low d4
5392 #define texels_high d5
5393 #define texels_wide_low d6
5394 #define texels_wide_high d7
5395 #define texels_wide q3
5398 setup_sprites_16bpp_flush:
5401 stmdb sp!, { r0 - r3, r12, r14 }
5402 bl flush_render_block_buffer
5403 ldmia sp!, { r0 - r3, r12, r14 }
5407 add block, psx_gpu, #psx_gpu_blocks_offset
5408 mov num_blocks, block_width
5412 function(setup_sprite_16bpp)
5413 stmdb sp!, { r4 - r11, r14 }
5414 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5417 add fb_ptr, fb_ptr, y, lsl #11
5419 ldr width, [ sp, #40 ]
5420 add fb_ptr, fb_ptr, x, lsl #1
5422 ldr height, [ sp, #44 ]
5423 and left_offset, u, #0x7
5425 add texture_offset_base, u, u
5426 add width_rounded, width, #7
5428 add texture_offset_base, v, lsl #11
5429 mov left_mask_bits, #0xFF
5431 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5432 add width_rounded, width_rounded, left_offset
5434 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5435 sub fb_ptr, fb_ptr, left_offset, lsl #1
5437 add texture_mask, texture_mask_width, texture_mask_width
5438 mov right_mask_bits, #0xFE
5440 and right_width, width_rounded, #0x7
5441 mvn left_mask_bits, left_mask_bits, lsl left_offset
5443 add texture_mask, texture_mask_height, lsl #11
5444 mov block_width, width_rounded, lsr #3
5446 mov right_mask_bits, right_mask_bits, lsl right_width
5447 movw fb_ptr_pitch, #(2048 + 16)
5449 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5450 vmov block_masks, left_mask_bits, right_mask_bits
5452 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5453 add block, psx_gpu, #psx_gpu_blocks_offset
5455 bic texture_offset_base, texture_offset_base, #0xF
5458 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5459 add block, block, num_blocks, lsl #6
5463 vext.32 block_masks_shifted, block_masks, block_masks, #1
5464 vorr.u32 block_masks, block_masks, block_masks_shifted
5465 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5468 add num_blocks, num_blocks, #1
5469 cmp num_blocks, #MAX_BLOCKS
5470 blgt setup_sprites_16bpp_flush
5472 and texture_block_ptr, texture_offset_base, texture_mask
5473 subs height, height, #1
5475 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5476 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5478 vst1.u32 { texels }, [ block, :128 ]
5479 add block, block, #40
5481 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5484 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5486 add block, block, #24
5487 add texture_offset_base, texture_offset_base, #2048
5488 add fb_ptr, fb_ptr, #2048
5489 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5492 ldmia sp!, { r4 - r11, pc }
5495 add num_blocks, num_blocks, block_width
5496 mov texture_offset, texture_offset_base
5498 cmp num_blocks, #MAX_BLOCKS
5499 blgt setup_sprites_16bpp_flush
5501 add texture_offset_base, texture_offset_base, #2048
5502 and texture_block_ptr, texture_offset, texture_mask
5504 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5505 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5507 vst1.u32 { texels }, [ block, :128 ]
5508 add block, block, #40
5510 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5511 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5514 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5515 subs blocks_remaining, block_width, #2
5517 add texture_offset, texture_offset, #16
5518 add fb_ptr, fb_ptr, #16
5520 vmov.u8 draw_mask_fb_ptr, #0
5522 add block, block, #24
5526 and texture_block_ptr, texture_offset, texture_mask
5527 subs blocks_remaining, blocks_remaining, #1
5529 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5530 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5532 vst1.u32 { texels }, [ block, :128 ]
5533 add block, block, #40
5535 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5538 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5540 add texture_offset, texture_offset, #16
5541 add fb_ptr, fb_ptr, #16
5543 add block, block, #24
5547 and texture_block_ptr, texture_offset, texture_mask
5548 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5550 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5551 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5553 vst1.u32 { texels }, [ block, :128 ]
5554 add block, block, #40
5556 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5557 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5559 add block, block, #24
5560 subs height, height, #1
5562 add fb_ptr, fb_ptr, fb_ptr_pitch
5563 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5567 ldmia sp!, { r4 - r11, pc }
5571 // FIXME: duplicate code with normal version :(
5572 #undef draw_mask_fb_ptr
5574 function(setup_sprite_16bpp_4x)
5575 stmdb sp!, { r4 - r11, r14 }
5576 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5579 add fb_ptr, fb_ptr, y, lsl #11
5581 ldr width, [ sp, #40 ]
5582 add fb_ptr, fb_ptr, x, lsl #1
5584 ldr height, [ sp, #44 ]
5585 and left_offset, u, #0x7
5587 add texture_offset_base, u, u
5588 add width_rounded, width, #7
5590 add texture_offset_base, v, lsl #11
5591 movw left_mask_bits, #0xFFFF
5593 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5594 add width_rounded, width_rounded, left_offset
5598 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5599 sub fb_ptr, fb_ptr, left_offset, lsl #1
5601 add texture_mask, texture_mask_width, texture_mask_width
5602 movw right_mask_bits, #0xFFFC
5604 and right_width, width_rounded, #0x7
5605 mvn left_mask_bits, left_mask_bits, lsl left_offset
5609 add texture_mask, texture_mask_height, lsl #11
5610 mov block_width, width_rounded, lsr #3
5612 mov right_mask_bits, right_mask_bits, lsl right_width
5613 movw fb_ptr_pitch, #(2048 + 16) * 2
5615 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4+1
5616 vmov block_masks, left_mask_bits, right_mask_bits
5618 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5619 add block, psx_gpu, #psx_gpu_blocks_offset
5621 bic texture_offset_base, texture_offset_base, #0xF
5624 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5625 add block, block, num_blocks, lsl #6
5630 vext.32 block_masks_shifted, block_masks, block_masks, #1
5631 vorr.u32 block_masks, block_masks, block_masks_shifted
5632 vdup.u8 draw_mask_fb_ptr_a, block_masks[0]
5633 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5636 add num_blocks, num_blocks, block_width
5637 cmp num_blocks, #MAX_BLOCKS
5638 blgt setup_sprites_16bpp_flush
5640 and texture_block_ptr, texture_offset_base, texture_mask
5641 subs height, height, #1
5643 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5644 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5646 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5648 add texture_offset_base, texture_offset_base, #2048
5649 add fb_ptr, fb_ptr, #2048*2
5650 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5653 ldmia sp!, { r4 - r11, pc }
5656 add num_blocks, num_blocks, block_width
5657 mov texture_offset, texture_offset_base
5659 vdup.u8 draw_mask_fb_ptr_a, block_masks[0] // left_mask_bits
5660 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5662 cmp num_blocks, #MAX_BLOCKS
5663 blgt setup_sprites_16bpp_flush
5665 add texture_offset_base, texture_offset_base, #2048
5666 and texture_block_ptr, texture_offset, texture_mask
5668 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5669 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5671 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5673 subs blocks_remaining, block_width, #2*4
5674 add texture_offset, texture_offset, #16
5676 vmov.u8 draw_mask_fb_ptr_a, #0
5677 vmov.u8 draw_mask_fb_ptr_b, #0
5679 add fb_ptr, fb_ptr, #16*2
5683 and texture_block_ptr, texture_offset, texture_mask
5684 subs blocks_remaining, blocks_remaining, #4
5686 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5687 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5689 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5690 add texture_offset, texture_offset, #16
5692 add fb_ptr, fb_ptr, #16*2
5696 vdup.u8 draw_mask_fb_ptr_a, block_masks[4] // right_mask_bits
5697 vdup.u8 draw_mask_fb_ptr_b, block_masks[5]
5699 and texture_block_ptr, texture_offset, texture_mask
5700 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5702 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5704 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5705 subs height, height, #1
5707 add fb_ptr, fb_ptr, fb_ptr_pitch
5708 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5712 ldmia sp!, { r4 - r11, pc }
5717 #undef right_mask_bits
5720 #undef blocks_remaining
5730 #define right_width r5
5731 #define right_mask_bits r6
5735 #define fb_ptr_pitch r12
5737 // referenced by setup_sprites_16bpp_flush
5738 #define num_blocks r4
5740 #define block_width r11
5745 #define blocks_remaining r6
5748 #define right_mask q1
5749 #define test_mask q2
5750 #define draw_mask q2
5751 #define draw_mask_bits_fb_ptr d6
5756 function(setup_sprite_untextured)
5757 ldrh r12, [ psx_gpu, #psx_gpu_render_state_offset ]
5758 tst r12, #(RENDER_STATE_MASK_EVALUATE | RENDER_FLAGS_MODULATE_TEXELS \
5759 | RENDER_FLAGS_BLEND)
5760 beq setup_sprite_untextured_simple
5762 stmdb sp!, { r4 - r11, r14 }
5764 ldr width, [ sp, #40 ]
5765 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5767 ldr height, [ sp, #44 ]
5768 add fb_ptr, fb_ptr, y, lsl #11
5770 add fb_ptr, fb_ptr, x, lsl #1
5771 sub right_width, width, #1
5773 ldr color, [ sp, #48 ]
5776 add block_width, width, #7
5780 mov right_mask_bits, #0xff
5782 sub fb_ptr_pitch, block_width, #1
5783 lsl right_mask_bits, right_width
5785 lsl fb_ptr_pitch, #3+1
5786 ubfx color_r, color, #3, #5
5788 rsb fb_ptr_pitch, #1024*2
5789 ubfx color_g, color, #11, #5
5791 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
5792 ubfx color_b, color, #19, #5
5794 vdup.u16 right_mask, right_mask_bits
5795 orr color, color_r, color_b, lsl #10
5797 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5798 orr color, color, color_g, lsl #5
5800 vtst.u16 right_mask, right_mask, test_mask
5801 add block, psx_gpu, #psx_gpu_blocks_offset
5803 vdup.u16 colors, color
5804 add block, block, num_blocks, lsl #6
5807 setup_sprite_untextured_height_loop:
5808 add num_blocks, block_width
5809 sub blocks_remaining, block_width, #1
5811 cmp num_blocks, #MAX_BLOCKS
5812 blgt setup_sprites_16bpp_flush
5814 cmp blocks_remaining, #0
5817 vmov.u8 draw_mask, #0 /* zero_mask */
5818 vmov.u8 draw_mask_bits_fb_ptr, #0
5821 vst1.u32 { draw_mask }, [ block, :128 ]!
5822 subs blocks_remaining, #1
5824 vst1.u32 { colors }, [ block, :128 ]
5825 add block, block, #24
5827 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5828 vst1.u32 { draw_mask_bits_fb_ptr }, [ block, :64 ]
5830 add block, block, #24
5835 vst1.u32 { right_mask }, [ block, :128 ]!
5838 vst1.u32 { colors }, [ block, :128 ]
5839 add block, block, #24
5841 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5842 vst1.u32 { draw_mask_bits_fb_ptr }, [ block, :64 ]
5844 add block, block, #24
5845 add fb_ptr, fb_ptr_pitch
5847 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5848 bgt setup_sprite_untextured_height_loop
5850 ldmia sp!, { r4 - r11, pc }
5854 #undef texture_page_ptr
5856 #undef dirty_textures_mask
5857 #undef current_texture_mask
5860 #define current_texture_page r1
5861 #define texture_page_ptr r2
5862 #define vram_ptr_a r3
5863 #define current_texture_page_x r12
5864 #define current_texture_page_y r4
5865 #define dirty_textures_mask r5
5869 #define current_texture_mask r9
5871 #define vram_ptr_b r11
5873 #define texel_block_a d0
5874 #define texel_block_b d1
5875 #define texel_block_expanded_a q1
5876 #define texel_block_expanded_b q2
5877 #define texel_block_expanded_ab q2
5878 #define texel_block_expanded_c q3
5879 #define texel_block_expanded_d q4
5880 #define texel_block_expanded_cd q3
5882 function(update_texture_4bpp_cache)
5883 stmdb sp!, { r4 - r11, r14 }
5886 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5888 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5889 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5891 and current_texture_page_x, current_texture_page, #0xF
5892 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
5894 mov current_texture_page_y, current_texture_page, lsr #4
5895 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5897 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5900 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5901 bic dirty_textures_mask, current_texture_mask
5904 str dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5909 add vram_ptr_b, vram_ptr_a, #2048
5912 vld1.u32 { texel_block_a }, [ vram_ptr_a, :64 ], c_4096
5913 vld1.u32 { texel_block_b }, [ vram_ptr_b, :64 ], c_4096
5915 vmovl.u8 texel_block_expanded_a, texel_block_a
5916 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5917 vmovl.u8 texel_block_expanded_c, texel_block_b
5918 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5920 vbic.u16 texel_block_expanded_a, #0x00F0
5921 vbic.u16 texel_block_expanded_b, #0x00F0
5922 vbic.u16 texel_block_expanded_c, #0x00F0
5923 vbic.u16 texel_block_expanded_d, #0x00F0
5925 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
5926 texel_block_expanded_b
5927 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
5928 texel_block_expanded_d
5930 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
5931 [ texture_page_ptr, :256 ]!
5933 subs sub_y, sub_y, #1
5937 add vram_ptr_a, vram_ptr_a, #8
5938 add vram_ptr_b, vram_ptr_b, #8
5940 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5941 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5943 subs tile_x, tile_x, #1
5947 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5948 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5950 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5951 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5953 subs tile_y, tile_y, #1
5957 ldmia sp!, { r4 - r11, pc }
5960 #undef current_texture_page
5963 #define texture_page r1
5964 #define texture_page_ptr r2
5965 #define vram_ptr_a r3
5966 #define texture_page_x r12
5967 #define texture_page_y r4
5968 #define current_texture_page r5
5973 #define vram_ptr_b r11
5985 function(update_texture_8bpp_cache_slice)
5986 stmdb sp!, { r4 - r11, r14 }
5989 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5990 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5992 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5995 and texture_page_x, texture_page, #0xF
5996 mov texture_page_y, texture_page, lsr #4
5998 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
6001 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
6002 eor current_texture_page, current_texture_page, texture_page
6004 ands current_texture_page, current_texture_page, #0x1
6007 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6010 add vram_ptr_b, vram_ptr_a, #2048
6013 vld1.u32 { texels_a }, [ vram_ptr_a, :128 ], c_4096
6014 vld1.u32 { texels_b }, [ vram_ptr_b, :128 ], c_4096
6015 vld1.u32 { texels_c }, [ vram_ptr_a, :128 ], c_4096
6016 vld1.u32 { texels_d }, [ vram_ptr_b, :128 ], c_4096
6018 vst1.u32 { texels_a, texels_b }, [ texture_page_ptr, :256 ]!
6019 vst1.u32 { texels_c, texels_d }, [ texture_page_ptr, :256 ]!
6021 subs sub_y, sub_y, #1
6026 add vram_ptr_a, vram_ptr_a, #16
6027 add vram_ptr_b, vram_ptr_b, #16
6029 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
6030 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
6032 subs tile_x, tile_x, #1
6037 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
6038 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
6040 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
6041 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
6043 subs tile_y, tile_y, #1
6044 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6049 ldmia sp!, { r4 - r11, pc }
6052 /* void scale2x_tiles8(void *dst, const void *src, int w8, int h) */
6053 function(scale2x_tiles8)
6057 add r12, r0, #1024*2
6061 vld1.u16 { q0 }, [ r1, :128 ]!
6062 vld1.u16 { q2 }, [ r1, :128 ]!
6068 vst1.u16 { q0, q1 }, [ r0, :128 ]!
6069 vst1.u16 { q0, q1 }, [ r12, :128 ]!
6071 vst1.u16 { q2, q3 }, [ r0, :128 ]!
6072 vst1.u16 { q2, q3 }, [ r12, :128 ]!
6079 sub r0, r2, lsl #4+1
6081 add r12, r0, #1024*2
6087 // vim:filetype=armasm