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
1178 strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1181 setup_spans_set_x4_alternate_no(none, down)
1182 subs height_minor_b, height_minor_b, #4
1186 setup_spans_epilogue()
1189 setup_spans_up_down_load_edge_set_b()
1190 setup_spans_prologue_b()
1195 #undef span_uvrg_offset
1196 #undef span_edge_data
1197 #undef span_b_offset
1202 #define num_spans r1
1203 #define span_uvrg_offset r2
1204 #define span_edge_data r3
1205 #define span_b_offset r4
1207 #define span_num_blocks r6
1211 #define dither_offset_ptr r10
1212 #define block_ptr_a r11
1214 #define num_blocks r14
1216 #define uvrg_dx_ptr r2
1217 #define texture_mask_ptr r3
1218 #define dither_shift r8
1219 #define dither_row r10
1224 #define block_ptr_b r10
1226 #define block_span_ptr r10
1227 #define right_mask r8
1246 #define b_whole_8 d14
1247 #define fb_mask_ptrs d15
1260 #define u_whole_low d16
1261 #define u_whole_high d17
1262 #define v_whole_low d18
1263 #define v_whole_high d19
1264 #define r_whole_low d20
1265 #define r_whole_high d21
1266 #define g_whole_low d22
1267 #define g_whole_high d23
1268 #define b_whole_low d24
1269 #define b_whole_high d25
1274 #define u_whole_8 d26
1275 #define v_whole_8 d27
1276 #define u_whole_8b d24
1277 #define r_whole_8 d24
1278 #define g_whole_8 d25
1280 #define uv_whole_8 q13
1281 #define uv_whole_8b q14
1283 #define dither_offsets q14
1284 #define texture_mask q15
1285 #define texture_mask_u d30
1286 #define texture_mask_v d31
1288 #define dither_offsets_short d28
1292 #define block_span q10
1297 #define draw_mask q1
1298 #define draw_mask_edge q13
1299 #define test_mask q0
1305 #define setup_blocks_texture_swizzled() \
1306 vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \
1307 vsli.u8 u_whole_8, v_whole_8, #4; \
1308 vsri.u8 v_whole_8, u_whole_8b, #4 \
1310 #define setup_blocks_texture_unswizzled() \
1313 #define setup_blocks_shaded_textured_builder(swizzling) \
1316 function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \
1317 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1318 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1320 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1321 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1323 cmp num_spans, #0; \
1326 stmdb sp!, { r4 - r11, r14 }; \
1327 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1329 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
1330 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1332 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1333 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1335 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1336 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1338 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
1339 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1341 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1344 vmov.u8 fb_mask_ptrs, #0; \
1346 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1347 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1349 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1350 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
1352 cmp span_num_blocks, #0; \
1355 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1356 add num_blocks, span_num_blocks, num_blocks; \
1358 cmp num_blocks, #MAX_BLOCKS; \
1362 ldr b, [ span_b_offset ]; \
1363 add fb_ptr, fb_ptr, y, lsl #11; \
1365 vdup.u32 v_left_x, left_x; \
1368 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1369 add fb_ptr, fb_ptr, left_x, lsl #1; \
1371 mla b, b_dx, left_x, b; \
1372 and dither_shift, left_x, #0x03; \
1374 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1375 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1377 mov dither_shift, dither_shift, lsl #3; \
1378 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1381 subs span_num_blocks, span_num_blocks, #1; \
1383 mov dither_row, dither_row, ror dither_shift; \
1384 mov b_dx4, b_dx, lsl #2; \
1386 vdup.u32 dither_offsets_short, dither_row; \
1387 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1389 vdup.u32 b_block, b; \
1390 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1392 vdup.u32 u_block, uv[0]; \
1393 mov b_dx8, b_dx, lsl #3; \
1395 vdup.u32 v_block, uv[1]; \
1396 vdup.u32 r_block, rg[0]; \
1397 vdup.u32 g_block, rg[1]; \
1399 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1401 vadd.u32 u_block, u_block, block_span; \
1402 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1404 vadd.u32 v_block, v_block, block_span; \
1405 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1407 vadd.u32 r_block, r_block, block_span; \
1408 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1410 vadd.u32 g_block, g_block, block_span; \
1411 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
1413 vadd.u32 b_block, b_block, block_span; \
1414 add block_ptr_b, block_ptr_a, #16; \
1416 vshrn.u32 u_whole_low, u_block, #16; \
1417 vshrn.u32 v_whole_low, v_block, #16; \
1418 vshrn.u32 r_whole_low, r_block, #16; \
1419 vshrn.u32 g_whole_low, g_block, #16; \
1421 vdup.u32 dx4, uv_dx4[0]; \
1422 vshrn.u32 b_whole_low, b_block, #16; \
1424 vaddhn.u32 u_whole_high, u_block, dx4; \
1425 vdup.u32 dx4, uv_dx4[1]; \
1427 vaddhn.u32 v_whole_high, v_block, dx4; \
1428 vdup.u32 dx4, rg_dx4[0]; \
1430 vaddhn.u32 r_whole_high, r_block, dx4; \
1431 vdup.u32 dx4, rg_dx4[1]; \
1433 vaddhn.u32 g_whole_high, g_block, dx4; \
1434 vdup.u32 dx4, b_dx4; \
1436 vaddhn.u32 b_whole_high, b_block, dx4; \
1437 vdup.u32 dx8, uv_dx8[0]; \
1439 vadd.u32 u_block, u_block, dx8; \
1440 vdup.u32 dx8, uv_dx8[1]; \
1442 vadd.u32 v_block, v_block, dx8; \
1443 vdup.u32 dx8, rg_dx8[0]; \
1445 vadd.u32 r_block, r_block, dx8; \
1446 vdup.u32 dx8, rg_dx8[1]; \
1448 vadd.u32 g_block, g_block, dx8; \
1449 vdup.u32 dx8, b_dx8; \
1451 vadd.u32 b_block, b_block, dx8; \
1452 vmovn.u16 u_whole_8, u_whole; \
1454 vmovn.u16 v_whole_8, v_whole; \
1456 vmovn.u16 b_whole_8, b_whole; \
1458 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1460 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1461 setup_blocks_texture_##swizzling(); \
1463 vmovn.u16 r_whole_8, r_whole; \
1467 vmovn.u16 g_whole_8, g_whole; \
1468 vshrn.u32 u_whole_low, u_block, #16; \
1470 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1471 vshrn.u32 v_whole_low, v_block, #16; \
1473 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1474 vshrn.u32 r_whole_low, r_block, #16; \
1476 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1477 vshrn.u32 g_whole_low, g_block, #16; \
1479 vdup.u32 dx4, uv_dx4[0]; \
1480 vshrn.u32 b_whole_low, b_block, #16; \
1482 vaddhn.u32 u_whole_high, u_block, dx4; \
1483 vdup.u32 dx4, uv_dx4[1]; \
1485 vaddhn.u32 v_whole_high, v_block, dx4; \
1486 vdup.u32 dx4, rg_dx4[0]; \
1488 vaddhn.u32 r_whole_high, r_block, dx4; \
1489 vdup.u32 dx4, rg_dx4[1]; \
1491 vaddhn.u32 g_whole_high, g_block, dx4; \
1492 vdup.u32 dx4, b_dx4; \
1494 vaddhn.u32 b_whole_high, b_block, dx4; \
1495 vdup.u32 dx8, uv_dx8[0]; \
1497 vadd.u32 u_block, u_block, dx8; \
1498 vdup.u32 dx8, uv_dx8[1]; \
1500 vadd.u32 v_block, v_block, dx8; \
1501 vdup.u32 dx8, rg_dx8[0]; \
1503 vadd.u32 r_block, r_block, dx8; \
1504 vdup.u32 dx8, rg_dx8[1]; \
1506 vadd.u32 g_block, g_block, dx8; \
1507 vdup.u32 dx8, b_dx8; \
1509 vadd.u32 b_block, b_block, dx8; \
1510 vmovn.u16 u_whole_8, u_whole; \
1512 add fb_ptr, fb_ptr, #16; \
1513 vmovn.u16 v_whole_8, v_whole; \
1515 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1516 vmovn.u16 b_whole_8, b_whole; \
1520 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1521 subs span_num_blocks, span_num_blocks, #1; \
1523 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1524 setup_blocks_texture_##swizzling(); \
1526 vmovn.u16 r_whole_8, r_whole; \
1530 vmovn.u16 g_whole_8, g_whole; \
1531 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1533 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1534 vdup.u8 draw_mask, right_mask; \
1536 vmov.u32 fb_mask_ptrs[0], right_mask; \
1537 vtst.u16 draw_mask, draw_mask, test_mask; \
1538 vzip.u8 u_whole_8, v_whole_8; \
1540 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1541 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1542 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1543 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1544 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1547 add span_uvrg_offset, span_uvrg_offset, #16; \
1548 add span_b_offset, span_b_offset, #4; \
1550 add span_edge_data, span_edge_data, #8; \
1551 subs num_spans, num_spans, #1; \
1553 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1556 ldmia sp!, { r4 - r11, pc }; \
1559 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1560 vpush { texture_mask }; \
1561 vpush { uvrg_dx4 }; \
1563 stmdb sp!, { r0 - r3, r12, r14 }; \
1564 bl flush_render_block_buffer; \
1565 ldmia sp!, { r0 - r3, r12, r14 }; \
1567 vpop { uvrg_dx4 }; \
1568 vpop { texture_mask }; \
1570 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1571 vmov.u8 fb_mask_ptrs, #0; \
1573 mov num_blocks, span_num_blocks; \
1574 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1578 setup_blocks_shaded_textured_builder(swizzled)
1579 setup_blocks_shaded_textured_builder(unswizzled)
1582 #define setup_blocks_unshaded_textured_builder(swizzling) \
1585 function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \
1586 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1587 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1589 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1590 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1592 cmp num_spans, #0; \
1595 stmdb sp!, { r4 - r11, r14 }; \
1596 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1598 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1600 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1601 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1603 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1604 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1606 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1608 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1611 vmov.u8 fb_mask_ptrs, #0; \
1613 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1614 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1616 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1617 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
1619 cmp span_num_blocks, #0; \
1622 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1623 add num_blocks, span_num_blocks, num_blocks; \
1625 cmp num_blocks, #MAX_BLOCKS; \
1629 add fb_ptr, fb_ptr, y, lsl #11; \
1631 vdup.u32 v_left_x, left_x; \
1634 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1635 add fb_ptr, fb_ptr, left_x, lsl #1; \
1637 and dither_shift, left_x, #0x03; \
1639 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1640 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1642 mov dither_shift, dither_shift, lsl #3; \
1643 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1646 subs span_num_blocks, span_num_blocks, #1; \
1648 mov dither_row, dither_row, ror dither_shift; \
1650 vdup.u32 dither_offsets_short, dither_row; \
1651 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1653 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1655 vdup.u32 u_block, uv[0]; \
1657 vdup.u32 v_block, uv[1]; \
1658 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1660 vadd.u32 u_block, u_block, block_span; \
1661 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1663 vadd.u32 v_block, v_block, block_span; \
1664 add block_ptr_b, block_ptr_a, #16; \
1666 vshrn.u32 u_whole_low, u_block, #16; \
1667 vshrn.u32 v_whole_low, v_block, #16; \
1669 vdup.u32 dx4, uv_dx4[0]; \
1671 vaddhn.u32 u_whole_high, u_block, dx4; \
1672 vdup.u32 dx4, uv_dx4[1]; \
1674 vaddhn.u32 v_whole_high, v_block, dx4; \
1675 vdup.u32 dx8, uv_dx8[0]; \
1677 vadd.u32 u_block, u_block, dx8; \
1678 vdup.u32 dx8, uv_dx8[1]; \
1680 vadd.u32 v_block, v_block, dx8; \
1681 vmovn.u16 u_whole_8, u_whole; \
1683 vmovn.u16 v_whole_8, v_whole; \
1686 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1688 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1689 setup_blocks_texture_##swizzling(); \
1694 vshrn.u32 u_whole_low, u_block, #16; \
1696 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1697 vshrn.u32 v_whole_low, v_block, #16; \
1699 add block_ptr_b, block_ptr_b, #32; \
1700 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1702 vdup.u32 dx4, uv_dx4[0]; \
1703 vaddhn.u32 u_whole_high, u_block, dx4; \
1704 vdup.u32 dx4, uv_dx4[1]; \
1706 vaddhn.u32 v_whole_high, v_block, dx4; \
1707 vdup.u32 dx8, uv_dx8[0]; \
1709 vadd.u32 u_block, u_block, dx8; \
1710 vdup.u32 dx8, uv_dx8[1]; \
1712 vadd.u32 v_block, v_block, dx8; \
1713 vmovn.u16 u_whole_8, u_whole; \
1715 add fb_ptr, fb_ptr, #16; \
1716 vmovn.u16 v_whole_8, v_whole; \
1718 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1721 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1722 subs span_num_blocks, span_num_blocks, #1; \
1724 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1725 setup_blocks_texture_##swizzling(); \
1730 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1732 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1733 vdup.u8 draw_mask, right_mask; \
1735 vmov.u32 fb_mask_ptrs[0], right_mask; \
1736 vtst.u16 draw_mask, draw_mask, test_mask; \
1737 vzip.u8 u_whole_8, v_whole_8; \
1739 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1740 add block_ptr_b, block_ptr_b, #32; \
1741 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1742 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1743 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1746 add span_uvrg_offset, span_uvrg_offset, #16; \
1747 add span_edge_data, span_edge_data, #8; \
1748 subs num_spans, num_spans, #1; \
1750 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1753 ldmia sp!, { r4 - r11, pc }; \
1756 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1757 vpush { texture_mask }; \
1758 vpush { uvrg_dx4 }; \
1760 stmdb sp!, { r0 - r3, r12, r14 }; \
1761 bl flush_render_block_buffer; \
1762 ldmia sp!, { r0 - r3, r12, r14 }; \
1764 vpop { uvrg_dx4 }; \
1765 vpop { texture_mask }; \
1767 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1768 vmov.u8 fb_mask_ptrs, #0; \
1770 mov num_blocks, span_num_blocks; \
1771 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1775 setup_blocks_unshaded_textured_builder(swizzled)
1776 setup_blocks_unshaded_textured_builder(unswizzled)
1781 function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect)
1782 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1783 veor.u32 draw_mask, draw_mask, draw_mask
1788 stmdb sp!, { r4 - r11, r14 }
1789 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1791 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1793 ubfx color_r, color, #3, #5
1794 ubfx color_g, color, #11, #5
1795 ubfx color_b, color, #19, #5
1797 orr color, color_r, color_b, lsl #10
1798 orr color, color, color_g, lsl #5
1800 vdup.u16 colors, color
1802 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1803 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1805 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1806 add block_ptr_a, block_ptr_a, num_blocks, lsl #6
1809 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1810 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1812 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
1814 cmp span_num_blocks, #0
1817 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1818 add num_blocks, span_num_blocks, num_blocks
1820 cmp num_blocks, #MAX_BLOCKS
1824 add fb_ptr, fb_ptr, y, lsl #11
1827 add fb_ptr, fb_ptr, left_x, lsl #1
1830 subs span_num_blocks, span_num_blocks, #1
1832 add block_ptr_b, block_ptr_a, #16
1835 vmov.u32 fb_mask_ptrs[1], fb_ptr
1839 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_32
1840 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1841 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1843 add fb_ptr, fb_ptr, #16
1844 add block_ptr_b, block_ptr_b, #32
1848 vmov.u32 fb_mask_ptrs[1], fb_ptr
1849 subs span_num_blocks, span_num_blocks, #1
1854 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1856 vdup.u8 draw_mask_edge, right_mask
1857 vtst.u16 draw_mask_edge, draw_mask_edge, test_mask
1859 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1860 vst1.u32 { draw_mask_edge }, [ block_ptr_a, :128 ], c_32
1861 add block_ptr_b, block_ptr_b, #32
1862 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1865 add span_edge_data, span_edge_data, #8
1866 subs num_spans, num_spans, #1
1868 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1871 ldmia sp!, { r4 - r11, pc }
1876 stmdb sp!, { r0 - r3, r12, r14 }
1877 bl flush_render_block_buffer
1878 ldmia sp!, { r0 - r3, r12, r14 }
1882 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1883 veor.u32 draw_mask, draw_mask, draw_mask
1885 mov num_blocks, span_num_blocks
1886 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1890 #define mask_msb_scalar r14
1892 #define msb_mask q15
1894 #define pixels_low d16
1896 #define msb_mask_low d30
1897 #define msb_mask_high d31
1902 function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct)
1903 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1908 stmdb sp!, { r4 - r11, r14 }
1910 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1912 ubfx color_r, color, #3, #5
1913 ubfx color_g, color, #11, #5
1915 ldrh mask_msb_scalar, [ psx_gpu, #psx_gpu_mask_msb_offset ]
1916 ubfx color_b, color, #19, #5
1918 orr color, color_r, color_b, lsl #10
1919 orr color, color, color_g, lsl #5
1920 orr color, color, mask_msb_scalar
1922 vdup.u16 colors, color
1924 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1925 orr color, color, lsl #16
1929 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1930 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1932 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
1934 cmp span_num_blocks, #0
1937 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1939 add fb_ptr, fb_ptr, y, lsl #11
1940 subs span_num_blocks, span_num_blocks, #1
1942 add fb_ptr, fb_ptr, left_x, lsl #1
1946 vst1.u32 { colors }, [ fb_ptr ]!
1947 subs span_num_blocks, span_num_blocks, #1
1952 ldrb right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1954 cmp right_mask, #0x0
1957 tst right_mask, #0xF
1958 streq color, [ fb_ptr ], #4
1959 moveq right_mask, right_mask, lsr #4
1960 streq color, [ fb_ptr ], #4
1962 tst right_mask, #0x3
1963 streq color, [ fb_ptr ], #4
1964 moveq right_mask, right_mask, lsr #2
1966 tst right_mask, #0x1
1967 streqh color, [ fb_ptr ]
1970 add span_edge_data, span_edge_data, #8
1971 subs num_spans, num_spans, #1
1974 ldmia sp!, { r4 - r11, pc }
1977 vst1.u32 { colors }, [ fb_ptr ]
1984 #define rg_dx_ptr r2
2002 #undef dither_offsets
2022 #define r_whole_low d6
2023 #define r_whole_high d7
2024 #define g_whole_low d8
2025 #define g_whole_high d9
2026 #define b_whole_low d10
2027 #define b_whole_high d11
2029 #define gb_whole_8 q6
2031 #define g_whole_8 d12
2032 #define b_whole_8 d13
2034 #define r_whole_8 d14
2046 #define block_span q5
2054 #define d128_0x7 q13
2058 #define dither_offsets q14
2059 #define draw_mask q15
2061 #define dither_offsets_low d28
2064 #define test_mask q10
2067 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2068 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2069 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2071 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2072 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2073 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2075 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2077 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2080 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2083 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2084 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2085 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2087 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2089 cmp num_spans, #0; \
2092 stmdb sp!, { r4 - r11, r14 }; \
2093 vshl.u32 rg_dx4, rg_dx, #2; \
2095 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2096 vshl.u32 rg_dx8, rg_dx, #3; \
2098 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2100 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2101 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2103 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2104 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2106 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2107 vmov.u8 d64_1, #1; \
2109 vmov.u8 d128_4, #4; \
2110 vmov.u8 d64_128, #128; \
2112 vmov.u8 d128_0x7, #0x7; \
2115 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2116 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2118 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2119 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
2121 cmp span_num_blocks, #0; \
2124 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2125 add num_blocks, span_num_blocks, num_blocks; \
2127 cmp num_blocks, #MAX_BLOCKS; \
2131 ldr b, [ span_b_offset ]; \
2132 add fb_ptr, fb_ptr, y, lsl #11; \
2134 vdup.u32 v_left_x, left_x; \
2137 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2138 add fb_ptr, fb_ptr, left_x, lsl #1; \
2140 mla b, b_dx, left_x, b; \
2141 and dither_shift, left_x, #0x03; \
2143 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2144 vshr.u32 rg_dx, rg_dx4, #2; \
2146 mov dither_shift, dither_shift, lsl #3; \
2147 vmla.u32 rg, rg_dx, v_left_x; \
2150 subs span_num_blocks, span_num_blocks, #1; \
2152 mov dither_row, dither_row, ror dither_shift; \
2153 mov b_dx4, b_dx, lsl #2; \
2155 vdup.u32 dither_offsets, dither_row; \
2156 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2158 vdup.u32 b_block, b; \
2159 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2161 mov b_dx8, b_dx, lsl #3; \
2162 vdup.u32 r_block, rg[0]; \
2163 vdup.u32 g_block, rg[1]; \
2165 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2167 vadd.u32 r_block, r_block, block_span; \
2168 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2170 vadd.u32 g_block, g_block, block_span; \
2171 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2173 vadd.u32 b_block, b_block, block_span; \
2174 add block_ptr_b, block_ptr_a, #16; \
2176 vshrn.u32 r_whole_low, r_block, #16; \
2177 vshrn.u32 g_whole_low, g_block, #16; \
2178 vshrn.u32 b_whole_low, b_block, #16; \
2179 vdup.u32 dx4, rg_dx4[0]; \
2181 vaddhn.u32 r_whole_high, r_block, dx4; \
2182 vdup.u32 dx4, rg_dx4[1]; \
2184 vaddhn.u32 g_whole_high, g_block, dx4; \
2185 vdup.u32 dx4, b_dx4; \
2187 vaddhn.u32 b_whole_high, b_block, dx4; \
2188 vdup.u32 dx8, rg_dx8[0]; \
2190 vadd.u32 r_block, r_block, dx8; \
2191 vdup.u32 dx8, rg_dx8[1]; \
2193 vadd.u32 g_block, g_block, dx8; \
2194 vdup.u32 dx8, b_dx8; \
2196 vadd.u32 b_block, b_block, dx8; \
2198 vmovn.u16 r_whole_8, r_whole; \
2199 vmovn.u16 g_whole_8, g_whole; \
2200 vmovn.u16 b_whole_8, b_whole; \
2203 veor.u32 draw_mask, draw_mask, draw_mask; \
2206 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2207 vshrn.u32 r_whole_low, r_block, #16; \
2209 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2210 vshrn.u32 g_whole_low, g_block, #16; \
2212 vshrn.u32 b_whole_low, b_block, #16; \
2213 str fb_ptr, [ block_ptr_a, #44 ]; \
2215 vdup.u32 dx4, rg_dx4[0]; \
2216 vshr.u8 r_whole_8, r_whole_8, #3; \
2217 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2219 vaddhn.u32 r_whole_high, r_block, dx4; \
2220 vdup.u32 dx4, rg_dx4[1]; \
2222 vaddhn.u32 g_whole_high, g_block, dx4; \
2223 vdup.u32 dx4, b_dx4; \
2225 vaddhn.u32 b_whole_high, b_block, dx4; \
2226 vdup.u32 dx8, rg_dx8[0]; \
2228 vmull.u8 pixels, r_whole_8, d64_1; \
2229 vmlal.u8 pixels, g_whole_8, d64_4; \
2230 vmlal.u8 pixels, b_whole_8, d64_128; \
2232 vadd.u32 r_block, r_block, dx8; \
2233 vdup.u32 dx8, rg_dx8[1]; \
2235 vadd.u32 g_block, g_block, dx8; \
2236 vdup.u32 dx8, b_dx8; \
2238 vadd.u32 b_block, b_block, dx8; \
2239 add fb_ptr, fb_ptr, #16; \
2241 vmovn.u16 r_whole_8, r_whole; \
2242 vmovn.u16 g_whole_8, g_whole; \
2243 vmovn.u16 b_whole_8, b_whole; \
2245 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2246 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2250 subs span_num_blocks, span_num_blocks, #1; \
2254 str fb_ptr, [ block_ptr_a, #44 ]; \
2255 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2257 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2258 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2260 vshr.u8 r_whole_8, r_whole_8, #3; \
2261 vdup.u8 draw_mask, right_mask; \
2263 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2264 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
2266 vtst.u16 draw_mask, draw_mask, test_mask; \
2268 vmull.u8 pixels, r_whole_8, d64_1; \
2269 vmlal.u8 pixels, g_whole_8, d64_4; \
2270 vmlal.u8 pixels, b_whole_8, d64_128; \
2272 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2273 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2276 add span_uvrg_offset, span_uvrg_offset, #16; \
2277 add span_b_offset, span_b_offset, #4; \
2279 add span_edge_data, span_edge_data, #8; \
2280 subs num_spans, num_spans, #1; \
2282 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2285 ldmia sp!, { r4 - r11, pc }; \
2288 /* TODO: Load from psx_gpu instead of saving/restoring these */\
2291 stmdb sp!, { r0 - r3, r12, r14 }; \
2292 bl flush_render_block_buffer; \
2293 ldmia sp!, { r0 - r3, r12, r14 }; \
2297 vmov.u8 d64_1, #1; \
2298 vmov.u8 d128_4, #4; \
2299 vmov.u8 d64_128, #128; \
2300 vmov.u8 d128_0x7, #0x7; \
2302 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2304 mov num_blocks, span_num_blocks; \
2305 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2309 setup_blocks_shaded_untextured_indirect_builder(undithered)
2310 setup_blocks_shaded_untextured_indirect_builder(dithered)
2315 #define mask_msb_ptr r14
2317 #define draw_mask q0
2318 #define pixels_low d16
2319 #define pixels_high d17
2323 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2326 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2327 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2328 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2330 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2332 cmp num_spans, #0; \
2335 stmdb sp!, { r4 - r11, r14 }; \
2336 vshl.u32 rg_dx4, rg_dx, #2; \
2338 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2339 vshl.u32 rg_dx8, rg_dx, #3; \
2341 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2342 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2344 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2345 vmov.u8 d64_1, #1; \
2347 vmov.u8 d128_4, #4; \
2348 vmov.u8 d64_128, #128; \
2350 vmov.u8 d128_0x7, #0x7; \
2351 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2352 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
2355 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2356 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2358 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2359 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
2361 cmp span_num_blocks, #0; \
2364 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2365 add fb_ptr, fb_ptr, y, lsl #11; \
2367 ldr b, [ span_b_offset ]; \
2368 vdup.u32 v_left_x, left_x; \
2371 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2372 add fb_ptr, fb_ptr, left_x, lsl #1; \
2374 mla b, b_dx, left_x, b; \
2375 and dither_shift, left_x, #0x03; \
2377 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2378 vshr.u32 rg_dx, rg_dx4, #2; \
2380 mov dither_shift, dither_shift, lsl #3; \
2381 vmla.u32 rg, rg_dx, v_left_x; \
2383 subs span_num_blocks, span_num_blocks, #1; \
2385 mov dither_row, dither_row, ror dither_shift; \
2386 mov b_dx4, b_dx, lsl #2; \
2388 vdup.u32 dither_offsets, dither_row; \
2389 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2391 vdup.u32 b_block, b; \
2392 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2394 mov b_dx8, b_dx, lsl #3; \
2395 vdup.u32 r_block, rg[0]; \
2396 vdup.u32 g_block, rg[1]; \
2398 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2400 vadd.u32 r_block, r_block, block_span; \
2401 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2403 vadd.u32 g_block, g_block, block_span; \
2404 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2406 vadd.u32 b_block, b_block, block_span; \
2407 add block_ptr_b, block_ptr_a, #16; \
2409 vshrn.u32 r_whole_low, r_block, #16; \
2410 vshrn.u32 g_whole_low, g_block, #16; \
2411 vshrn.u32 b_whole_low, b_block, #16; \
2412 vdup.u32 dx4, rg_dx4[0]; \
2414 vaddhn.u32 r_whole_high, r_block, dx4; \
2415 vdup.u32 dx4, rg_dx4[1]; \
2417 vaddhn.u32 g_whole_high, g_block, dx4; \
2418 vdup.u32 dx4, b_dx4; \
2420 vaddhn.u32 b_whole_high, b_block, dx4; \
2421 vdup.u32 dx8, rg_dx8[0]; \
2423 vadd.u32 r_block, r_block, dx8; \
2424 vdup.u32 dx8, rg_dx8[1]; \
2426 vadd.u32 g_block, g_block, dx8; \
2427 vdup.u32 dx8, b_dx8; \
2429 vadd.u32 b_block, b_block, dx8; \
2431 vmovn.u16 r_whole_8, r_whole; \
2432 vmovn.u16 g_whole_8, g_whole; \
2433 vmovn.u16 b_whole_8, b_whole; \
2438 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2439 vshrn.u32 r_whole_low, r_block, #16; \
2441 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2442 vshrn.u32 g_whole_low, g_block, #16; \
2444 vshrn.u32 b_whole_low, b_block, #16; \
2446 vdup.u32 dx4, rg_dx4[0]; \
2447 vshr.u8 r_whole_8, r_whole_8, #3; \
2448 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2450 vaddhn.u32 r_whole_high, r_block, dx4; \
2451 vdup.u32 dx4, rg_dx4[1]; \
2453 vmov pixels, msb_mask; \
2454 vaddhn.u32 g_whole_high, g_block, dx4; \
2455 vdup.u32 dx4, b_dx4; \
2457 vaddhn.u32 b_whole_high, b_block, dx4; \
2458 vdup.u32 dx8, rg_dx8[0]; \
2460 vmlal.u8 pixels, r_whole_8, d64_1; \
2461 vmlal.u8 pixels, g_whole_8, d64_4; \
2462 vmlal.u8 pixels, b_whole_8, d64_128; \
2464 vadd.u32 r_block, r_block, dx8; \
2465 vdup.u32 dx8, rg_dx8[1]; \
2467 vadd.u32 g_block, g_block, dx8; \
2468 vdup.u32 dx8, b_dx8; \
2470 vadd.u32 b_block, b_block, dx8; \
2472 vmovn.u16 r_whole_8, r_whole; \
2473 vmovn.u16 g_whole_8, g_whole; \
2474 vmovn.u16 b_whole_8, b_whole; \
2476 vst1.u32 { pixels }, [ fb_ptr ]!; \
2477 subs span_num_blocks, span_num_blocks, #1; \
2481 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2483 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2484 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2486 vshr.u8 r_whole_8, r_whole_8, #3; \
2487 rbit right_mask, right_mask; \
2488 vmov pixels, msb_mask; \
2489 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2490 clz right_mask, right_mask; \
2492 vmlal.u8 pixels, r_whole_8, d64_1; \
2493 vmlal.u8 pixels, g_whole_8, d64_4; \
2494 vmlal.u8 pixels, b_whole_8, d64_128; \
2496 ldr pc, [ pc, right_mask, lsl #2 ]; \
2509 vst1.u16 { pixels_low[0] }, [ fb_ptr ]; \
2513 vst1.u32 { pixels_low[0] }, [ fb_ptr ]; \
2517 vst1.u32 { pixels_low[0] }, [ fb_ptr ]!; \
2518 vst1.u16 { pixels_low[2] }, [ fb_ptr ]; \
2522 vst1.u32 { pixels_low }, [ fb_ptr ]; \
2526 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2527 vst1.u16 { pixels_high[0] }, [ fb_ptr ]; \
2531 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2532 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2536 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2537 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2538 vst1.u16 { pixels_high[2] }, [ fb_ptr ]; \
2542 vst1.u32 { pixels }, [ fb_ptr ]; \
2546 add span_uvrg_offset, span_uvrg_offset, #16; \
2547 add span_b_offset, span_b_offset, #4; \
2549 add span_edge_data, span_edge_data, #8; \
2550 subs num_spans, num_spans, #1; \
2554 ldmia sp!, { r4 - r11, pc } \
2556 setup_blocks_shaded_untextured_direct_builder(undithered)
2557 setup_blocks_shaded_untextured_direct_builder(dithered)
2566 #define block_ptr r1
2567 #define num_blocks r2
2580 #define texture_ptr r11
2594 #define pixels_d r10
2598 #define clut_ptr r12
2599 #define current_texture_mask r5
2600 #define dirty_textures_mask r6
2604 #define clut_low_a d2
2605 #define clut_low_b d3
2606 #define clut_high_a d4
2607 #define clut_high_b d5
2612 #define texels_low d6
2613 #define texels_high d7
2617 function(texture_blocks_untextured)
2623 function(texture_blocks_4bpp)
2624 stmdb sp!, { r3 - r11, r14 }
2625 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2627 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2628 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2630 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2631 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]
2633 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2634 vuzp.u8 clut_a, clut_b
2636 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
2637 tst dirty_textures_mask, current_texture_mask
2643 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2645 uxtah uv_0, texture_ptr, uv_01
2646 uxtah uv_1, texture_ptr, uv_01, ror #16
2648 uxtah uv_2, texture_ptr, uv_23
2649 uxtah uv_3, texture_ptr, uv_23, ror #16
2651 uxtah uv_4, texture_ptr, uv_45
2652 ldrb pixel_0, [ uv_0 ]
2654 uxtah uv_5, texture_ptr, uv_45, ror #16
2655 ldrb pixel_1, [ uv_1 ]
2657 uxtah uv_6, texture_ptr, uv_67
2658 ldrb pixel_2, [ uv_2 ]
2660 uxtah uv_7, texture_ptr, uv_67, ror #16
2661 ldrb pixel_3, [ uv_3 ]
2663 ldrb pixel_4, [ uv_4 ]
2664 subs num_blocks, num_blocks, #1
2666 ldrb pixel_5, [ uv_5 ]
2667 orr pixels_a, pixel_0, pixel_1, lsl #8
2669 ldrb pixel_6, [ uv_6 ]
2670 orr pixels_b, pixel_4, pixel_5, lsl #8
2672 ldrb pixel_7, [ uv_7 ]
2673 orr pixels_a, pixels_a, pixel_2, lsl #16
2675 orr pixels_b, pixels_b, pixel_6, lsl #16
2676 orr pixels_a, pixels_a, pixel_3, lsl #24
2678 orr pixels_b, pixels_b, pixel_7, lsl #24
2679 vmov.u32 texels, pixels_a, pixels_b
2681 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2682 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2684 vst2.u8 { texels_low, texels_high }, [ block_ptr, :128 ], c_64
2687 ldmia sp!, { r3 - r11, pc }
2690 stmdb sp!, { r1 - r2 }
2691 bl update_texture_4bpp_cache
2694 ldmia sp!, { r1 - r2 }
2700 function(texture_blocks_8bpp)
2701 stmdb sp!, { r3 - r11, r14 }
2702 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2704 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2705 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2707 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2708 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2710 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]
2711 tst dirty_textures_mask, current_texture_mask
2717 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2719 uxtah uv_0, texture_ptr, uv_01
2720 uxtah uv_1, texture_ptr, uv_01, ror #16
2722 uxtah uv_2, texture_ptr, uv_23
2723 uxtah uv_3, texture_ptr, uv_23, ror #16
2725 uxtah uv_4, texture_ptr, uv_45
2726 ldrb pixel_0, [ uv_0 ]
2728 uxtah uv_5, texture_ptr, uv_45, ror #16
2729 ldrb pixel_1, [ uv_1 ]
2731 uxtah uv_6, texture_ptr, uv_67
2732 ldrb pixel_2, [ uv_2 ]
2734 uxtah uv_7, texture_ptr, uv_67, ror #16
2735 ldrb pixel_3, [ uv_3 ]
2737 ldrb pixel_4, [ uv_4 ]
2738 add pixel_0, pixel_0, pixel_0
2740 ldrb pixel_5, [ uv_5 ]
2741 add pixel_1, pixel_1, pixel_1
2743 ldrb pixel_6, [ uv_6 ]
2744 add pixel_2, pixel_2, pixel_2
2746 ldrb pixel_7, [ uv_7 ]
2747 add pixel_3, pixel_3, pixel_3
2749 ldrh pixel_0, [ clut_ptr, pixel_0 ]
2750 add pixel_4, pixel_4, pixel_4
2752 ldrh pixel_1, [ clut_ptr, pixel_1 ]
2753 add pixel_5, pixel_5, pixel_5
2755 ldrh pixel_2, [ clut_ptr, pixel_2 ]
2756 add pixel_6, pixel_6, pixel_6
2758 ldrh pixel_3, [ clut_ptr, pixel_3 ]
2759 add pixel_7, pixel_7, pixel_7
2761 ldrh pixel_4, [ clut_ptr, pixel_4 ]
2762 orr pixels_a, pixel_0, pixel_1, lsl #16
2764 ldrh pixel_5, [ clut_ptr, pixel_5 ]
2765 orr pixels_c, pixel_2, pixel_3, lsl #16
2767 ldrh pixel_6, [ clut_ptr, pixel_6 ]
2768 subs num_blocks, num_blocks, #1
2770 ldrh pixel_7, [ clut_ptr, pixel_7 ]
2771 orr pixels_b, pixel_4, pixel_5, lsl #16
2773 orr pixels_d, pixel_6, pixel_7, lsl #16
2774 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2776 add block_ptr, block_ptr, #64
2779 ldmia sp!, { r3 - r11, pc }
2782 stmdb sp!, { r1 - r2, r12 }
2784 bl update_texture_8bpp_cache
2786 ldmia sp!, { r1 - r2, r12 }
2816 #define block_ptr r1
2817 #define num_blocks r2
2861 #define texture_ptr r12
2866 function(texture_blocks_16bpp)
2867 stmdb sp!, { r3 - r11, r14 }
2868 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2870 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2871 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2874 ldrh uv_0, [ block_ptr ]
2875 subs num_blocks, num_blocks, #1
2877 ldrh uv_1, [ block_ptr, #2 ]
2879 and v_0, uv_0, #0xFF00
2880 and v_1, uv_1, #0xFF00
2882 and u_0, uv_0, #0xFF
2883 and u_1, uv_1, #0xFF
2885 add uv_0, u_0, v_0, lsl #2
2886 ldrh uv_2, [ block_ptr, #4 ]
2888 add uv_1, u_1, v_1, lsl #2
2889 ldrh uv_3, [ block_ptr, #6 ]
2891 add uv_0, uv_0, uv_0
2892 add uv_1, uv_1, uv_1
2894 and v_2, uv_2, #0xFF00
2895 and v_3, uv_3, #0xFF00
2897 and u_2, uv_2, #0xFF
2898 and u_3, uv_3, #0xFF
2900 add uv_2, u_2, v_2, lsl #2
2901 ldrh uv_4, [ block_ptr, #8 ]
2903 add uv_3, u_3, v_3, lsl #2
2904 ldrh uv_5, [ block_ptr, #10 ]
2906 add uv_2, uv_2, uv_2
2907 add uv_3, uv_3, uv_3
2909 and v_4, uv_4, #0xFF00
2910 and v_5, uv_5, #0xFF00
2912 and u_4, uv_4, #0xFF
2913 and u_5, uv_5, #0xFF
2915 add uv_4, u_4, v_4, lsl #2
2916 ldrh uv_6, [ block_ptr, #12 ]
2918 add uv_5, u_5, v_5, lsl #2
2919 ldrh uv_7, [ block_ptr, #14 ]
2921 add uv_4, uv_4, uv_4
2922 ldrh pixel_0, [ texture_ptr, uv_0 ]
2924 add uv_5, uv_5, uv_5
2925 ldrh pixel_1, [ texture_ptr, uv_1 ]
2927 and v_6, uv_6, #0xFF00
2928 ldrh pixel_2, [ texture_ptr, uv_2 ]
2930 and v_7, uv_7, #0xFF00
2931 ldrh pixel_3, [ texture_ptr, uv_3 ]
2933 and u_6, uv_6, #0xFF
2934 ldrh pixel_4, [ texture_ptr, uv_4 ]
2936 and u_7, uv_7, #0xFF
2937 ldrh pixel_5, [ texture_ptr, uv_5 ]
2939 add uv_6, u_6, v_6, lsl #2
2940 add uv_7, u_7, v_7, lsl #2
2942 add uv_6, uv_6, uv_6
2943 add uv_7, uv_7, uv_7
2945 orr pixels_a, pixel_0, pixel_1, lsl #16
2946 orr pixels_b, pixel_2, pixel_3, lsl #16
2948 ldrh pixel_6, [ texture_ptr, uv_6 ]
2949 orr pixels_c, pixel_4, pixel_5, lsl #16
2951 ldrh pixel_7, [ texture_ptr, uv_7 ]
2952 orr pixels_d, pixel_6, pixel_7, lsl #16
2954 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
2955 add block_ptr, block_ptr, #64
2959 ldmia sp!, { r3 - r11, pc }
2974 #undef msb_mask_high
2982 #define num_blocks r1
2983 #define color_ptr r2
2984 #define colors_scalar r2
2985 #define colors_scalar_compare r3
2986 #define mask_msb_ptr r2
2988 #define block_ptr_load_a r0
2989 #define block_ptr_store r3
2990 #define block_ptr_load_b r12
2995 #define draw_mask_bits_scalar r5
2997 #define d128_0x07 q0
2998 #define d128_0x1F q1
2999 #define d128_0x8000 q2
3000 #define test_mask q3
3002 #define colors_rg q5
3003 #define colors_b_dm_bits q6
3004 #define texels_rg q7
3007 #define pixels_b q10
3009 #define zero_mask q4
3010 #define draw_mask q12
3011 #define msb_mask q13
3013 #define fb_pixels q8
3015 #define pixels_gb_low q9
3017 #define colors_r d10
3018 #define colors_g d11
3019 #define colors_b d12
3020 #define draw_mask_bits d13
3021 #define texels_r d14
3022 #define texels_g d15
3023 #define pixels_r_low d16
3024 #define pixels_g_low d18
3025 #define pixels_b_low d19
3026 #define msb_mask_low d26
3027 #define msb_mask_high d27
3032 #define texels_b d31
3034 #define shade_blocks_textured_modulated_prologue_indirect() \
3036 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3038 #define shade_blocks_textured_modulated_prologue_direct() \
3039 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3040 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ] \
3043 #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \
3045 #define shade_blocks_textured_false_modulation_check_undithered(target) \
3046 ldr colors_scalar, [ psx_gpu, #psx_gpu_triangle_color_offset ]; \
3047 movw colors_scalar_compare, #0x8080; \
3049 movt colors_scalar_compare, #0x80; \
3050 cmp colors_scalar, colors_scalar_compare; \
3051 beq shade_blocks_textured_unmodulated_##target \
3053 #define shade_blocks_textured_false_modulation_check_dithered(target) \
3055 #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \
3056 shade_blocks_textured_false_modulation_check_##dithering(target); \
3057 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3058 vld1.u32 { colors_r[] }, [ color_ptr, :32 ]; \
3059 vdup.u8 colors_g, colors_r[1]; \
3060 vdup.u8 colors_b, colors_r[2]; \
3061 vdup.u8 colors_r, colors_r[0] \
3064 #define shade_blocks_textured_modulated_load_dithered(target) \
3065 vld1.u32 { target }, [ block_ptr_load_b, :128 ] \
3067 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3068 vld1.u32 { target }, [ block_ptr_load_b, :128 ], c_32 \
3070 #define shade_blocks_textured_modulated_load_undithered(target) \
3072 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3073 add block_ptr_load_b, block_ptr_load_b, #32 \
3075 #define shade_blocks_textured_modulate_dithered(channel) \
3076 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3078 #define shade_blocks_textured_modulate_undithered(channel) \
3079 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3082 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3083 vst1.u32 { draw_mask }, [ block_ptr_store, :128 ]! \
3085 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3086 ldr fb_ptr, [ block_ptr_load_b, #(offset - 64) ]; \
3087 vld1.u32 { fb_pixels }, [ fb_ptr ]; \
3088 vbit.u16 pixels, fb_pixels, draw_mask \
3090 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3091 vst1.u32 { pixels }, [ block_ptr_store, :128 ], c_48 \
3093 #define shade_blocks_textured_modulated_store_pixels_direct() \
3094 vst1.u32 { pixels }, [ fb_ptr ] \
3097 #define shade_blocks_textured_modulated_load_rg_shaded() \
3098 vld1.u32 { colors_r, colors_g }, [ block_ptr_load_b, :128 ], c_32 \
3100 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3101 add block_ptr_load_b, block_ptr_load_b, #32 \
3103 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3104 vld1.u32 { colors_b, draw_mask_bits }, [ block_ptr_load_a, :128 ], c_32 \
3106 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3107 ldr draw_mask_bits_scalar, [ block_ptr_load_a, #8 ]; \
3108 add block_ptr_load_a, block_ptr_load_a, #32 \
3110 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3111 vdup.u16 draw_mask, draw_mask_bits[0] \
3113 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3114 vdup.u16 draw_mask, draw_mask_bits_scalar \
3117 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3119 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3120 vorr.u16 pixels, pixels, msb_mask \
3123 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3126 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3127 shade_blocks_textured_modulated_prologue_##shading(dithering, target); \
3128 stmdb sp!, { r4 - r5, lr }; \
3129 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3131 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
3133 shade_blocks_textured_modulated_prologue_##target(); \
3135 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3138 add block_ptr_load_b, block_ptr_load_a, #16; \
3139 vmov.u8 d64_1, #1; \
3140 vmov.u8 d64_4, #4; \
3141 vmov.u8 d64_128, #128; \
3143 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3144 vmov.u8 d128_0x07, #0x07; \
3146 shade_blocks_textured_modulated_load_rg_##shading(); \
3147 vmov.u8 d128_0x1F, #0x1F; \
3149 shade_blocks_textured_modulated_load_bdm_##shading(); \
3150 vmov.u16 d128_0x8000, #0x8000; \
3152 vmovn.u16 texels_r, texels; \
3153 vshrn.u16 texels_g, texels, #5; \
3155 vshrn.u16 texels_b, texels, #7; \
3156 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3158 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3159 vtst.u16 draw_mask, draw_mask, test_mask; \
3161 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3162 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3164 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3165 vshr.u8 texels_b, texels_b, #3; \
3167 shade_blocks_textured_modulate_##dithering(r); \
3168 shade_blocks_textured_modulate_##dithering(g); \
3169 shade_blocks_textured_modulate_##dithering(b); \
3171 vand.u16 pixels, texels, d128_0x8000; \
3172 vceq.u16 zero_mask, texels, #0; \
3174 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3175 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3176 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3178 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3179 vorr.u16 draw_mask, draw_mask, zero_mask; \
3180 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3181 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3183 subs num_blocks, num_blocks, #1; \
3189 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3190 shade_blocks_textured_modulated_load_rg_##shading(); \
3191 vshrn.u16 texels_g, texels, #5; \
3193 shade_blocks_textured_modulated_load_bdm_##shading(); \
3194 vshrn.u16 texels_b, texels, #7; \
3196 pld [ block_ptr_load_a ]; \
3197 vmovn.u16 texels_r, texels; \
3198 vmlal.u8 pixels, pixels_r_low, d64_1; \
3200 vmlal.u8 pixels, pixels_g_low, d64_4; \
3201 vmlal.u8 pixels, pixels_b_low, d64_128; \
3202 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3204 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3205 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3207 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3208 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3210 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3211 vtst.u16 draw_mask, draw_mask, test_mask; \
3213 shade_blocks_textured_modulated_store_pixels_##target(); \
3214 vshr.u8 texels_b, texels_b, #3; \
3216 shade_blocks_textured_modulate_##dithering(r); \
3217 shade_blocks_textured_modulate_##dithering(g); \
3218 shade_blocks_textured_modulate_##dithering(b); \
3220 vand.u16 pixels, texels, d128_0x8000; \
3221 vceq.u16 zero_mask, texels, #0; \
3223 subs num_blocks, num_blocks, #1; \
3225 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3226 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3227 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3229 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3230 vorr.u16 draw_mask, draw_mask, zero_mask; \
3231 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3232 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3237 vmlal.u8 pixels, pixels_r_low, d64_1; \
3238 vmlal.u8 pixels, pixels_g_low, d64_4; \
3239 vmlal.u8 pixels, pixels_b_low, d64_128; \
3241 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3242 shade_blocks_textured_modulated_store_pixels_##target(); \
3244 ldmia sp!, { r4 - r5, pc } \
3247 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3248 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3249 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3250 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3252 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3253 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3254 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3255 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3273 #undef msb_mask_high
3276 #define num_blocks r1
3277 #define mask_msb_ptr r2
3278 #define color_ptr r3
3280 #define block_ptr_load r0
3281 #define draw_mask_store_ptr r3
3282 #define draw_mask_bits_ptr r12
3283 #define draw_mask_ptr r12
3284 #define pixel_store_ptr r14
3286 #define fb_ptr_cmp r4
3289 #define fb_ptr_next r14
3293 #define test_mask q0
3295 #define draw_mask q2
3296 #define zero_mask q3
3297 #define draw_mask_combined q4
3298 #define fb_pixels q5
3299 #define fb_pixels_next q6
3302 #define draw_mask_low d4
3303 #define draw_mask_high d5
3304 #define msb_mask_low d14
3305 #define msb_mask_high d15
3308 function(shade_blocks_textured_unmodulated_indirect)
3309 str r14, [ sp, #-4 ]
3310 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3312 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3313 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3315 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3316 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3319 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3321 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3322 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3323 [ draw_mask_bits_ptr, :16 ], c_64
3324 vceq.u16 zero_mask, pixels, #0
3326 vtst.u16 draw_mask, draw_mask, test_mask
3327 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3329 subs num_blocks, num_blocks, #1
3333 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3334 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3336 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3337 [ draw_mask_bits_ptr, :16 ], c_64
3338 vceq.u16 zero_mask, pixels, #0
3340 vtst.u16 draw_mask, draw_mask, test_mask
3341 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3343 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3344 subs num_blocks, num_blocks, #1
3349 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3350 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3357 function(shade_blocks_textured_unmodulated_direct)
3358 stmdb sp!, { r4, r14 }
3359 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3361 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3362 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3364 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3367 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3368 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3370 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3371 [ draw_mask_bits_ptr, :16 ], c_64
3372 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3374 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3375 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3376 vceq.u16 zero_mask, pixels, #0
3377 vtst.u16 draw_mask, draw_mask, test_mask
3379 subs num_blocks, num_blocks, #1
3383 mov fb_ptr, fb_ptr_next
3384 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3386 vorr.u16 pixels, pixels, msb_mask
3388 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3389 vmov fb_pixels, fb_pixels_next
3391 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3392 [ draw_mask_bits_ptr, :16 ], c_64
3393 vbif.u16 fb_pixels, pixels, draw_mask_combined
3395 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3397 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3398 add fb_ptr_cmp, fb_ptr_cmp, #14
3402 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3403 vceq.u16 zero_mask, pixels, #0
3405 vst1.u16 { fb_pixels }, [ fb_ptr ]
3406 vtst.u16 draw_mask, draw_mask, test_mask
3409 subs num_blocks, num_blocks, #1
3413 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3414 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3416 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3418 ldmia sp!, { r4, pc }
3421 vst1.u16 { fb_pixels }, [ fb_ptr ]
3422 vceq.u16 zero_mask, pixels, #0
3424 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3425 vtst.u16 draw_mask, draw_mask, test_mask
3430 function(shade_blocks_unshaded_untextured_indirect)
3435 function(shade_blocks_unshaded_untextured_direct)
3436 stmdb sp!, { r4, r14 }
3437 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3439 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3440 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3442 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3443 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3445 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3446 vld1.u16 { pixels }, [ color_ptr, :128 ]
3449 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3451 vorr.u16 pixels, pixels, msb_mask
3452 subs num_blocks, num_blocks, #1
3454 ldr fb_ptr_next, [ block_ptr_load ], #64
3456 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3460 vmov fb_pixels, fb_pixels_next
3461 mov fb_ptr, fb_ptr_next
3462 ldr fb_ptr_next, [ block_ptr_load ], #64
3464 vbif.u16 fb_pixels, pixels, draw_mask
3465 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3467 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3468 add fb_ptr_cmp, fb_ptr_cmp, #14
3472 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3473 vst1.u16 { fb_pixels }, [ fb_ptr ]
3476 subs num_blocks, num_blocks, #1
3480 vbif.u16 fb_pixels_next, pixels, draw_mask
3481 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3483 ldmia sp!, { r4, pc }
3486 vst1.u16 { fb_pixels }, [ fb_ptr ]
3487 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3491 #undef draw_mask_ptr
3498 #define num_blocks r1
3499 #define msb_mask_ptr r2
3500 #define pixel_ptr r3
3501 #define draw_mask_ptr r0
3504 #define fb_ptr_next r14
3505 #define fb_ptr_cmp r4
3513 #undef msb_mask_high
3514 #undef draw_mask_next
3517 #undef fb_pixels_next
3520 #define draw_mask q1
3522 #define fb_pixels q3
3523 #define blend_pixels q4
3524 #define pixels_no_msb q5
3525 #define blend_mask q6
3526 #define fb_pixels_no_msb q7
3527 #define d128_0x8000 q8
3528 #define d128_0x0421 q9
3529 #define fb_pixels_next q10
3530 #define blend_pixels_next q11
3531 #define pixels_next q12
3532 #define draw_mask_next q13
3533 #define write_mask q14
3535 #define pixels_rb q5
3536 #define pixels_mg q7
3538 #define d128_0x7C1F q8
3539 #define d128_0x03E0 q9
3540 #define fb_pixels_rb q10
3541 #define fb_pixels_g q11
3542 #define fb_pixels_masked q11
3543 #define d128_0x83E0 q15
3544 #define pixels_fourth q7
3545 #define d128_0x1C07 q12
3546 #define d128_0x00E0 q13
3547 #define d128_0x80E0 q13
3549 #define msb_mask_low d0
3550 #define msb_mask_high d1
3552 #define blend_blocks_average_set_blend_mask_textured(source) \
3553 vclt.s16 blend_mask, source, #0 \
3555 #define blend_blocks_average_set_stp_bit_textured() \
3556 vorr.u16 blend_pixels, #0x8000 \
3558 #define blend_blocks_average_combine_textured(source) \
3559 vbif.u16 blend_pixels, source, blend_mask \
3561 #define blend_blocks_average_set_blend_mask_untextured(source) \
3563 #define blend_blocks_average_set_stp_bit_untextured() \
3565 #define blend_blocks_average_combine_untextured(source) \
3567 #define blend_blocks_average_mask_set_on() \
3568 vclt.s16 write_mask, fb_pixels_next, #0 \
3570 #define blend_blocks_average_mask_copy_on() \
3571 vorr.u16 draw_mask, draw_mask_next, write_mask \
3573 #define blend_blocks_average_mask_copy_b_on() \
3574 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3576 #define blend_blocks_average_mask_set_off() \
3578 #define blend_blocks_average_mask_copy_off() \
3579 vmov draw_mask, draw_mask_next \
3581 #define blend_blocks_average_mask_copy_b_off() \
3583 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3586 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3587 stmdb sp!, { r4, r14 }; \
3588 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3589 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3591 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3592 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3594 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3597 vmov.u16 d128_0x8000, #0x8000; \
3598 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3599 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3601 vmov.u16 d128_0x0421, #0x0400; \
3602 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3604 vorr.u16 d128_0x0421, #0x0021; \
3605 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3607 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3608 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3609 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3610 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3611 blend_blocks_average_mask_set_##mask_evaluate(); \
3612 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3614 subs num_blocks, num_blocks, #1; \
3618 mov fb_ptr, fb_ptr_next; \
3619 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3621 vmov pixels, pixels_next; \
3622 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3624 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3626 blend_blocks_average_mask_copy_##mask_evaluate(); \
3627 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3629 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3630 blend_blocks_average_set_stp_bit_##texturing(); \
3631 vmov fb_pixels, fb_pixels_next; \
3632 blend_blocks_average_combine_##texturing(pixels); \
3634 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3635 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3636 cmp fb_ptr_cmp, #28; \
3639 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3640 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3642 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3643 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3645 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3646 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3648 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3649 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3650 blend_blocks_average_mask_set_##mask_evaluate(); \
3651 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3654 subs num_blocks, num_blocks, #1; \
3658 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3659 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3661 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3662 blend_blocks_average_set_stp_bit_##texturing(); \
3663 blend_blocks_average_combine_##texturing(pixels_next); \
3665 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3666 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3667 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3669 ldmia sp!, { r4, pc }; \
3672 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3673 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3674 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3676 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3677 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3678 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3679 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3680 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3681 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3685 blend_blocks_average_builder(textured, off)
3686 blend_blocks_average_builder(untextured, off)
3687 blend_blocks_average_builder(textured, on)
3688 blend_blocks_average_builder(untextured, on)
3691 #define blend_blocks_add_mask_set_on() \
3692 vclt.s16 write_mask, fb_pixels, #0 \
3694 #define blend_blocks_add_mask_copy_on() \
3695 vorr.u16 draw_mask, draw_mask, write_mask \
3697 #define blend_blocks_add_mask_set_off() \
3699 #define blend_blocks_add_mask_copy_off() \
3702 #define blend_blocks_add_textured_builder(mask_evaluate) \
3705 function(blend_blocks_textured_add_##mask_evaluate) \
3706 stmdb sp!, { r4, r14 }; \
3707 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3708 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3710 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3711 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3713 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3716 vmov.u16 d128_0x7C1F, #0x7C00; \
3717 vmov.u16 d128_0x03E0, #0x0300; \
3718 vmov.u16 d128_0x83E0, #0x8000; \
3719 vorr.u16 d128_0x03E0, #0x00E0; \
3720 vorr.u16 d128_0x7C1F, #0x001F; \
3721 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3723 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3724 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3725 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3726 vclt.s16 blend_mask, pixels, #0; \
3727 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3728 blend_blocks_add_mask_set_##mask_evaluate(); \
3729 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3731 blend_blocks_add_mask_copy_##mask_evaluate(); \
3732 vorr.u16 pixels, pixels, msb_mask; \
3733 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3734 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3735 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3736 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3737 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3738 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3739 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3740 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3742 subs num_blocks, num_blocks, #1; \
3746 mov fb_ptr, fb_ptr_next; \
3748 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3750 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3751 vclt.s16 blend_mask, pixels, #0; \
3753 vorr.u16 pixels, pixels, msb_mask; \
3754 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3755 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3757 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3758 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3760 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3761 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3762 cmp fb_ptr_cmp, #28; \
3765 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3766 blend_blocks_add_mask_set_##mask_evaluate(); \
3767 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3768 blend_blocks_add_mask_copy_##mask_evaluate(); \
3769 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3770 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3771 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3774 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3775 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3776 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3777 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3778 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3780 subs num_blocks, num_blocks, #1; \
3784 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3785 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3786 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3788 ldmia sp!, { r4, pc }; \
3791 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3792 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3794 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3795 blend_blocks_add_mask_set_##mask_evaluate(); \
3796 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3797 blend_blocks_add_mask_copy_##mask_evaluate(); \
3798 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3802 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3805 function(blend_blocks_untextured_add_##mask_evaluate) \
3806 stmdb sp!, { r4, r14 }; \
3807 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3808 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3810 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3811 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3813 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3816 vmov.u16 d128_0x7C1F, #0x7C00; \
3817 vmov.u16 d128_0x03E0, #0x0300; \
3818 vorr.u16 d128_0x7C1F, #0x001F; \
3819 vorr.u16 d128_0x03E0, #0x00E0; \
3821 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3822 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3823 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3824 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3825 blend_blocks_add_mask_set_##mask_evaluate(); \
3826 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3828 blend_blocks_add_mask_copy_##mask_evaluate(); \
3829 vand.u16 pixels_g, pixels, d128_0x03E0; \
3830 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3831 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3832 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3833 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3834 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3835 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3837 subs num_blocks, num_blocks, #1; \
3841 mov fb_ptr, fb_ptr_next; \
3843 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3845 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3847 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3848 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3849 vand.u16 pixels_g, pixels, d128_0x03E0; \
3851 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3852 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3854 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3855 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3856 cmp fb_ptr_cmp, #28; \
3859 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3860 blend_blocks_add_mask_set_##mask_evaluate(); \
3861 blend_blocks_add_mask_copy_##mask_evaluate(); \
3862 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3863 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3864 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3867 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3868 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3869 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3870 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3871 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3873 subs num_blocks, num_blocks, #1; \
3877 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3878 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3879 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3880 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3882 ldmia sp!, { r4, pc }; \
3885 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3886 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3888 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3889 blend_blocks_add_mask_set_##mask_evaluate(); \
3890 blend_blocks_add_mask_copy_##mask_evaluate(); \
3891 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3895 blend_blocks_add_textured_builder(off)
3896 blend_blocks_add_textured_builder(on)
3897 blend_blocks_add_untextured_builder(off)
3898 blend_blocks_add_untextured_builder(on)
3900 #define blend_blocks_subtract_set_blend_mask_textured() \
3901 vclt.s16 blend_mask, pixels_next, #0 \
3903 #define blend_blocks_subtract_combine_textured() \
3904 vbif.u16 blend_pixels, pixels, blend_mask \
3906 #define blend_blocks_subtract_set_stb_textured() \
3907 vorr.u16 blend_pixels, #0x8000 \
3909 #define blend_blocks_subtract_msb_mask_textured() \
3910 vorr.u16 pixels, pixels_next, msb_mask \
3912 #define blend_blocks_subtract_set_blend_mask_untextured() \
3914 #define blend_blocks_subtract_combine_untextured() \
3916 #define blend_blocks_subtract_set_stb_untextured() \
3917 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3919 #define blend_blocks_subtract_msb_mask_untextured() \
3922 #define blend_blocks_subtract_mask_set_on() \
3923 vclt.s16 write_mask, fb_pixels, #0 \
3925 #define blend_blocks_subtract_mask_copy_on() \
3926 vorr.u16 draw_mask, draw_mask_next, write_mask \
3928 #define blend_blocks_subtract_mask_set_off() \
3930 #define blend_blocks_subtract_mask_copy_off() \
3931 vmov draw_mask, draw_mask_next \
3934 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
3937 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
3938 stmdb sp!, { r4, r14 }; \
3939 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3940 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3942 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3943 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3945 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3948 vmov.u16 d128_0x7C1F, #0x7C00; \
3949 vmov.u16 d128_0x03E0, #0x0300; \
3950 vorr.u16 d128_0x7C1F, #0x001F; \
3951 vorr.u16 d128_0x03E0, #0x00E0; \
3953 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3954 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3955 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3956 blend_blocks_subtract_set_blend_mask_##texturing(); \
3957 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3958 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3959 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3961 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3962 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3963 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3964 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3965 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3967 subs num_blocks, num_blocks, #1; \
3971 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
3972 mov fb_ptr, fb_ptr_next; \
3973 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3975 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3976 blend_blocks_subtract_msb_mask_##texturing(); \
3978 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3979 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3980 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3981 blend_blocks_subtract_set_stb_##texturing(); \
3982 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3983 blend_blocks_subtract_combine_##texturing(); \
3984 blend_blocks_subtract_set_blend_mask_##texturing(); \
3985 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3987 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3988 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3989 cmp fb_ptr_cmp, #28; \
3992 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3993 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3994 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3995 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3996 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3997 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3998 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4001 subs num_blocks, num_blocks, #1; \
4005 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4007 blend_blocks_subtract_msb_mask_##texturing(); \
4008 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4009 blend_blocks_subtract_set_stb_##texturing(); \
4010 blend_blocks_subtract_combine_##texturing(); \
4011 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4012 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4014 ldmia sp!, { r4, pc }; \
4017 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4018 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4019 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4020 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4021 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4022 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4023 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4027 blend_blocks_subtract_builder(textured, off)
4028 blend_blocks_subtract_builder(textured, on)
4029 blend_blocks_subtract_builder(untextured, off)
4030 blend_blocks_subtract_builder(untextured, on)
4033 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4036 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4037 stmdb sp!, { r4, r14 }; \
4038 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4039 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4041 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4042 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4044 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4047 vmov.u16 d128_0x7C1F, #0x7C00; \
4048 vmov.u16 d128_0x03E0, #0x0300; \
4049 vmov.u16 d128_0x1C07, #0x1C00; \
4050 vmov.u16 d128_0x00E0, #0x00E0; \
4051 vorr.u16 d128_0x7C1F, #0x001F; \
4052 vorr.u16 d128_0x03E0, #0x00E0; \
4053 vorr.u16 d128_0x1C07, #0x0007; \
4055 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4056 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4057 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4058 vclt.s16 blend_mask, pixels, #0; \
4059 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4060 blend_blocks_add_mask_set_##mask_evaluate(); \
4061 vshr.s16 pixels_fourth, pixels, #2; \
4062 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4064 blend_blocks_add_mask_copy_##mask_evaluate(); \
4065 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4066 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4067 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4068 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4069 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4070 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4071 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4073 subs num_blocks, num_blocks, #1; \
4077 mov fb_ptr, fb_ptr_next; \
4078 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4080 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4081 vbif.u16 blend_pixels, pixels, blend_mask; \
4083 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4084 vclt.s16 blend_mask, pixels, #0; \
4085 vshr.s16 pixels_fourth, pixels, #2; \
4086 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4087 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4089 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4090 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4092 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4093 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4094 cmp fb_ptr_cmp, #28; \
4097 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4098 blend_blocks_add_mask_set_##mask_evaluate(); \
4099 blend_blocks_add_mask_copy_##mask_evaluate(); \
4100 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4101 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4102 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4105 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4106 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4107 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4108 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4109 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4111 subs num_blocks, num_blocks, #1; \
4115 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4116 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4117 vbif.u16 blend_pixels, pixels, blend_mask; \
4118 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4119 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4121 ldmia sp!, { r4, pc }; \
4124 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4125 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4127 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4128 blend_blocks_add_mask_set_##mask_evaluate(); \
4129 blend_blocks_add_mask_copy_##mask_evaluate(); \
4130 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4135 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4138 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4139 stmdb sp!, { r4, r14 }; \
4140 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4141 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4143 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4144 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4146 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4149 vmov.u16 d128_0x7C1F, #0x7C00; \
4150 vmov.u16 d128_0x03E0, #0x0300; \
4151 vmov.u16 d128_0x1C07, #0x1C00; \
4152 vmov.u16 d128_0x00E0, #0x00E0; \
4153 vorr.u16 d128_0x7C1F, #0x001F; \
4154 vorr.u16 d128_0x03E0, #0x00E0; \
4155 vorr.u16 d128_0x1C07, #0x0007; \
4157 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4158 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4159 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4160 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4161 blend_blocks_add_mask_set_##mask_evaluate(); \
4162 vshr.s16 pixels_fourth, pixels, #2; \
4163 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4165 blend_blocks_add_mask_copy_##mask_evaluate(); \
4166 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4167 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4168 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4169 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4170 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4171 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4172 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4174 subs num_blocks, num_blocks, #1; \
4178 mov fb_ptr, fb_ptr_next; \
4179 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4181 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4183 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4184 vshr.s16 pixels_fourth, pixels, #2; \
4185 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4186 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4188 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4189 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4191 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4192 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4193 cmp fb_ptr_cmp, #28; \
4196 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4197 blend_blocks_add_mask_set_##mask_evaluate(); \
4198 blend_blocks_add_mask_copy_##mask_evaluate(); \
4199 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4200 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4201 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4204 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4205 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4206 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4207 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4208 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4210 subs num_blocks, num_blocks, #1; \
4214 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4215 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4216 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4217 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4219 ldmia sp!, { r4, pc }; \
4222 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4223 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4225 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4226 blend_blocks_add_mask_set_##mask_evaluate(); \
4227 blend_blocks_add_mask_copy_##mask_evaluate(); \
4228 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4232 blend_blocks_add_fourth_textured_builder(off)
4233 blend_blocks_add_fourth_textured_builder(on)
4234 blend_blocks_add_fourth_untextured_builder(off)
4235 blend_blocks_add_fourth_untextured_builder(on)
4237 // TODO: Optimize this more. Need a scene that actually uses it for
4242 function(blend_blocks_textured_unblended_on)
4243 stmdb sp!, { r4, r14 }
4244 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4245 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
4247 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4248 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
4250 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4253 ldr fb_ptr, [ pixel_ptr, #28 ]
4254 vld1.u16 { fb_pixels }, [ fb_ptr ]
4255 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4256 vclt.s16 write_mask, fb_pixels, #0
4257 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4259 subs num_blocks, num_blocks, #1
4263 vorr.u16 draw_mask, draw_mask, write_mask
4264 vbif.u16 fb_pixels, pixels, draw_mask
4265 vst1.u16 { fb_pixels }, [ fb_ptr ]
4267 ldr fb_ptr, [ pixel_ptr, #28 ]
4268 vld1.u16 { fb_pixels }, [ fb_ptr ]
4269 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4270 vclt.s16 write_mask, fb_pixels, #0
4271 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4273 subs num_blocks, num_blocks, #1
4277 vorr.u16 draw_mask, draw_mask, write_mask
4278 vbif.u16 fb_pixels, pixels, draw_mask
4279 vst1.u16 { fb_pixels }, [ fb_ptr ]
4281 ldmia sp!, { r4, pc }
4284 function(blend_blocks_textured_unblended_off)
4294 vld1.u32 { u_whole_8, v_whole_8 }, [ r1, :128 ], r3
4314 #define num_width r12
4324 function(render_block_fill_body)
4325 vdup.u16 colors_a, color
4328 vmov colors_b, colors_a
4329 sub pitch, pitch, width, lsl #1
4331 mov num_width, width
4334 vst1.u32 { colors_a, colors_b }, [ vram_ptr, :256 ]!
4336 subs num_width, num_width, #16
4339 add vram_ptr, vram_ptr, pitch
4340 mov num_width, width
4342 subs height, height, #1
4356 #undef dirty_textures_mask
4358 #undef current_texture_mask
4369 #define offset_u_right r10
4370 #define width_rounded r11
4371 #define height_rounded r12
4373 #define texture_offset_base r1
4374 #define tile_width r2
4375 #define tile_height r3
4376 #define num_blocks r4
4378 #define sub_tile_height r6
4380 #define texture_mask r8
4381 #define column_data r9
4382 #define texture_offset r10
4383 #define tiles_remaining r11
4384 #define fb_ptr_advance_column r12
4385 #define texture_block_ptr r14
4387 #define texture_page_ptr r3
4388 #define left_block_mask r4
4389 #define right_block_mask r5
4390 #define texture_mask_rev r10
4391 #define control_mask r11
4393 #define dirty_textures_mask r4
4395 #define current_texture_mask r6
4409 #define draw_masks_fb_ptrs q1
4411 #define draw_mask_fb_ptr_left d2
4412 #define draw_mask_fb_ptr_right d3
4414 #define draw_mask_fb_ptr_left_a d2
4415 #define draw_mask_fb_ptr_left_b d3
4416 #define draw_mask_fb_ptr_right_a d10
4417 #define draw_mask_fb_ptr_right_b d11
4418 #define draw_masks_fb_ptrs2 q5
4420 #define clut_low_a d4
4421 #define clut_low_b d5
4422 #define clut_high_a d6
4423 #define clut_high_b d7
4425 #define block_masks d8
4426 #define block_masks_shifted d9
4431 #define texels_low d12
4432 #define texels_high d13
4434 #define texels_wide_low d14
4435 #define texels_wide_high d15
4436 #define texels_wide q7
4439 setup_sprite_flush_blocks:
4442 stmdb sp!, { r0 - r3, r12, r14 }
4443 bl flush_render_block_buffer
4444 ldmia sp!, { r0 - r3, r12, r14 }
4448 add block, psx_gpu, #psx_gpu_blocks_offset
4452 setup_sprite_update_texture_4bpp_cache:
4453 stmdb sp!, { r0 - r3, r14 }
4454 bl update_texture_4bpp_cache
4455 ldmia sp!, { r0 - r3, pc }
4458 setup_sprite_update_texture_8bpp_cache:
4459 stmdb sp!, { r0 - r3, r14 }
4460 bl update_texture_8bpp_cache
4461 ldmia sp!, { r0 - r3, pc }
4464 #define setup_sprite_tiled_initialize_4bpp() \
4465 ldr dirty_textures_mask, \
4466 [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]; \
4467 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4469 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4470 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4472 tst current_texture_mask, dirty_textures_mask; \
4473 vuzp.u8 clut_a, clut_b; \
4475 blne setup_sprite_update_texture_4bpp_cache \
4477 #define setup_sprite_tiled_initialize_8bpp() \
4478 ldr dirty_textures_mask, \
4479 [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]; \
4480 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4482 tst current_texture_mask, dirty_textures_mask; \
4483 blne setup_sprite_update_texture_8bpp_cache \
4486 #define setup_sprite_block_count_single() \
4489 #define setup_sprite_block_count_double() \
4490 sub_tile_height, lsl #1 \
4492 #define setup_sprite_tile_add_blocks(type) \
4493 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4494 cmp num_blocks, #MAX_BLOCKS; \
4496 movgt num_blocks, setup_sprite_block_count_##type(); \
4497 blgt setup_sprite_flush_blocks \
4500 #define setup_sprite_tile_full_4bpp(edge) \
4501 setup_sprite_tile_add_blocks(double); \
4504 and texture_block_ptr, texture_offset, texture_mask; \
4505 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4508 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4509 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4511 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4512 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4514 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4515 add texture_block_ptr, texture_offset, #8; \
4517 and texture_block_ptr, texture_block_ptr, texture_mask; \
4518 add block, block, #40; \
4520 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4521 add fb_ptr, fb_ptr, #16; \
4523 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4524 add block, block, #24; \
4526 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4527 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4530 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4531 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4533 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4534 add block, block, #40; \
4536 add texture_offset, texture_offset, #0x10; \
4537 add fb_ptr, fb_ptr, #(2048 - 16); \
4539 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4540 add block, block, #24; \
4542 subs sub_tile_height, sub_tile_height, #1; \
4545 add texture_offset, texture_offset, #0xF00; \
4546 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4549 #define setup_sprite_tile_half_4bpp(edge) \
4550 setup_sprite_tile_add_blocks(single); \
4553 and texture_block_ptr, texture_offset, texture_mask; \
4554 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4557 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4558 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4560 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4561 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4563 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4564 add block, block, #40; \
4566 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4567 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4569 add block, block, #24; \
4570 add texture_offset, texture_offset, #0x10; \
4572 add fb_ptr, fb_ptr, #2048; \
4573 subs sub_tile_height, sub_tile_height, #1; \
4577 add texture_offset, texture_offset, #0xF00; \
4578 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4581 #define setup_sprite_tile_full_8bpp(edge) \
4582 setup_sprite_tile_add_blocks(double); \
4583 add block, block, #16; \
4586 and texture_block_ptr, texture_offset, texture_mask; \
4587 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4590 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4591 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4593 add texture_block_ptr, texture_offset, #8; \
4594 vst1.u32 { texels }, [ block, :64 ]; \
4596 and texture_block_ptr, texture_block_ptr, texture_mask; \
4597 add block, block, #24; \
4599 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4601 add fb_ptr, fb_ptr, #16; \
4602 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4604 add block, block, #40; \
4605 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4608 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4609 vst1.u32 { texels }, [ block, :64 ]; \
4610 add block, block, #24; \
4612 add texture_offset, texture_offset, #0x10; \
4613 add fb_ptr, fb_ptr, #(2048 - 16); \
4615 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4616 add block, block, #40; \
4618 subs sub_tile_height, sub_tile_height, #1; \
4621 sub block, block, #16; \
4622 add texture_offset, texture_offset, #0xF00; \
4623 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4626 #define setup_sprite_tile_half_8bpp(edge) \
4627 setup_sprite_tile_add_blocks(single); \
4628 add block, block, #16; \
4631 and texture_block_ptr, texture_offset, texture_mask; \
4632 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4635 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4636 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4638 vst1.u32 { texels }, [ block, :64 ]; \
4639 add block, block, #24; \
4641 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4642 add block, block, #40; \
4644 add texture_offset, texture_offset, #0x10; \
4645 add fb_ptr, fb_ptr, #2048; \
4647 subs sub_tile_height, sub_tile_height, #1; \
4650 sub block, block, #16; \
4651 add texture_offset, texture_offset, #0xF00; \
4652 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4655 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4656 add texture_offset, texture_offset_base, #8; \
4657 add fb_ptr, fb_ptr, #16 \
4659 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4660 mov texture_offset, texture_offset_base \
4662 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4663 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4665 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4666 mov texture_offset, texture_offset_base \
4668 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4669 sub fb_ptr, fb_ptr, #16 \
4671 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4673 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4674 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4676 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4679 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \
4681 mov sub_tile_height, column_data; \
4682 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4683 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4684 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4686 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \
4688 and sub_tile_height, column_data, #0xFF; \
4689 mov tiles_remaining, column_data, lsr #16; \
4690 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4691 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4693 subs tiles_remaining, tiles_remaining, #1; \
4697 mov sub_tile_height, #16; \
4698 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4699 subs tiles_remaining, tiles_remaining, #1; \
4703 uxtb sub_tile_height, column_data, ror #8; \
4704 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4705 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4708 #define setup_sprite_column_data_single() \
4709 mov column_data, height; \
4710 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] \
4712 #define setup_sprite_column_data_multi() \
4713 and height_rounded, height_rounded, #0xF; \
4714 rsb column_data, offset_v, #16; \
4716 add height_rounded, height_rounded, #1; \
4717 sub tile_height, tile_height, #1; \
4719 orr column_data, column_data, tile_height, lsl #16; \
4720 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]; \
4722 orr column_data, column_data, height_rounded, lsl #8 \
4724 #define setup_sprite_setup_left_draw_mask_fb_ptr() \
4725 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4726 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4728 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column() \
4729 mov fb_ptr_advance_column, #32; \
4730 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4732 sub fb_ptr_advance_column, height, lsl #11; \
4733 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4735 #define setup_sprite_setup_right_draw_mask_fb_ptr() \
4736 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4737 vdup.u8 draw_mask_fb_ptr_right, block_masks[5] \
4739 #define setup_sprite_tile_column_width_single(tm, multi_height, edge_mode, \
4741 setup_sprite_##tm##_single_##multi_height##_##edge_mode##_##edge##x4mode: \
4742 setup_sprite_column_data_##multi_height(); \
4743 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4744 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4745 setup_sprite_setup_left_draw_mask_fb_ptr##x4mode(); \
4747 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, tm, x4mode); \
4748 ldmia sp!, { r4 - r11, pc } \
4750 #define setup_sprite_tiled_advance_column() \
4751 add texture_offset_base, texture_offset_base, #0x100; \
4752 tst texture_offset_base, #0xF00; \
4753 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4755 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4756 right_mode, x4mode) \
4757 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode##x4mode:\
4758 setup_sprite_column_data_##multi_height(); \
4760 setup_sprite_setup_left_draw_mask_fb_ptr_advance_column##x4mode(); \
4762 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm, x4mode);\
4764 subs tile_width, tile_width, #2; \
4765 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4769 vmov.u8 draw_masks_fb_ptrs, #0; \
4770 vmov.u8 draw_masks_fb_ptrs2, #0; \
4773 setup_sprite_tiled_advance_column(); \
4774 setup_sprite_tile_column_height_##multi_height(full, none, tm, x4mode); \
4775 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4776 subs tile_width, tile_width, #1; \
4780 setup_sprite_setup_right_draw_mask_fb_ptr##x4mode(); \
4782 setup_sprite_tiled_advance_column(); \
4783 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm, x4mode);\
4784 ldmia sp!, { r4 - r11, pc } \
4787 #define setup_sprite_offset_u_adjust() \
4789 #define setup_sprite_get_left_block_mask() \
4790 and left_block_mask, left_block_mask, #0xFF \
4792 #define setup_sprite_compare_left_block_mask() \
4793 cmp left_block_mask, #0xFF \
4795 #define setup_sprite_get_right_block_mask() \
4796 uxtb right_block_mask, right_block_mask, ror #8 \
4798 #define setup_sprite_compare_right_block_mask() \
4799 cmp right_block_mask, #0xFF \
4804 #define fb_ptr2 column_data
4806 #define setup_sprite_offset_u_adjust_4x() \
4807 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4808 lsl offset_u_right, #1; \
4810 add offset_u_right, #1 \
4812 #define setup_sprite_get_left_block_mask_4x() \
4813 sxth left_block_mask, left_block_mask \
4815 #define setup_sprite_compare_left_block_mask_4x() \
4816 cmp left_block_mask, #0xFFFFFFFF \
4818 #define setup_sprite_get_right_block_mask_4x() \
4819 sxth right_block_mask, right_block_mask, ror #16 \
4821 #define setup_sprite_compare_right_block_mask_4x() \
4822 cmp right_block_mask, #0xFFFFFFFF \
4825 #define widen_texels_16bpp(texels_) \
4826 vmov texels_wide_low, texels_; \
4827 vmov texels_wide_high, texels_; \
4828 vzip.16 texels_wide_low, texels_wide_high \
4830 #define widen_texels_8bpp(texels_) \
4831 vmov texels_wide_low, texels_; \
4832 vmov texels_wide_high, texels_; \
4833 vzip.8 texels_wide_low, texels_wide_high \
4835 #define write_block_16bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4836 vst1.u32 { texels_ }, [ block_, :128 ]; \
4837 add block_, block_, #40; \
4839 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4840 vst1.u32 { draw_mask_fb_ptr_ }, [ block_, :64 ]; \
4841 add block_, block_, #24 \
4843 /* assumes 16-byte offset already added to block_ */
4844 #define write_block_8bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4845 vst1.u32 { texels_ }, [ block_, :64 ]; \
4846 add block_, block_, #24; \
4848 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4849 vst1.u32 { draw_mask_fb_ptr_ }, [ block_, :64 ]; \
4850 add block_, block_, #40 \
4852 #define do_texture_block_16bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4853 draw_mask_fb_ptr_b_) \
4854 widen_texels_16bpp(texels_low); \
4855 add fb_ptr_tmp, fb_ptr, #1024*2; \
4857 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr); \
4859 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4860 widen_texels_16bpp(texels_high); \
4862 add fb_ptr_tmp, fb_ptr, #8*2; \
4863 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4865 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4866 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4868 #define do_texture_block_8bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4869 draw_mask_fb_ptr_b_) \
4870 widen_texels_8bpp(texels); \
4871 add fb_ptr_tmp, fb_ptr, #1024*2; \
4873 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr); \
4874 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4876 add fb_ptr_tmp, fb_ptr, #8*2; \
4877 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4879 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4880 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4883 #define setup_sprite_tiled_initialize_4bpp_4x() \
4884 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4885 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4887 vuzp.u8 clut_a, clut_b \
4889 #define setup_sprite_tiled_initialize_8bpp_4x() \
4892 #define setup_sprite_block_count_single_4x() \
4893 sub_tile_height, lsl #2 \
4895 #define setup_sprite_block_count_double_4x() \
4896 sub_tile_height, lsl #(1+2) \
4898 #define setup_sprite_tile_full_4bpp_4x(edge) \
4899 setup_sprite_tile_add_blocks(double_4x); \
4900 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4903 and texture_block_ptr, texture_offset, texture_mask; \
4906 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4907 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4909 add texture_block_ptr, texture_offset, #8; \
4910 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4912 and texture_block_ptr, texture_block_ptr, texture_mask; \
4913 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4915 vzip.8 texels_low, texels_high; \
4916 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
4917 draw_mask_fb_ptr_left_b); \
4919 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4920 add fb_ptr, fb_ptr, #16*2; \
4922 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4923 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4926 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4928 vzip.8 texels_low, texels_high; \
4929 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
4930 draw_mask_fb_ptr_right_b); \
4932 add texture_offset, texture_offset, #0x10; \
4933 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
4935 subs sub_tile_height, sub_tile_height, #1; \
4938 ldr column_data, [sp], #8; /* fb_ptr2 */ \
4939 add texture_offset, texture_offset, #0xF00; \
4940 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4943 #define setup_sprite_tile_half_4bpp_4x(edge) \
4944 setup_sprite_tile_add_blocks(single_4x); \
4945 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4948 and texture_block_ptr, texture_offset, texture_mask; \
4951 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4952 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4954 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4955 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4957 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4958 add texture_offset, texture_offset, #0x10; \
4960 vzip.8 texels_low, texels_high; \
4961 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
4962 draw_mask_fb_ptr_##edge##_b); \
4964 add fb_ptr, fb_ptr, #2048 * 2; \
4965 subs sub_tile_height, sub_tile_height, #1; \
4969 ldr column_data, [sp], #8; /* fb_ptr2 */ \
4970 add texture_offset, texture_offset, #0xF00; \
4971 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4974 #define setup_sprite_tile_full_8bpp_4x(edge) \
4975 setup_sprite_tile_add_blocks(double_4x); \
4976 add block, block, #16; \
4977 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4980 and texture_block_ptr, texture_offset, texture_mask; \
4983 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4984 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4986 add texture_block_ptr, texture_offset, #8; \
4987 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
4988 draw_mask_fb_ptr_left_b); \
4990 and texture_block_ptr, texture_block_ptr, texture_mask; \
4992 add fb_ptr, fb_ptr, #16*2; \
4993 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4995 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4998 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
4999 draw_mask_fb_ptr_right_b); \
5001 add texture_offset, texture_offset, #0x10; \
5002 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
5004 subs sub_tile_height, sub_tile_height, #1; \
5007 sub block, block, #16; \
5008 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5009 add texture_offset, texture_offset, #0xF00; \
5010 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
5013 #define setup_sprite_tile_half_8bpp_4x(edge) \
5014 setup_sprite_tile_add_blocks(single_4x); \
5015 add block, block, #16; \
5016 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5019 and texture_block_ptr, texture_offset, texture_mask; \
5022 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5023 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
5025 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5026 draw_mask_fb_ptr_##edge##_b); \
5028 add texture_offset, texture_offset, #0x10; \
5029 add fb_ptr, fb_ptr, #2048 * 2; \
5031 subs sub_tile_height, sub_tile_height, #1; \
5034 sub block, block, #16; \
5035 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5036 add texture_offset, texture_offset, #0xF00; \
5037 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
5040 #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \
5041 add texture_offset, texture_offset_base, #8; \
5042 add fb_ptr, fb_ptr, #16 * 2 \
5044 #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \
5045 mov texture_offset, texture_offset_base \
5047 #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \
5048 setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \
5050 #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \
5051 mov texture_offset, texture_offset_base \
5053 #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \
5054 sub fb_ptr, fb_ptr, #16 * 2 \
5056 #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \
5058 #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \
5059 setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \
5061 #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \
5064 #define setup_sprite_setup_left_draw_mask_fb_ptr_4x() \
5065 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5066 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5067 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5068 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5070 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column_4x() \
5071 mov fb_ptr_advance_column, #32 * 2; \
5072 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5073 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5074 sub fb_ptr_advance_column, height, lsl #11 + 1; \
5075 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5076 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5078 #define setup_sprite_setup_right_draw_mask_fb_ptr_4x() \
5079 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[4]; \
5080 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[5]; \
5081 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[6]; \
5082 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[7] \
5090 // [ sp + 4 ]: width
5091 // [ sp + 8 ]: height
5092 // [ sp + 12 ]: color (unused)
5094 #define setup_sprite_tiled_builder(texture_mode, x4mode) \
5096 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \
5098 setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \
5100 setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \
5102 setup_sprite_tile_column_width_single(texture_mode, single, full, none, \
5104 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \
5106 setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \
5108 setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \
5110 setup_sprite_tile_column_width_single(texture_mode, single, half, right, \
5112 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \
5114 setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \
5116 setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \
5118 setup_sprite_tile_column_width_single(texture_mode, single, half, left, \
5120 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \
5122 setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \
5127 function(setup_sprite_##texture_mode##x4mode) \
5128 stmdb sp!, { r4 - r11, r14 }; \
5129 setup_sprite_tiled_initialize_##texture_mode##x4mode(); \
5131 ldr v, [ sp, #36 ]; \
5132 and offset_u, u, #0xF; \
5134 ldr width, [ sp, #40 ]; \
5135 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
5137 ldr height, [ sp, #44 ]; \
5138 add fb_ptr, fb_ptr, y, lsl #11; \
5140 add fb_ptr, fb_ptr, x, lsl #1; \
5141 and offset_v, v, #0xF; \
5143 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
5144 add width_rounded, offset_u, width; \
5146 add height_rounded, offset_v, height; \
5147 add width_rounded, width_rounded, #15; \
5149 add height_rounded, height_rounded, #15; \
5150 mov tile_width, width_rounded, lsr #4; \
5152 /* texture_offset_base = VH-VL-00-00 */\
5153 mov texture_offset_base, v, lsl #8; \
5154 and offset_u_right, width_rounded, #0xF; \
5156 /* texture_offset_base = VH-UH-UL-00 */\
5157 bfi texture_offset_base, u, #4, #8; \
5158 mov right_block_mask, #0xFFFFFFFE; \
5160 setup_sprite_offset_u_adjust##x4mode(); \
5162 /* texture_offset_base = VH-UH-VL-00 */\
5163 bfi texture_offset_base, v, #4, #4; \
5164 mov left_block_mask, #0xFFFFFFFF; \
5166 mov tile_height, height_rounded, lsr #4; \
5167 mvn left_block_mask, left_block_mask, lsl offset_u; \
5169 /* texture_mask = HH-HL-WH-WL */\
5170 ldrh texture_mask, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]; \
5171 mov right_block_mask, right_block_mask, lsl offset_u_right; \
5173 /* texture_mask_rev = WH-WL-HH-HL */\
5174 rev16 texture_mask_rev, texture_mask; \
5175 vmov block_masks, left_block_mask, right_block_mask; \
5177 /* texture_mask = HH-HL-HL-WL */\
5178 bfi texture_mask, texture_mask_rev, #4, #4; \
5179 /* texture_mask_rev = 00-00-00-WH */\
5180 mov texture_mask_rev, texture_mask_rev, lsr #12; \
5182 /* texture_mask = HH-WH-HL-WL */\
5183 bfi texture_mask, texture_mask_rev, #8, #4; \
5184 setup_sprite_get_left_block_mask##x4mode(); \
5186 mov control_mask, #0; \
5187 setup_sprite_compare_left_block_mask##x4mode(); \
5189 setup_sprite_get_right_block_mask##x4mode(); \
5190 orreq control_mask, control_mask, #0x4; \
5192 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
5193 setup_sprite_compare_right_block_mask##x4mode(); \
5195 orreq control_mask, control_mask, #0x8; \
5196 cmp tile_width, #1; \
5198 add block, psx_gpu, #psx_gpu_blocks_offset; \
5199 orreq control_mask, control_mask, #0x1; \
5201 cmp tile_height, #1; \
5202 add block, block, num_blocks, lsl #6; \
5204 orreq control_mask, control_mask, #0x2; \
5205 ldr pc, [ pc, control_mask, lsl #2 ]; \
5208 .word setup_sprite_##texture_mode##_multi_multi_full_full##x4mode; \
5209 .word setup_sprite_##texture_mode##_single_multi_full_none##x4mode; \
5210 .word setup_sprite_##texture_mode##_multi_single_full_full##x4mode; \
5211 .word setup_sprite_##texture_mode##_single_single_full_none##x4mode; \
5212 .word setup_sprite_##texture_mode##_multi_multi_half_full##x4mode; \
5213 .word setup_sprite_##texture_mode##_single_multi_half_right##x4mode; \
5214 .word setup_sprite_##texture_mode##_multi_single_half_full##x4mode; \
5215 .word setup_sprite_##texture_mode##_single_single_half_right##x4mode; \
5216 .word setup_sprite_##texture_mode##_multi_multi_full_half##x4mode; \
5217 .word setup_sprite_##texture_mode##_single_multi_half_left##x4mode; \
5218 .word setup_sprite_##texture_mode##_multi_single_full_half##x4mode; \
5219 .word setup_sprite_##texture_mode##_single_single_half_left##x4mode; \
5220 .word setup_sprite_##texture_mode##_multi_multi_half_half##x4mode; \
5222 .word setup_sprite_##texture_mode##_multi_single_half_half##x4mode; \
5225 setup_sprite_tiled_builder(4bpp,);
5226 setup_sprite_tiled_builder(8bpp,);
5228 #undef draw_mask_fb_ptr_left
5229 #undef draw_mask_fb_ptr_right
5231 setup_sprite_tiled_builder(4bpp, _4x);
5232 setup_sprite_tiled_builder(8bpp, _4x);
5240 #define block_ptr r0
5241 #define num_blocks r1
5243 #define texel_shift_mask r3
5244 #define block_pixels_a r4
5245 #define block_pixels_b r5
5254 #define texels_01 r6
5255 #define texels_23 r7
5256 #define texels_45 r8
5257 #define texels_67 r9
5259 function(texture_sprite_blocks_8bpp)
5260 stmdb sp!, { r4 - r11, r14 }
5261 movw texel_shift_mask, #(0xFF << 1)
5263 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5264 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
5266 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
5267 ldr block_pixels_a, [ block_ptr, #16 ]
5270 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
5271 ldr block_pixels_b, [ block_ptr, #20 ]
5273 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
5274 ldrh texel_0, [ clut_ptr, texel_0 ]
5276 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
5277 ldrh texel_1, [ clut_ptr, texel_1 ]
5279 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
5280 ldr block_pixels_a, [ block_ptr, #(64 + 16) ]
5282 ldrh texel_2, [ clut_ptr, texel_2 ]
5283 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
5285 ldrh texel_3, [ clut_ptr, texel_3 ]
5286 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
5288 ldrh texel_4, [ clut_ptr, texel_4 ]
5289 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
5291 ldrh texel_5, [ clut_ptr, texel_5 ]
5292 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
5294 ldrh texel_6, [ clut_ptr, texel_6 ]
5295 orr texels_01, texel_0, texel_1, lsl #16
5297 ldrh texel_7, [ clut_ptr, texel_7 ]
5298 orr texels_23, texel_2, texel_3, lsl #16
5300 orr texels_45, texel_4, texel_5, lsl #16
5301 str texels_01, [ block_ptr, #0 ]
5303 orr texels_67, texel_6, texel_7, lsl #16
5304 str texels_23, [ block_ptr, #4 ]
5306 subs num_blocks, num_blocks, #1
5307 str texels_45, [ block_ptr, #8 ]
5309 str texels_67, [ block_ptr, #12 ]
5310 add block_ptr, block_ptr, #64
5314 ldmia sp!, { r4 - r11, pc }
5317 #undef width_rounded
5320 #undef texture_offset
5323 #undef texels_wide_low
5324 #undef texels_wide_high
5335 #define left_offset r8
5336 #define width_rounded r9
5337 #define right_width r10
5339 #define block_width r11
5341 #define texture_offset_base r1
5342 #define texture_mask r2
5343 #define texture_page_ptr r3
5344 #define num_blocks r4
5347 #define texture_offset r8
5348 #define blocks_remaining r9
5350 #define fb_ptr_pitch r12
5351 #define texture_block_ptr r14
5353 #define texture_mask_width r2
5354 #define texture_mask_height r3
5355 #define left_mask_bits r4
5356 #define right_mask_bits r5
5360 #undef block_masks_shifted
5363 #define block_masks d0
5364 #define block_masks_shifted d1
5365 #define draw_mask_fb_ptr d2
5368 #define draw_mask_fb_ptr_a d2
5369 #define draw_mask_fb_ptr_b d3
5370 #define texels_low d4
5371 #define texels_high d5
5372 #define texels_wide_low d6
5373 #define texels_wide_high d7
5374 #define texels_wide q3
5377 setup_sprites_16bpp_flush:
5380 stmdb sp!, { r0 - r3, r12, r14 }
5381 bl flush_render_block_buffer
5382 ldmia sp!, { r0 - r3, r12, r14 }
5386 add block, psx_gpu, #psx_gpu_blocks_offset
5387 mov num_blocks, block_width
5391 function(setup_sprite_16bpp)
5392 stmdb sp!, { r4 - r11, r14 }
5393 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5396 add fb_ptr, fb_ptr, y, lsl #11
5398 ldr width, [ sp, #40 ]
5399 add fb_ptr, fb_ptr, x, lsl #1
5401 ldr height, [ sp, #44 ]
5402 and left_offset, u, #0x7
5404 add texture_offset_base, u, u
5405 add width_rounded, width, #7
5407 add texture_offset_base, v, lsl #11
5408 mov left_mask_bits, #0xFF
5410 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5411 add width_rounded, width_rounded, left_offset
5413 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5414 sub fb_ptr, fb_ptr, left_offset, lsl #1
5416 add texture_mask, texture_mask_width, texture_mask_width
5417 mov right_mask_bits, #0xFE
5419 and right_width, width_rounded, #0x7
5420 mvn left_mask_bits, left_mask_bits, lsl left_offset
5422 add texture_mask, texture_mask_height, lsl #11
5423 mov block_width, width_rounded, lsr #3
5425 mov right_mask_bits, right_mask_bits, lsl right_width
5426 movw fb_ptr_pitch, #(2048 + 16)
5428 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5429 vmov block_masks, left_mask_bits, right_mask_bits
5431 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5432 add block, psx_gpu, #psx_gpu_blocks_offset
5434 bic texture_offset_base, texture_offset_base, #0xF
5437 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5438 add block, block, num_blocks, lsl #6
5442 vext.32 block_masks_shifted, block_masks, block_masks, #1
5443 vorr.u32 block_masks, block_masks, block_masks_shifted
5444 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5447 add num_blocks, num_blocks, #1
5448 cmp num_blocks, #MAX_BLOCKS
5449 blgt setup_sprites_16bpp_flush
5451 and texture_block_ptr, texture_offset_base, texture_mask
5452 subs height, height, #1
5454 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5455 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5457 vst1.u32 { texels }, [ block, :128 ]
5458 add block, block, #40
5460 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5463 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5465 add block, block, #24
5466 add texture_offset_base, texture_offset_base, #2048
5467 add fb_ptr, fb_ptr, #2048
5468 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5471 ldmia sp!, { r4 - r11, pc }
5474 add num_blocks, num_blocks, block_width
5475 mov texture_offset, texture_offset_base
5477 cmp num_blocks, #MAX_BLOCKS
5478 blgt setup_sprites_16bpp_flush
5480 add texture_offset_base, texture_offset_base, #2048
5481 and texture_block_ptr, texture_offset, texture_mask
5483 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5484 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5486 vst1.u32 { texels }, [ block, :128 ]
5487 add block, block, #40
5489 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5490 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5493 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5494 subs blocks_remaining, block_width, #2
5496 add texture_offset, texture_offset, #16
5497 add fb_ptr, fb_ptr, #16
5499 vmov.u8 draw_mask_fb_ptr, #0
5501 add block, block, #24
5505 and texture_block_ptr, texture_offset, texture_mask
5506 subs blocks_remaining, blocks_remaining, #1
5508 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5509 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5511 vst1.u32 { texels }, [ block, :128 ]
5512 add block, block, #40
5514 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5517 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5519 add texture_offset, texture_offset, #16
5520 add fb_ptr, fb_ptr, #16
5522 add block, block, #24
5526 and texture_block_ptr, texture_offset, texture_mask
5527 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5529 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5530 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5532 vst1.u32 { texels }, [ block, :128 ]
5533 add block, block, #40
5535 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5536 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5538 add block, block, #24
5539 subs height, height, #1
5541 add fb_ptr, fb_ptr, fb_ptr_pitch
5542 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5546 ldmia sp!, { r4 - r11, pc }
5550 // FIXME: duplicate code with normal version :(
5551 #undef draw_mask_fb_ptr
5553 function(setup_sprite_16bpp_4x)
5554 stmdb sp!, { r4 - r11, r14 }
5555 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5558 add fb_ptr, fb_ptr, y, lsl #11
5560 ldr width, [ sp, #40 ]
5561 add fb_ptr, fb_ptr, x, lsl #1
5563 ldr height, [ sp, #44 ]
5564 and left_offset, u, #0x7
5566 add texture_offset_base, u, u
5567 add width_rounded, width, #7
5569 add texture_offset_base, v, lsl #11
5570 movw left_mask_bits, #0xFFFF
5572 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5573 add width_rounded, width_rounded, left_offset
5577 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5578 sub fb_ptr, fb_ptr, left_offset, lsl #1
5580 add texture_mask, texture_mask_width, texture_mask_width
5581 movw right_mask_bits, #0xFFFC
5583 and right_width, width_rounded, #0x7
5584 mvn left_mask_bits, left_mask_bits, lsl left_offset
5588 add texture_mask, texture_mask_height, lsl #11
5589 mov block_width, width_rounded, lsr #3
5591 mov right_mask_bits, right_mask_bits, lsl right_width
5592 movw fb_ptr_pitch, #(2048 + 16) * 2
5594 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4+1
5595 vmov block_masks, left_mask_bits, right_mask_bits
5597 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5598 add block, psx_gpu, #psx_gpu_blocks_offset
5600 bic texture_offset_base, texture_offset_base, #0xF
5603 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5604 add block, block, num_blocks, lsl #6
5609 vext.32 block_masks_shifted, block_masks, block_masks, #1
5610 vorr.u32 block_masks, block_masks, block_masks_shifted
5611 vdup.u8 draw_mask_fb_ptr_a, block_masks[0]
5612 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5615 add num_blocks, num_blocks, block_width
5616 cmp num_blocks, #MAX_BLOCKS
5617 blgt setup_sprites_16bpp_flush
5619 and texture_block_ptr, texture_offset_base, texture_mask
5620 subs height, height, #1
5622 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5623 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5625 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5627 add texture_offset_base, texture_offset_base, #2048
5628 add fb_ptr, fb_ptr, #2048*2
5629 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5632 ldmia sp!, { r4 - r11, pc }
5635 add num_blocks, num_blocks, block_width
5636 mov texture_offset, texture_offset_base
5638 vdup.u8 draw_mask_fb_ptr_a, block_masks[0] // left_mask_bits
5639 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5641 cmp num_blocks, #MAX_BLOCKS
5642 blgt setup_sprites_16bpp_flush
5644 add texture_offset_base, texture_offset_base, #2048
5645 and texture_block_ptr, texture_offset, texture_mask
5647 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5648 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5650 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5652 subs blocks_remaining, block_width, #2*4
5653 add texture_offset, texture_offset, #16
5655 vmov.u8 draw_mask_fb_ptr_a, #0
5656 vmov.u8 draw_mask_fb_ptr_b, #0
5658 add fb_ptr, fb_ptr, #16*2
5662 and texture_block_ptr, texture_offset, texture_mask
5663 subs blocks_remaining, blocks_remaining, #4
5665 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5666 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5668 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5669 add texture_offset, texture_offset, #16
5671 add fb_ptr, fb_ptr, #16*2
5675 vdup.u8 draw_mask_fb_ptr_a, block_masks[4] // right_mask_bits
5676 vdup.u8 draw_mask_fb_ptr_b, block_masks[5]
5678 and texture_block_ptr, texture_offset, texture_mask
5679 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5681 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5683 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5684 subs height, height, #1
5686 add fb_ptr, fb_ptr, fb_ptr_pitch
5687 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5691 ldmia sp!, { r4 - r11, pc }
5696 #undef right_mask_bits
5699 #undef blocks_remaining
5709 #define right_width r5
5710 #define right_mask_bits r6
5714 #define fb_ptr_pitch r12
5716 // referenced by setup_sprites_16bpp_flush
5717 #define num_blocks r4
5719 #define block_width r11
5724 #define blocks_remaining r6
5727 #define right_mask q1
5728 #define test_mask q2
5729 #define draw_mask q2
5730 #define draw_mask_bits_fb_ptr d6
5735 function(setup_sprite_untextured)
5736 ldrh r12, [ psx_gpu, #psx_gpu_render_state_offset ]
5737 tst r12, #(RENDER_STATE_MASK_EVALUATE | RENDER_FLAGS_MODULATE_TEXELS \
5738 | RENDER_FLAGS_BLEND)
5739 beq setup_sprite_untextured_simple
5741 stmdb sp!, { r4 - r11, r14 }
5743 ldr width, [ sp, #40 ]
5744 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5746 ldr height, [ sp, #44 ]
5747 add fb_ptr, fb_ptr, y, lsl #11
5749 add fb_ptr, fb_ptr, x, lsl #1
5750 sub right_width, width, #1
5752 ldr color, [ sp, #48 ]
5755 add block_width, width, #7
5759 mov right_mask_bits, #0xff
5761 sub fb_ptr_pitch, block_width, #1
5762 lsl right_mask_bits, right_width
5764 lsl fb_ptr_pitch, #3+1
5765 ubfx color_r, color, #3, #5
5767 rsb fb_ptr_pitch, #1024*2
5768 ubfx color_g, color, #11, #5
5770 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
5771 ubfx color_b, color, #19, #5
5773 vdup.u16 right_mask, right_mask_bits
5774 orr color, color_r, color_b, lsl #10
5776 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5777 orr color, color, color_g, lsl #5
5779 vtst.u16 right_mask, right_mask, test_mask
5780 add block, psx_gpu, #psx_gpu_blocks_offset
5782 vdup.u16 colors, color
5783 add block, block, num_blocks, lsl #6
5786 setup_sprite_untextured_height_loop:
5787 add num_blocks, block_width
5788 sub blocks_remaining, block_width, #1
5790 cmp num_blocks, #MAX_BLOCKS
5791 blgt setup_sprites_16bpp_flush
5793 cmp blocks_remaining, #0
5796 vmov.u8 draw_mask, #0 /* zero_mask */
5797 vmov.u8 draw_mask_bits_fb_ptr, #0
5800 vst1.u32 { draw_mask }, [ block, :128 ]!
5801 subs blocks_remaining, #1
5803 vst1.u32 { colors }, [ block, :128 ]
5804 add block, block, #24
5806 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5807 vst1.u32 { draw_mask_bits_fb_ptr }, [ block, :64 ]
5809 add block, block, #24
5814 vst1.u32 { right_mask }, [ block, :128 ]!
5817 vst1.u32 { colors }, [ block, :128 ]
5818 add block, block, #24
5820 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5821 vst1.u32 { draw_mask_bits_fb_ptr }, [ block, :64 ]
5823 add block, block, #24
5824 add fb_ptr, fb_ptr_pitch
5826 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5827 bgt setup_sprite_untextured_height_loop
5829 ldmia sp!, { r4 - r11, pc }
5833 #undef texture_page_ptr
5835 #undef dirty_textures_mask
5836 #undef current_texture_mask
5839 #define current_texture_page r1
5840 #define texture_page_ptr r2
5841 #define vram_ptr_a r3
5842 #define current_texture_page_x r12
5843 #define current_texture_page_y r4
5844 #define dirty_textures_mask r5
5848 #define current_texture_mask r9
5850 #define vram_ptr_b r11
5852 #define texel_block_a d0
5853 #define texel_block_b d1
5854 #define texel_block_expanded_a q1
5855 #define texel_block_expanded_b q2
5856 #define texel_block_expanded_ab q2
5857 #define texel_block_expanded_c q3
5858 #define texel_block_expanded_d q4
5859 #define texel_block_expanded_cd q3
5861 function(update_texture_4bpp_cache)
5862 stmdb sp!, { r4 - r11, r14 }
5865 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5867 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5868 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5870 and current_texture_page_x, current_texture_page, #0xF
5871 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
5873 mov current_texture_page_y, current_texture_page, lsr #4
5874 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5876 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5879 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5880 bic dirty_textures_mask, current_texture_mask
5883 str dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5888 add vram_ptr_b, vram_ptr_a, #2048
5891 vld1.u32 { texel_block_a }, [ vram_ptr_a, :64 ], c_4096
5892 vld1.u32 { texel_block_b }, [ vram_ptr_b, :64 ], c_4096
5894 vmovl.u8 texel_block_expanded_a, texel_block_a
5895 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5896 vmovl.u8 texel_block_expanded_c, texel_block_b
5897 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5899 vbic.u16 texel_block_expanded_a, #0x00F0
5900 vbic.u16 texel_block_expanded_b, #0x00F0
5901 vbic.u16 texel_block_expanded_c, #0x00F0
5902 vbic.u16 texel_block_expanded_d, #0x00F0
5904 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
5905 texel_block_expanded_b
5906 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
5907 texel_block_expanded_d
5909 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
5910 [ texture_page_ptr, :256 ]!
5912 subs sub_y, sub_y, #1
5916 add vram_ptr_a, vram_ptr_a, #8
5917 add vram_ptr_b, vram_ptr_b, #8
5919 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5920 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5922 subs tile_x, tile_x, #1
5926 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5927 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5929 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5930 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5932 subs tile_y, tile_y, #1
5936 ldmia sp!, { r4 - r11, pc }
5939 #undef current_texture_page
5942 #define texture_page r1
5943 #define texture_page_ptr r2
5944 #define vram_ptr_a r3
5945 #define texture_page_x r12
5946 #define texture_page_y r4
5947 #define current_texture_page r5
5952 #define vram_ptr_b r11
5964 function(update_texture_8bpp_cache_slice)
5965 stmdb sp!, { r4 - r11, r14 }
5968 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5969 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5971 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5974 and texture_page_x, texture_page, #0xF
5975 mov texture_page_y, texture_page, lsr #4
5977 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
5980 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
5981 eor current_texture_page, current_texture_page, texture_page
5983 ands current_texture_page, current_texture_page, #0x1
5986 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
5989 add vram_ptr_b, vram_ptr_a, #2048
5992 vld1.u32 { texels_a }, [ vram_ptr_a, :128 ], c_4096
5993 vld1.u32 { texels_b }, [ vram_ptr_b, :128 ], c_4096
5994 vld1.u32 { texels_c }, [ vram_ptr_a, :128 ], c_4096
5995 vld1.u32 { texels_d }, [ vram_ptr_b, :128 ], c_4096
5997 vst1.u32 { texels_a, texels_b }, [ texture_page_ptr, :256 ]!
5998 vst1.u32 { texels_c, texels_d }, [ texture_page_ptr, :256 ]!
6000 subs sub_y, sub_y, #1
6005 add vram_ptr_a, vram_ptr_a, #16
6006 add vram_ptr_b, vram_ptr_b, #16
6008 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
6009 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
6011 subs tile_x, tile_x, #1
6016 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
6017 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
6019 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
6020 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
6022 subs tile_y, tile_y, #1
6023 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6028 ldmia sp!, { r4 - r11, pc }
6031 /* void scale2x_tiles8(void *dst, const void *src, int w8, int h) */
6032 function(scale2x_tiles8)
6036 add r12, r0, #1024*2
6040 vld1.u16 { q0 }, [ r1, :128 ]!
6041 vld1.u16 { q2 }, [ r1, :128 ]!
6047 vst1.u16 { q0, q1 }, [ r0, :128 ]!
6048 vst1.u16 { q0, q1 }, [ r12, :128 ]!
6050 vst1.u16 { q2, q3 }, [ r0, :128 ]!
6051 vst1.u16 { q2, q3 }, [ r12, :128 ]!
6058 sub r0, r2, lsl #4+1
6060 add r12, r0, #1024*2
6066 // vim:filetype=armasm