2 * Copyright (C) 2011 Gilead Kutnick "Exophase" <exophase@gmail.com>
3 * Copyright (C) 2012 GraÅžvydas Ignotas "notaz" <notasas@gmail.com>
5 * This program is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU General Public License as
7 * published by the Free Software Foundation; either version 2 of
8 * the License, or (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License for more details.
18 #define MAX_BLOCKS_PER_ROW 128
20 #define RENDER_STATE_MASK_EVALUATE 0x20
21 #define RENDER_FLAGS_MODULATE_TEXELS 0x1
22 #define RENDER_FLAGS_BLEND 0x2
23 #define RENDER_INTERLACE_ENABLED 0x1
25 #include "psx_gpu_offsets.h"
27 #define psx_gpu_b_dx_offset (psx_gpu_b_block_span_offset + 4)
29 #define edge_data_left_x_offset 0
30 #define edge_data_num_blocks_offset 2
31 #define edge_data_right_mask_offset 4
32 #define edge_data_y_offset 6
78 #define gw_by_l gw_bx_l
103 #define uvrg_xxxx1 q4
108 #define uvrg_xxxx2 q5
113 #define yyyy_uvrg0 q6
118 #define yyyy_uvrg1 q7
123 #define yyyy_uvrg2 q8
150 #define uvrg_base q11
193 #define uvrgb_phase q13
197 #define function(name) \
206 function(compute_all_gradients)
207 // First compute the triangle area reciprocal and shift. The division will
208 // happen concurrently with much of the work which follows.
209 @ r12 = psx_gpu->triangle_area
210 ldr r12, [ psx_gpu, #psx_gpu_triangle_area_offset ]
211 stmdb sp!, { r4 - r11, lr }
213 @ load exponent of 62 into upper half of double
215 clz r14, r12 @ r14 = shift
217 movt r4, #((62 + 1023) << 4)
218 mov r12, r12, lsl r14 @ r12 = triangle_area_normalized
220 @ load area normalized into lower half of double
222 vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n
224 movt r4, #((1022 + 31) << 4)
227 add r4, r4, r12, lsr #11
230 vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n
232 // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) =
233 // ( d0 * d1 ) - ( d2 * d3 ) =
234 // ( m0 ) - ( m1 ) = gradient
236 // This is split to do 12 elements at a time over three sets: a, b, and c.
237 // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so
238 // two of the slots are unused.
240 // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as
243 // First type is: uvrg bxxx xxxx
244 // Second type is: yyyy ybyy uvrg
245 // Since x_a and y_c are the same the same variable is used for both.
247 vld1.u32 { v0 }, [ v_a, : 128 ] @ v0 = { uvrg0, b0, x0, y0 }
248 ldrsh x0, [ v_a, #8 ] @ load x0
250 vld1.u32 { v1 }, [ v_b, : 128 ] @ v1 = { uvrg1, b1, x1, y1}
251 ldrh x1, [ v_b, #8 ] @ load x1
253 vld1.u32 { v2 }, [ v_c, : 128 ] @ v2 = { uvrg2, b2, x2, y2 }
254 ldrh x2, [ v_c, #8 ] @ load x2
256 vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- }
257 ldrh y0, [ v_a, #10 ] @ load y0
259 vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- }
260 ldrh y1, [ v_b, #10 ] @ load y1
262 vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- }
263 ldrh y2, [ v_c, #10 ] @ load y2
265 vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 }
266 vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 }
268 orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 }
269 pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 }
271 vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 }
272 vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 }
274 vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 }
275 vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 }
277 ldrb b2, [ v_c, #4 ] @ load b2
278 orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 }
280 ldrb b1, [ v_b, #4 ] @ load b1
281 orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 }
283 vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 }
284 vsub.s16 d0_ab, x1_ab, x0_ab
286 ldrb b0, [ v_a, #4 ] @ load b0
287 orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 }
289 vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 }
290 vsub.s16 d2_ab, x2_ab, x1_ab
292 vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 }
293 vsub.s16 d1_ab, y2_ab, y1_ab
295 orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 }
296 ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 }
298 ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 }
299 ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 }
301 vsub.s16 d3_ab, y1_ab, y0_ab
302 smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) -
303 @ ((x2 - X1) * (b1 - b0))
304 vmull.s16 ga_uvrg_x, d0_a, d1_a
305 smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) -
306 @ ((b2 - b1) * (y1 - y0))
307 vmlsl.s16 ga_uvrg_x, d2_a, d3_a
308 movs gs_bx, ga_bx, asr #31
310 vmull.s16 ga_uvrg_y, d0_b, d1_b
311 rsbmi ga_bx, ga_bx, #0
313 @ r12 = psx_gpu->uvrgb_phase
314 ldr r12, [ psx_gpu, #psx_gpu_uvrgb_phase_offset ]
316 vmlsl.s16 ga_uvrg_y, d2_b, d3_b
317 movs gs_by, ga_by, asr #31
319 vshr.u64 d0, d30, #22
320 add b_base, r12, b0, lsl #16
322 vdup.u32 uvrgb_phase, r12
324 rsbmi ga_by, ga_by, #0
325 vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0
327 @ r12 = psx_gpu->triangle_winding_offset
328 ldrb r12, [ psx_gpu, #psx_gpu_triangle_winding_offset ]
329 vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0
331 rsb r12, r12, #0 @ r12 = -(triangle->winding)
333 vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w }
334 sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS)
336 vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16
337 vdup.u32 r_shift, r14 @ r_shift = { shift, shift, shift, shift }
339 vadd.u32 uvrg_base, uvrgb_phase
340 vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x)
342 vmov area_r_s, s0 @ area_r_s = triangle_reciprocal
343 vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y)
345 vmull.u32 gw_rg_x, ga_rg_x, d0[0]
346 vmull.u32 gw_uv_x, ga_uv_x, d0[0]
347 vmull.u32 gw_rg_y, ga_rg_y, d0[0]
348 vmull.u32 gw_uv_y, ga_uv_y, d0[0]
350 vshl.u64 gw_rg_x, gw_rg_x, r_shift
351 vshl.u64 gw_uv_x, gw_uv_x, r_shift
352 vshl.u64 gw_rg_y, gw_rg_y, r_shift
353 vshl.u64 gw_uv_y, gw_uv_y, r_shift
355 veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask
356 vmovn.u64 g_uv_x, gw_uv_x
358 veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask
359 vmovn.u64 g_rg_x, gw_rg_x
361 veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
362 vmovn.u64 g_uv_y, gw_uv_y
364 vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
365 vmovn.u64 g_rg_y, gw_rg_y
367 veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
368 mov ga_bx, ga_bx, lsl #13
370 vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
371 mov ga_by, ga_by, lsl #13
374 umull gw_bx_l, gw_bx_h, ga_bx, area_r_s
376 vshl.u32 g_uvrg_x, g_uvrg_x, #4
377 vshl.u32 g_uvrg_y, g_uvrg_y, #4
379 umull gw_by_l, gw_by_h, ga_by, area_r_s
380 vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0]
382 eor gs_bx, gs_bx, r12
383 vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1
385 veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0
386 eor gs_by, gs_by, r12
388 rsb r11, r14, #0 @ r11 = negative shift for scalar lsr
389 add store_a, psx_gpu, #psx_gpu_uvrg_offset
391 sub r11, r11, #(32 - 13)
393 add store_b, store_a, #16
396 vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1
397 vst1.u32 { uvrg_base }, [ store_a, : 128 ], store_inc
399 vst1.u32 { uvrg_dx1 }, [ store_b, : 128 ], store_inc
400 mov g_bx, gw_bx_h, lsr r11
402 vst1.u32 { g_uvrg_y }, [ store_a, : 128 ], store_inc
403 mov g_by, gw_by_h, lsr r11
405 vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \
406 [ store_b, : 128 ], store_inc
407 eor g_bx, g_bx, gs_bx
409 vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \
410 [ store_b, : 128 ], store_inc
411 sub g_bx, g_bx, gs_bx
414 eor g_by, g_by, gs_by
416 mls b_base, g_bx, x0, b_base
417 sub g_by, g_by, gs_by
422 add g_bx2, g_bx, g_bx
423 add g_bx3, g_bx, g_bx2
425 stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by }
427 ldmia sp!, { r4 - r11, pc }
444 #define height_minor_a r7
445 #define height_minor_b r8
446 #define height_major r9
449 #define reciprocal_table_ptr r10
451 #define edge_alt_low r4
452 #define edge_alt_high r5
453 #define edge_dx_dy_alt r6
454 #define edge_shift_alt r10
456 #define edge_dx_dy_alt_low r4
457 #define edge_dx_dy_alt_high r5
459 #define span_edge_data r4
460 #define span_uvrg_offset r5
461 #define span_b_offset r6
469 #define alternate_x q0
470 #define alternate_dx_dy q1
471 #define alternate_x_32 q2
473 #define alternate_x_low d0
474 #define alternate_x_high d1
475 #define alternate_dx_dy_low d2
476 #define alternate_dx_dy_high d3
477 #define alternate_x_32_low d4
478 #define alternate_x_32_high d5
482 #define left_dx_dy q5
483 #define right_dx_dy q6
485 #define right_edge q8
487 #define left_x_low d6
488 #define left_x_high d7
489 #define right_x_low d8
490 #define right_x_high d9
491 #define left_dx_dy_low d10
492 #define left_dx_dy_high d11
493 #define right_dx_dy_low d12
494 #define right_dx_dy_high d13
495 #define left_edge_low d14
496 #define left_edge_high d15
497 #define right_edge_low d16
498 #define right_edge_high d17
500 #define y_mid_point d18
503 #define left_right_x_16 q11
504 #define span_shifts_y q12
507 #define span_shifts d24
512 #define left_right_x_16_low d22
513 #define left_right_x_16_high d23
518 #define alternate_x_16 d4
521 #define v_clip_low d6
523 #define right_x_32 q10
524 #define left_x_32 q11
525 #define alternate_select d24
527 #define right_x_32_low d20
528 #define right_x_32_high d21
529 #define left_x_32_low d22
530 #define left_x_32_high d23
533 #define edges_dx_dy d2
534 #define edge_shifts d3
535 #define edge_shifts_64 q2
537 #define edges_xy_left d0
538 #define edges_xy_right d1
540 #define height_reciprocals d6
548 #define heights_b d12
549 #define edges_dx_dy_64 q10
551 #define edges_dx_dy_64_left d20
552 #define edges_dx_dy_64_right d21
555 #define setup_spans_prologue() \
556 stmdb sp!, { r4 - r11, lr }; \
558 ldrsh x_a, [ v_a, #8 ]; \
559 ldrsh x_b, [ v_b, #8 ]; \
560 ldrsh x_c, [ v_c, #8 ]; \
561 ldrsh y_a, [ v_a, #10 ]; \
562 ldrsh y_b, [ v_b, #10 ]; \
563 ldrsh y_c, [ v_c, #10 ]; \
565 add temp, psx_gpu, #psx_gpu_uvrg_offset; \
566 vld1.32 { uvrg }, [ temp ]; \
567 add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \
568 vld1.32 { uvrg_dy }, [ temp ]; \
569 ldr reciprocal_table_ptr, [ psx_gpu, #psx_gpu_reciprocal_table_ptr_offset ]; \
571 vmov.u32 c_0x01, #0x01 \
573 #define setup_spans_load_b() \
574 ldr b, [ psx_gpu, #psx_gpu_b_offset ]; \
575 ldr b_dy, [ psx_gpu, #psx_gpu_b_dy_offset ] \
577 #define setup_spans_prologue_b() \
578 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
579 add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \
581 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
582 vmov.u16 c_0x0004, #0x0004; \
584 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
585 vmov.u16 c_0x0001, #0x0001; \
587 vld1.u16 { left_edge_low[], left_edge_high[] }, [ temp ]; \
588 add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \
590 vld1.u16 { right_edge_low[], right_edge_high[] }, [ temp ]; \
591 vadd.u16 right_edge, right_edge, c_0x0001; \
593 vmov.u16 c_0x0007, #0x0007; \
594 vmvn.u16 c_0xFFFE, #0x0001 \
597 #define compute_edge_delta_x2() \
598 ldr temp, [ reciprocal_table_ptr, height, lsl #2 ]; \
600 vdup.u32 heights, height; \
601 vsub.u32 widths, x_ends, x_starts; \
603 vdup.u32 edge_shifts, temp; \
604 vsub.u32 heights_b, heights, c_0x01; \
605 vshr.u32 height_reciprocals, edge_shifts, #10; \
607 vmla.s32 heights_b, x_starts, heights; \
608 vbic.u16 edge_shifts, #0xE0; \
609 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
610 vmull.s32 edges_xy, heights_b, height_reciprocals \
613 #define height_reciprocal_alt r11
614 #define height_b_alt r12
616 #define compute_edge_delta_x3(start_c, height_a, height_b) \
617 vmov heights, height_a, height_b; \
618 ldr temp, [ reciprocal_table_ptr, height_a, lsl #2 ]; \
619 vmov.u32 edge_shifts[0], temp; \
620 ldr temp, [ reciprocal_table_ptr, height_b, lsl #2 ]; \
621 vmov.u32 edge_shifts[1], temp; \
622 ldr edge_shift_alt, [ reciprocal_table_ptr, height_minor_b, lsl #2 ]; \
624 vsub.u32 widths, x_ends, x_starts; \
625 sub width_alt, x_c, start_c; \
627 vsub.u32 heights_b, heights, c_0x01; \
628 sub height_b_alt, height_minor_b, #1; \
630 vshr.u32 height_reciprocals, edge_shifts, #10; \
631 lsr height_reciprocal_alt, edge_shift_alt, #10; \
633 vmla.s32 heights_b, x_starts, heights; \
634 mla height_b_alt, height_minor_b, start_c, height_b_alt; \
636 vbic.u16 edge_shifts, #0xE0; \
637 and edge_shift_alt, edge_shift_alt, #0x1F; \
639 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
640 mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \
642 vmull.s32 edges_xy, heights_b, height_reciprocals; \
643 smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \
646 #define setup_spans_adjust_y_up() \
647 vsub.u32 y_x4, y_x4, c_0x0004 \
649 #define setup_spans_adjust_y_down() \
650 vadd.u32 y_x4, y_x4, c_0x0004 \
652 #define setup_spans_adjust_interpolants_up() \
653 vsub.u32 uvrg, uvrg, uvrg_dy; \
656 #define setup_spans_adjust_interpolants_down() \
657 vadd.u32 uvrg, uvrg, uvrg_dy; \
661 #define setup_spans_clip_interpolants_increment() \
662 mla b, b_dy, clip, b; \
663 vmla.s32 uvrg, uvrg_dy, v_clip \
665 #define setup_spans_clip_interpolants_decrement() \
666 mls b, b_dy, clip, b; \
667 vmls.s32 uvrg, uvrg_dy, v_clip \
669 #define setup_spans_clip_alternate_yes() \
670 smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \
672 #define setup_spans_clip_alternate_no() \
674 #define setup_spans_clip(direction, alternate_active) \
675 vdup.u32 v_clip, clip; \
676 setup_spans_clip_alternate_##alternate_active(); \
677 setup_spans_clip_interpolants_##direction(); \
678 vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \
681 #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \
682 vmovl.s32 edge_shifts_64, edge_shifts; \
683 vmovl.s32 edges_dx_dy_64, edges_dx_dy; \
685 vshl.s64 edges_xy, edges_xy, edge_shifts_64; \
686 vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \
688 vmov left_x_low, edges_xy_##left_index; \
689 vmov right_x_low, edges_xy_##right_index; \
691 vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \
692 vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \
693 vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \
694 vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \
696 vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \
697 vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \
699 vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \
700 vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \
703 #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \
704 setup_spans_adjust_edges_alternate_no(left_index, right_index); \
706 vdup.u16 y_mid_point, y_b; \
707 rsb temp, edge_shift_alt, #32; \
709 lsl edge_alt_high, edge_alt_high, edge_shift_alt; \
710 orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \
711 lsl edge_alt_low, edge_alt_low, edge_shift_alt; \
712 vmov alternate_x_low, edge_alt_low, edge_alt_high; \
714 asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \
715 lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \
716 vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \
717 vmov alternate_dx_dy_high, alternate_dx_dy_low; \
719 vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \
720 vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \
723 #define setup_spans_y_select_up() \
724 vclt.s16 alternate_select, y_x4, y_mid_point \
726 #define setup_spans_y_select_down() \
727 vcgt.s16 alternate_select, y_x4, y_mid_point \
730 #define setup_spans_alternate_select_left() \
731 vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \
733 #define setup_spans_alternate_select_right() \
734 vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \
737 #define setup_spans_set_x4_alternate_yes(alternate, direction) \
738 vshrn.s64 alternate_x_32_low, alternate_x, #32; \
739 vshrn.s64 left_x_32_low, left_x, #32; \
740 vshrn.s64 right_x_32_low, right_x, #32; \
742 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
743 vadd.u64 left_x, left_x, left_dx_dy; \
744 vadd.u64 right_x, right_x, right_dx_dy; \
746 vshrn.s64 alternate_x_32_high, alternate_x, #32; \
747 vshrn.s64 left_x_32_high, left_x, #32; \
748 vshrn.s64 right_x_32_high, right_x, #32; \
750 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
751 vadd.u64 left_x, left_x, left_dx_dy; \
752 vadd.u64 right_x, right_x, right_dx_dy; \
754 vmovn.u32 alternate_x_16, alternate_x_32; \
755 setup_spans_y_select_##direction(); \
756 vmovn.u32 left_right_x_16_low, left_x_32; \
758 vmovn.u32 left_right_x_16_high, right_x_32; \
759 setup_spans_alternate_select_##alternate(); \
761 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
762 str b, [ span_b_offset ], #4; \
763 setup_spans_adjust_interpolants_##direction(); \
765 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
767 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
768 str b, [ span_b_offset ], #4; \
769 setup_spans_adjust_interpolants_##direction(); \
771 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
773 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
774 str b, [ span_b_offset ], #4; \
775 setup_spans_adjust_interpolants_##direction(); \
777 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
778 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
779 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
781 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
782 str b, [ span_b_offset ], #4; \
783 setup_spans_adjust_interpolants_##direction(); \
785 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
786 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
788 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
790 setup_spans_adjust_y_##direction() \
793 #define setup_spans_set_x4_alternate_no(alternate, direction) \
794 vshrn.s64 left_x_32_low, left_x, #32; \
795 vshrn.s64 right_x_32_low, right_x, #32; \
797 vadd.u64 left_x, left_x, left_dx_dy; \
798 vadd.u64 right_x, right_x, right_dx_dy; \
800 vshrn.s64 left_x_32_high, left_x, #32; \
801 vshrn.s64 right_x_32_high, right_x, #32; \
803 vadd.u64 left_x, left_x, left_dx_dy; \
804 vadd.u64 right_x, right_x, right_dx_dy; \
806 vmovn.u32 left_right_x_16_low, left_x_32; \
807 vmovn.u32 left_right_x_16_high, right_x_32; \
809 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
810 str b, [ span_b_offset ], #4; \
811 setup_spans_adjust_interpolants_##direction(); \
813 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
815 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
816 str b, [ span_b_offset ], #4; \
817 setup_spans_adjust_interpolants_##direction(); \
819 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
821 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
822 str b, [ span_b_offset ], #4; \
823 setup_spans_adjust_interpolants_##direction(); \
825 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
826 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
827 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
829 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
830 str b, [ span_b_offset ], #4; \
831 setup_spans_adjust_interpolants_##direction(); \
833 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
834 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
836 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
838 setup_spans_adjust_y_##direction() \
841 #define edge_adjust_low r11
842 #define edge_adjust_high r12
844 #define setup_spans_alternate_adjust_yes() \
845 smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \
846 subs edge_alt_low, edge_alt_low, edge_adjust_low; \
847 sbc edge_alt_high, edge_alt_high, edge_adjust_high \
849 #define setup_spans_alternate_adjust_no() \
852 #define setup_spans_down(left_index, right_index, alternate, alternate_active) \
853 setup_spans_alternate_adjust_##alternate_active(); \
854 setup_spans_load_b(); \
856 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
857 subs y_c, y_c, temp; \
858 subgt height, height, y_c; \
859 addgt height, height, #1; \
861 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
862 subs clip, temp, y_a; \
865 sub height, height, clip; \
866 add y_a, y_a, clip; \
867 setup_spans_clip(increment, alternate_active); \
873 orr temp, y_a, y_a, lsl #16; \
874 add temp, temp, #(1 << 16); \
876 add y_a, y_a, #(2 << 16); \
877 vmov y_x4, temp, y_a; \
879 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
881 setup_spans_prologue_b(); \
883 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
886 setup_spans_set_x4_alternate_##alternate_active(alternate, down); \
887 subs height, height, #4; \
893 #define setup_spans_alternate_pre_increment_yes() \
894 adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \
895 adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \
897 #define setup_spans_alternate_pre_increment_no() \
900 #define setup_spans_up_decrement_yes() \
901 suble height, height, #1 \
903 #define setup_spans_up_decrement_no() \
906 #define setup_spans_up(left_index, right_index, alternate, alternate_active) \
907 setup_spans_alternate_adjust_##alternate_active(); \
908 setup_spans_load_b(); \
911 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
912 subs temp, temp, y_c; \
913 subgt height, height, temp; \
914 setup_spans_up_decrement_##alternate_active(); \
916 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
917 subs clip, y_a, temp; \
920 sub height, height, clip; \
921 sub y_a, y_a, clip; \
922 setup_spans_clip(decrement, alternate_active); \
928 orr temp, y_a, y_a, lsl #16; \
929 sub temp, temp, #(1 << 16); \
931 sub y_a, y_a, #(2 << 16); \
932 vmov y_x4, temp, y_a; \
934 vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \
936 setup_spans_alternate_pre_increment_##alternate_active(); \
937 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
939 setup_spans_adjust_interpolants_up(); \
940 setup_spans_prologue_b(); \
942 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
945 setup_spans_set_x4_alternate_##alternate_active(alternate, up); \
946 subs height, height, #4; \
952 #define setup_spans_epilogue() \
953 ldmia sp!, { r4 - r11, pc } \
956 #define setup_spans_up_up(minor, major) \
957 setup_spans_prologue(); \
958 sub height_minor_a, y_a, y_b; \
959 sub height_minor_b, y_b, y_c; \
960 sub height, y_a, y_c; \
962 vdup.u32 x_starts, x_a; \
963 vmov x_ends, x_c, x_b; \
965 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
966 setup_spans_up(major, minor, minor, yes); \
967 setup_spans_epilogue() \
969 function(setup_spans_up_left)
970 setup_spans_up_up(left, right)
972 function(setup_spans_up_right)
973 setup_spans_up_up(right, left)
975 #define setup_spans_down_down(minor, major) \
976 setup_spans_prologue(); \
977 sub height_minor_a, y_b, y_a; \
978 sub height_minor_b, y_c, y_b; \
979 sub height, y_c, y_a; \
981 vdup.u32 x_starts, x_a; \
982 vmov x_ends, x_c, x_b; \
984 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
985 setup_spans_down(major, minor, minor, yes); \
986 setup_spans_epilogue() \
988 function(setup_spans_down_left)
989 setup_spans_down_down(left, right)
991 function(setup_spans_down_right)
992 setup_spans_down_down(right, left)
995 #define setup_spans_up_flat() \
996 sub height, y_a, y_c; \
998 compute_edge_delta_x2(); \
999 setup_spans_up(left, right, none, no); \
1000 setup_spans_epilogue() \
1002 function(setup_spans_up_a)
1003 setup_spans_prologue()
1005 vmov x_starts, x_a, x_b
1006 vdup.u32 x_ends, x_c
1008 setup_spans_up_flat()
1010 function(setup_spans_up_b)
1011 setup_spans_prologue()
1013 vdup.u32 x_starts, x_a
1014 vmov x_ends, x_b, x_c
1016 setup_spans_up_flat()
1018 #define setup_spans_down_flat() \
1019 sub height, y_c, y_a; \
1021 compute_edge_delta_x2(); \
1022 setup_spans_down(left, right, none, no); \
1023 setup_spans_epilogue() \
1025 function(setup_spans_down_a)
1026 setup_spans_prologue()
1028 vmov x_starts, x_a, x_b
1029 vdup.u32 x_ends, x_c
1031 setup_spans_down_flat()
1033 function(setup_spans_down_b)
1034 setup_spans_prologue()
1036 vdup.u32 x_starts, x_a
1037 vmov x_ends, x_b, x_c
1039 setup_spans_down_flat()
1044 #define edges_xy_b q11
1045 #define edges_dx_dy_b d26
1046 #define edge_shifts_b d27
1047 #define edges_dx_dy_and_shifts_b q13
1048 #define height_increment d20
1050 #define edges_dx_dy_and_shifts q1
1052 #define edges_xy_b_left d22
1053 #define edges_xy_b_right d23
1055 #define setup_spans_up_down_load_edge_set_b() \
1056 vmov edges_xy, edges_xy_b; \
1057 vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \
1060 function(setup_spans_up_down)
1061 setup_spans_prologue()
1063 // s32 middle_y = y_a;
1064 sub height_minor_a, y_a, y_b
1065 sub height_minor_b, y_c, y_a
1066 sub height_major, y_c, y_b
1068 vmov x_starts, x_a, x_c
1069 vdup.u32 x_ends, x_b
1071 compute_edge_delta_x3(x_a, height_minor_a, height_major)
1074 vmov height_increment, temp, height_minor_b
1075 vmlal.s32 edges_xy, edges_dx_dy, height_increment
1077 vmov edges_xy_b_left, edge_alt_low, edge_alt_high
1078 vmov edges_xy_b_right, edges_xy_right
1080 vmov edge_shifts_b, edge_shifts
1081 vmov.u32 edge_shifts_b[0], edge_shift_alt
1083 vneg.s32 edges_dx_dy_b, edges_dx_dy
1084 vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt
1088 setup_spans_load_b()
1091 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1092 subs temp, temp, y_b
1093 subgt height_minor_a, height_minor_a, temp
1095 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1096 subs clip, y_a, temp
1099 sub height_minor_a, height_minor_a, clip
1101 setup_spans_clip(decrement, no)
1104 cmp height_minor_a, #0
1107 orr temp, y_a, y_a, lsl #16
1108 sub temp, temp, #(1 << 16)
1110 sub y_a, y_a, #(2 << 16)
1111 vmov y_x4, temp, y_a
1113 vaddw.s32 edges_xy, edges_xy, edges_dx_dy
1115 strh height_minor_a, [ psx_gpu, #psx_gpu_num_spans_offset ]
1117 setup_spans_adjust_edges_alternate_no(left, right);
1118 setup_spans_adjust_interpolants_up()
1119 setup_spans_up_down_load_edge_set_b()
1121 setup_spans_prologue_b()
1125 setup_spans_set_x4_alternate_no(none, up)
1126 subs height_minor_a, height_minor_a, #4
1129 add span_edge_data, span_edge_data, height_minor_a, lsl #3
1130 add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4
1131 add span_b_offset, span_b_offset, height_minor_a, lsl #2
1134 add temp, psx_gpu, #psx_gpu_uvrg_offset
1135 vld1.32 { uvrg }, [ temp ]
1138 setup_spans_load_b()
1140 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1142 subgt height_minor_b, height_minor_b, y_c
1143 addgt height_minor_b, height_minor_b, #1
1145 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1146 subs clip, temp, y_a
1149 sub height_minor_b, height_minor_b, clip
1151 setup_spans_clip(increment, no)
1154 cmp height_minor_b, #0
1157 orr temp, y_a, y_a, lsl #16
1158 add temp, temp, #(1 << 16)
1160 add y_a, y_a, #(2 << 16)
1161 vmov y_x4, temp, y_a
1163 setup_spans_adjust_edges_alternate_no(left, right)
1165 ldrh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1166 add temp, temp, height_minor_b
1168 cmp temp, #MAX_SPANS
1171 strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1174 setup_spans_set_x4_alternate_no(none, down)
1175 subs height_minor_b, height_minor_b, #4
1179 setup_spans_epilogue()
1182 setup_spans_up_down_load_edge_set_b()
1183 setup_spans_prologue_b()
1187 // FIXME: overflow corner case
1188 sub temp, temp, height_minor_b
1189 bics height_minor_b, #3
1190 add temp, temp, height_minor_b
1191 strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
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, 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 strheq 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 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 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3396 pld [ fb_ptr_next, #64 ]
3398 add fb_ptr_cmp, fb_ptr_cmp, #14
3399 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3404 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3405 vceq.u16 zero_mask, pixels, #0
3407 vst1.u16 { fb_pixels }, [ fb_ptr ]
3408 vtst.u16 draw_mask, draw_mask, test_mask
3411 subs num_blocks, num_blocks, #1
3415 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3416 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3418 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3420 ldmia sp!, { r4, pc }
3423 vst1.u16 { fb_pixels }, [ fb_ptr ]
3424 vceq.u16 zero_mask, pixels, #0
3426 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3427 vtst.u16 draw_mask, draw_mask, test_mask
3432 function(shade_blocks_unshaded_untextured_indirect)
3437 function(shade_blocks_unshaded_untextured_direct)
3438 stmdb sp!, { r4, r14 }
3439 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3441 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3442 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3444 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3445 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3447 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3448 vld1.u16 { pixels }, [ color_ptr, :128 ]
3451 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3453 vorr.u16 pixels, pixels, msb_mask
3454 subs num_blocks, num_blocks, #1
3456 ldr fb_ptr_next, [ block_ptr_load ], #64
3458 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3462 vmov fb_pixels, fb_pixels_next
3463 mov fb_ptr, fb_ptr_next
3464 ldr fb_ptr_next, [ block_ptr_load ], #64
3466 vbif.u16 fb_pixels, pixels, draw_mask
3467 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3469 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3470 add fb_ptr_cmp, fb_ptr_cmp, #14
3474 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3475 vst1.u16 { fb_pixels }, [ fb_ptr ]
3478 subs num_blocks, num_blocks, #1
3482 vbif.u16 fb_pixels_next, pixels, draw_mask
3483 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3485 ldmia sp!, { r4, pc }
3488 vst1.u16 { fb_pixels }, [ fb_ptr ]
3489 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3493 #undef draw_mask_ptr
3500 #define num_blocks r1
3501 #define msb_mask_ptr r2
3502 #define pixel_ptr r3
3503 #define draw_mask_ptr r0
3506 #define fb_ptr_next r14
3507 #define fb_ptr_cmp r4
3515 #undef msb_mask_high
3516 #undef draw_mask_next
3519 #undef fb_pixels_next
3522 #define draw_mask q1
3524 #define fb_pixels q3
3525 #define blend_pixels q4
3526 #define pixels_no_msb q5
3527 #define blend_mask q6
3528 #define fb_pixels_no_msb q7
3529 #define d128_0x8000 q8
3530 #define d128_0x0421 q9
3531 #define fb_pixels_next q10
3532 #define blend_pixels_next q11
3533 #define pixels_next q12
3534 #define draw_mask_next q13
3535 #define write_mask q14
3537 #define pixels_rb q5
3538 #define pixels_mg q7
3540 #define d128_0x7C1F q8
3541 #define d128_0x03E0 q9
3542 #define fb_pixels_rb q10
3543 #define fb_pixels_g q11
3544 #define fb_pixels_masked q11
3545 #define d128_0x83E0 q15
3546 #define pixels_fourth q7
3547 #define d128_0x1C07 q12
3548 #define d128_0x00E0 q13
3549 #define d128_0x80E0 q13
3551 #define msb_mask_low d0
3552 #define msb_mask_high d1
3554 #define blend_blocks_average_set_blend_mask_textured(source) \
3555 vclt.s16 blend_mask, source, #0 \
3557 #define blend_blocks_average_set_stp_bit_textured() \
3558 vorr.u16 blend_pixels, #0x8000 \
3560 #define blend_blocks_average_combine_textured(source) \
3561 vbif.u16 blend_pixels, source, blend_mask \
3563 #define blend_blocks_average_set_blend_mask_untextured(source) \
3565 #define blend_blocks_average_set_stp_bit_untextured() \
3567 #define blend_blocks_average_combine_untextured(source) \
3569 #define blend_blocks_average_mask_set_on() \
3570 vclt.s16 write_mask, fb_pixels_next, #0 \
3572 #define blend_blocks_average_mask_copy_on() \
3573 vorr.u16 draw_mask, draw_mask_next, write_mask \
3575 #define blend_blocks_average_mask_copy_b_on() \
3576 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3578 #define blend_blocks_average_mask_set_off() \
3580 #define blend_blocks_average_mask_copy_off() \
3581 vmov draw_mask, draw_mask_next \
3583 #define blend_blocks_average_mask_copy_b_off() \
3585 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3588 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3589 stmdb sp!, { r4, r14 }; \
3590 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3591 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3593 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3594 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3596 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3599 vmov.u16 d128_0x8000, #0x8000; \
3600 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3601 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3603 vmov.u16 d128_0x0421, #0x0400; \
3604 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3606 vorr.u16 d128_0x0421, #0x0021; \
3607 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3609 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3610 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3611 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3612 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3613 blend_blocks_average_mask_set_##mask_evaluate(); \
3614 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3616 subs num_blocks, num_blocks, #1; \
3620 mov fb_ptr, fb_ptr_next; \
3621 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3623 vmov pixels, pixels_next; \
3624 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3626 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3628 blend_blocks_average_mask_copy_##mask_evaluate(); \
3629 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3631 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3632 blend_blocks_average_set_stp_bit_##texturing(); \
3633 vmov fb_pixels, fb_pixels_next; \
3634 blend_blocks_average_combine_##texturing(pixels); \
3636 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3637 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3638 cmp fb_ptr_cmp, #28; \
3641 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3642 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3644 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3645 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3647 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3648 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3650 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3651 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3652 blend_blocks_average_mask_set_##mask_evaluate(); \
3653 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3656 subs num_blocks, num_blocks, #1; \
3660 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3661 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3663 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3664 blend_blocks_average_set_stp_bit_##texturing(); \
3665 blend_blocks_average_combine_##texturing(pixels_next); \
3667 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3668 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3669 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3671 ldmia sp!, { r4, pc }; \
3674 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3675 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3676 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3678 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3679 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3680 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3681 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3682 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3683 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3687 blend_blocks_average_builder(textured, off)
3688 blend_blocks_average_builder(untextured, off)
3689 blend_blocks_average_builder(textured, on)
3690 blend_blocks_average_builder(untextured, on)
3693 #define blend_blocks_add_mask_set_on() \
3694 vclt.s16 write_mask, fb_pixels, #0 \
3696 #define blend_blocks_add_mask_copy_on() \
3697 vorr.u16 draw_mask, draw_mask, write_mask \
3699 #define blend_blocks_add_mask_set_off() \
3701 #define blend_blocks_add_mask_copy_off() \
3704 #define blend_blocks_add_textured_builder(mask_evaluate) \
3707 function(blend_blocks_textured_add_##mask_evaluate) \
3708 stmdb sp!, { r4, r14 }; \
3709 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3710 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3712 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3713 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3715 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3718 vmov.u16 d128_0x7C1F, #0x7C00; \
3719 vmov.u16 d128_0x03E0, #0x0300; \
3720 vmov.u16 d128_0x83E0, #0x8000; \
3721 vorr.u16 d128_0x03E0, #0x00E0; \
3722 vorr.u16 d128_0x7C1F, #0x001F; \
3723 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3725 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3726 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3727 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3728 vclt.s16 blend_mask, pixels, #0; \
3729 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3730 blend_blocks_add_mask_set_##mask_evaluate(); \
3731 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3733 blend_blocks_add_mask_copy_##mask_evaluate(); \
3734 vorr.u16 pixels, pixels, msb_mask; \
3735 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3736 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3737 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3738 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3739 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3740 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3741 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3742 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3744 subs num_blocks, num_blocks, #1; \
3748 mov fb_ptr, fb_ptr_next; \
3750 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3752 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3753 vclt.s16 blend_mask, pixels, #0; \
3755 vorr.u16 pixels, pixels, msb_mask; \
3756 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3757 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3759 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3760 pld [ fb_ptr_next, #64 ]; \
3762 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3763 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3765 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3766 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3768 cmp fb_ptr_cmp, #28; \
3771 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3772 blend_blocks_add_mask_set_##mask_evaluate(); \
3773 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3774 blend_blocks_add_mask_copy_##mask_evaluate(); \
3775 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3776 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3777 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3780 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3781 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3782 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3783 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3784 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3786 subs num_blocks, num_blocks, #1; \
3790 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3791 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3792 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3794 ldmia sp!, { r4, pc }; \
3797 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3798 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3800 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3801 blend_blocks_add_mask_set_##mask_evaluate(); \
3802 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3803 blend_blocks_add_mask_copy_##mask_evaluate(); \
3804 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3808 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3811 function(blend_blocks_untextured_add_##mask_evaluate) \
3812 stmdb sp!, { r4, r14 }; \
3813 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3814 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3816 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3817 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3819 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3822 vmov.u16 d128_0x7C1F, #0x7C00; \
3823 vmov.u16 d128_0x03E0, #0x0300; \
3824 vorr.u16 d128_0x7C1F, #0x001F; \
3825 vorr.u16 d128_0x03E0, #0x00E0; \
3827 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3828 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3829 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3830 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3831 blend_blocks_add_mask_set_##mask_evaluate(); \
3832 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3834 blend_blocks_add_mask_copy_##mask_evaluate(); \
3835 vand.u16 pixels_g, pixels, d128_0x03E0; \
3836 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3837 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3838 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3839 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3840 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3841 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3843 subs num_blocks, num_blocks, #1; \
3847 mov fb_ptr, fb_ptr_next; \
3849 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3851 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3853 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3854 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3855 vand.u16 pixels_g, pixels, d128_0x03E0; \
3857 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3858 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3860 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3861 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3862 cmp fb_ptr_cmp, #28; \
3865 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3866 blend_blocks_add_mask_set_##mask_evaluate(); \
3867 blend_blocks_add_mask_copy_##mask_evaluate(); \
3868 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3869 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3870 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3873 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3874 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3875 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3876 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3877 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3879 subs num_blocks, num_blocks, #1; \
3883 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3884 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3885 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3886 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3888 ldmia sp!, { r4, pc }; \
3891 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3892 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3894 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3895 blend_blocks_add_mask_set_##mask_evaluate(); \
3896 blend_blocks_add_mask_copy_##mask_evaluate(); \
3897 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3901 blend_blocks_add_textured_builder(off)
3902 blend_blocks_add_textured_builder(on)
3903 blend_blocks_add_untextured_builder(off)
3904 blend_blocks_add_untextured_builder(on)
3906 #define blend_blocks_subtract_set_blend_mask_textured() \
3907 vclt.s16 blend_mask, pixels_next, #0 \
3909 #define blend_blocks_subtract_combine_textured() \
3910 vbif.u16 blend_pixels, pixels, blend_mask \
3912 #define blend_blocks_subtract_set_stb_textured() \
3913 vorr.u16 blend_pixels, #0x8000 \
3915 #define blend_blocks_subtract_msb_mask_textured() \
3916 vorr.u16 pixels, pixels_next, msb_mask \
3918 #define blend_blocks_subtract_set_blend_mask_untextured() \
3920 #define blend_blocks_subtract_combine_untextured() \
3922 #define blend_blocks_subtract_set_stb_untextured() \
3923 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3925 #define blend_blocks_subtract_msb_mask_untextured() \
3928 #define blend_blocks_subtract_mask_set_on() \
3929 vclt.s16 write_mask, fb_pixels, #0 \
3931 #define blend_blocks_subtract_mask_copy_on() \
3932 vorr.u16 draw_mask, draw_mask_next, write_mask \
3934 #define blend_blocks_subtract_mask_set_off() \
3936 #define blend_blocks_subtract_mask_copy_off() \
3937 vmov draw_mask, draw_mask_next \
3940 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
3943 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
3944 stmdb sp!, { r4, r14 }; \
3945 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3946 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3948 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3949 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3951 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3954 vmov.u16 d128_0x7C1F, #0x7C00; \
3955 vmov.u16 d128_0x03E0, #0x0300; \
3956 vorr.u16 d128_0x7C1F, #0x001F; \
3957 vorr.u16 d128_0x03E0, #0x00E0; \
3959 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3960 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3961 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3962 blend_blocks_subtract_set_blend_mask_##texturing(); \
3963 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3964 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3965 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3967 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3968 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3969 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3970 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3971 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3973 subs num_blocks, num_blocks, #1; \
3977 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
3978 mov fb_ptr, fb_ptr_next; \
3979 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3981 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3982 blend_blocks_subtract_msb_mask_##texturing(); \
3984 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3985 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3986 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3987 blend_blocks_subtract_set_stb_##texturing(); \
3988 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3989 blend_blocks_subtract_combine_##texturing(); \
3990 blend_blocks_subtract_set_blend_mask_##texturing(); \
3991 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3993 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3994 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3995 cmp fb_ptr_cmp, #28; \
3998 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3999 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4000 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4001 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4002 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4003 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4004 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4007 subs num_blocks, num_blocks, #1; \
4011 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4013 blend_blocks_subtract_msb_mask_##texturing(); \
4014 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4015 blend_blocks_subtract_set_stb_##texturing(); \
4016 blend_blocks_subtract_combine_##texturing(); \
4017 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4018 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4020 ldmia sp!, { r4, pc }; \
4023 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4024 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4025 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4026 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4027 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4028 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4029 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4033 blend_blocks_subtract_builder(textured, off)
4034 blend_blocks_subtract_builder(textured, on)
4035 blend_blocks_subtract_builder(untextured, off)
4036 blend_blocks_subtract_builder(untextured, on)
4039 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4042 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4043 stmdb sp!, { r4, r14 }; \
4044 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4045 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4047 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4048 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4050 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4053 vmov.u16 d128_0x7C1F, #0x7C00; \
4054 vmov.u16 d128_0x03E0, #0x0300; \
4055 vmov.u16 d128_0x1C07, #0x1C00; \
4056 vmov.u16 d128_0x00E0, #0x00E0; \
4057 vorr.u16 d128_0x7C1F, #0x001F; \
4058 vorr.u16 d128_0x03E0, #0x00E0; \
4059 vorr.u16 d128_0x1C07, #0x0007; \
4061 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4062 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4063 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4064 vclt.s16 blend_mask, pixels, #0; \
4065 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4066 blend_blocks_add_mask_set_##mask_evaluate(); \
4067 vshr.s16 pixels_fourth, pixels, #2; \
4068 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4070 blend_blocks_add_mask_copy_##mask_evaluate(); \
4071 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4072 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4073 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4074 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4075 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4076 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4077 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4079 subs num_blocks, num_blocks, #1; \
4083 mov fb_ptr, fb_ptr_next; \
4084 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4086 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4087 vbif.u16 blend_pixels, pixels, blend_mask; \
4089 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4090 vclt.s16 blend_mask, pixels, #0; \
4091 vshr.s16 pixels_fourth, pixels, #2; \
4092 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4093 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4095 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4096 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4098 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4099 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4100 cmp fb_ptr_cmp, #28; \
4103 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4104 blend_blocks_add_mask_set_##mask_evaluate(); \
4105 blend_blocks_add_mask_copy_##mask_evaluate(); \
4106 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4107 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4108 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4111 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4112 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4113 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4114 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4115 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4117 subs num_blocks, num_blocks, #1; \
4121 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4122 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4123 vbif.u16 blend_pixels, pixels, blend_mask; \
4124 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4125 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4127 ldmia sp!, { r4, pc }; \
4130 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4131 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4133 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4134 blend_blocks_add_mask_set_##mask_evaluate(); \
4135 blend_blocks_add_mask_copy_##mask_evaluate(); \
4136 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4141 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4144 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4145 stmdb sp!, { r4, r14 }; \
4146 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4147 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4149 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4150 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4152 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4155 vmov.u16 d128_0x7C1F, #0x7C00; \
4156 vmov.u16 d128_0x03E0, #0x0300; \
4157 vmov.u16 d128_0x1C07, #0x1C00; \
4158 vmov.u16 d128_0x00E0, #0x00E0; \
4159 vorr.u16 d128_0x7C1F, #0x001F; \
4160 vorr.u16 d128_0x03E0, #0x00E0; \
4161 vorr.u16 d128_0x1C07, #0x0007; \
4163 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4164 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4165 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4166 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4167 blend_blocks_add_mask_set_##mask_evaluate(); \
4168 vshr.s16 pixels_fourth, pixels, #2; \
4169 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4171 blend_blocks_add_mask_copy_##mask_evaluate(); \
4172 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4173 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4174 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4175 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4176 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4177 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4178 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4180 subs num_blocks, num_blocks, #1; \
4184 mov fb_ptr, fb_ptr_next; \
4185 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4187 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4189 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4190 vshr.s16 pixels_fourth, pixels, #2; \
4191 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4192 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4194 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4195 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4197 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4198 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4199 cmp fb_ptr_cmp, #28; \
4202 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4203 blend_blocks_add_mask_set_##mask_evaluate(); \
4204 blend_blocks_add_mask_copy_##mask_evaluate(); \
4205 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4206 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4207 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4210 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4211 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4212 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4213 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4214 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4216 subs num_blocks, num_blocks, #1; \
4220 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4221 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4222 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4223 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4225 ldmia sp!, { r4, pc }; \
4228 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4229 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4231 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4232 blend_blocks_add_mask_set_##mask_evaluate(); \
4233 blend_blocks_add_mask_copy_##mask_evaluate(); \
4234 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4238 blend_blocks_add_fourth_textured_builder(off)
4239 blend_blocks_add_fourth_textured_builder(on)
4240 blend_blocks_add_fourth_untextured_builder(off)
4241 blend_blocks_add_fourth_untextured_builder(on)
4243 // TODO: Optimize this more. Need a scene that actually uses it for
4248 function(blend_blocks_textured_unblended_on)
4249 stmdb sp!, { r4, r14 }
4250 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4251 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
4253 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4254 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
4256 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4259 ldr fb_ptr, [ pixel_ptr, #28 ]
4260 vld1.u16 { fb_pixels }, [ fb_ptr ]
4261 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4262 vclt.s16 write_mask, fb_pixels, #0
4263 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4265 subs num_blocks, num_blocks, #1
4269 vorr.u16 pixels, pixels, msb_mask
4270 vorr.u16 draw_mask, draw_mask, write_mask
4271 vbif.u16 fb_pixels, pixels, draw_mask
4272 vst1.u16 { fb_pixels }, [ fb_ptr ]
4274 ldr fb_ptr, [ pixel_ptr, #28 ]
4275 vld1.u16 { fb_pixels }, [ fb_ptr ]
4276 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4277 vclt.s16 write_mask, fb_pixels, #0
4278 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4280 subs num_blocks, num_blocks, #1
4284 vorr.u16 pixels, pixels, msb_mask
4285 vorr.u16 draw_mask, draw_mask, write_mask
4286 vbif.u16 fb_pixels, pixels, draw_mask
4287 vst1.u16 { fb_pixels }, [ fb_ptr ]
4289 ldmia sp!, { r4, pc }
4292 function(blend_blocks_textured_unblended_off)
4302 vld1.u32 { u_whole_8, v_whole_8 }, [ r1, :128 ], r3
4322 #define num_width r12
4332 function(render_block_fill_body)
4333 vdup.u16 colors_a, color
4336 vmov colors_b, colors_a
4337 sub pitch, pitch, width, lsl #1
4339 mov num_width, width
4342 vst1.u32 { colors_a, colors_b }, [ vram_ptr, :256 ]!
4344 subs num_width, num_width, #16
4347 add vram_ptr, vram_ptr, pitch
4348 mov num_width, width
4350 subs height, height, #1
4364 #undef dirty_textures_mask
4366 #undef current_texture_mask
4377 #define offset_u_right r10
4378 #define width_rounded r11
4379 #define height_rounded r12
4381 #define texture_offset_base r1
4382 #define tile_width r2
4383 #define tile_height r3
4384 #define num_blocks r4
4386 #define sub_tile_height r6
4388 #define texture_mask r8
4389 #define column_data r9
4390 #define texture_offset r10
4391 #define tiles_remaining r11
4392 #define fb_ptr_advance_column r12
4393 #define texture_block_ptr r14
4395 #define texture_page_ptr r3
4396 #define left_block_mask r4
4397 #define right_block_mask r5
4398 #define texture_mask_rev r10
4399 #define control_mask r11
4401 #define dirty_textures_mask r4
4403 #define current_texture_mask r6
4417 #define draw_masks_fb_ptrs q1
4419 #define draw_mask_fb_ptr_left d2
4420 #define draw_mask_fb_ptr_right d3
4422 #define draw_mask_fb_ptr_left_a d2
4423 #define draw_mask_fb_ptr_left_b d3
4424 #define draw_mask_fb_ptr_right_a d10
4425 #define draw_mask_fb_ptr_right_b d11
4426 #define draw_masks_fb_ptrs2 q5
4428 #define clut_low_a d4
4429 #define clut_low_b d5
4430 #define clut_high_a d6
4431 #define clut_high_b d7
4433 #define block_masks d8
4434 #define block_masks_shifted d9
4439 #define texels_low d12
4440 #define texels_high d13
4442 #define texels_wide_low d14
4443 #define texels_wide_high d15
4444 #define texels_wide q7
4447 setup_sprite_flush_blocks:
4450 stmdb sp!, { r0 - r3, r12, r14 }
4451 bl flush_render_block_buffer
4452 ldmia sp!, { r0 - r3, r12, r14 }
4456 add block, psx_gpu, #psx_gpu_blocks_offset
4460 setup_sprite_update_texture_4bpp_cache:
4461 stmdb sp!, { r0 - r3, r14 }
4462 bl update_texture_4bpp_cache
4463 ldmia sp!, { r0 - r3, pc }
4466 setup_sprite_update_texture_8bpp_cache:
4467 stmdb sp!, { r0 - r3, r14 }
4468 bl update_texture_8bpp_cache
4469 ldmia sp!, { r0 - r3, pc }
4472 #define setup_sprite_tiled_initialize_4bpp() \
4473 ldr dirty_textures_mask, \
4474 [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]; \
4475 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4477 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4478 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4480 tst current_texture_mask, dirty_textures_mask; \
4481 vuzp.u8 clut_a, clut_b; \
4483 blne setup_sprite_update_texture_4bpp_cache \
4485 #define setup_sprite_tiled_initialize_8bpp() \
4486 ldr dirty_textures_mask, \
4487 [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]; \
4488 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4490 tst current_texture_mask, dirty_textures_mask; \
4491 blne setup_sprite_update_texture_8bpp_cache \
4494 #define setup_sprite_block_count_single() \
4497 #define setup_sprite_block_count_double() \
4498 sub_tile_height, lsl #1 \
4500 #define setup_sprite_tile_add_blocks(type) \
4501 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4502 cmp num_blocks, #MAX_BLOCKS; \
4504 movgt num_blocks, setup_sprite_block_count_##type(); \
4505 blgt setup_sprite_flush_blocks \
4508 #define setup_sprite_tile_full_4bpp(edge) \
4509 setup_sprite_tile_add_blocks(double); \
4512 and texture_block_ptr, texture_offset, texture_mask; \
4513 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4516 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4517 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4519 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4520 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4522 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4523 add texture_block_ptr, texture_offset, #8; \
4525 and texture_block_ptr, texture_block_ptr, texture_mask; \
4526 add block, block, #40; \
4528 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4529 add fb_ptr, fb_ptr, #16; \
4531 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4532 add block, block, #24; \
4534 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4535 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4538 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4539 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4541 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4542 add block, block, #40; \
4544 add texture_offset, texture_offset, #0x10; \
4545 add fb_ptr, fb_ptr, #(2048 - 16); \
4547 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4548 add block, block, #24; \
4550 subs sub_tile_height, sub_tile_height, #1; \
4553 add texture_offset, texture_offset, #0xF00; \
4554 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4557 #define setup_sprite_tile_half_4bpp(edge) \
4558 setup_sprite_tile_add_blocks(single); \
4561 and texture_block_ptr, texture_offset, texture_mask; \
4562 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4565 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4566 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4568 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4569 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4571 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4572 add block, block, #40; \
4574 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4575 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4577 add block, block, #24; \
4578 add texture_offset, texture_offset, #0x10; \
4580 add fb_ptr, fb_ptr, #2048; \
4581 subs sub_tile_height, sub_tile_height, #1; \
4585 add texture_offset, texture_offset, #0xF00; \
4586 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4589 #define setup_sprite_tile_full_8bpp(edge) \
4590 setup_sprite_tile_add_blocks(double); \
4591 add block, block, #16; \
4594 and texture_block_ptr, texture_offset, texture_mask; \
4595 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4598 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4599 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4601 add texture_block_ptr, texture_offset, #8; \
4602 vst1.u32 { texels }, [ block, :64 ]; \
4604 and texture_block_ptr, texture_block_ptr, texture_mask; \
4605 add block, block, #24; \
4607 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4609 add fb_ptr, fb_ptr, #16; \
4610 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4612 add block, block, #40; \
4613 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4616 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4617 vst1.u32 { texels }, [ block, :64 ]; \
4618 add block, block, #24; \
4620 add texture_offset, texture_offset, #0x10; \
4621 add fb_ptr, fb_ptr, #(2048 - 16); \
4623 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4624 add block, block, #40; \
4626 subs sub_tile_height, sub_tile_height, #1; \
4629 sub block, block, #16; \
4630 add texture_offset, texture_offset, #0xF00; \
4631 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4634 #define setup_sprite_tile_half_8bpp(edge) \
4635 setup_sprite_tile_add_blocks(single); \
4636 add block, block, #16; \
4639 and texture_block_ptr, texture_offset, texture_mask; \
4640 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4643 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4644 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4646 vst1.u32 { texels }, [ block, :64 ]; \
4647 add block, block, #24; \
4649 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4650 add block, block, #40; \
4652 add texture_offset, texture_offset, #0x10; \
4653 add fb_ptr, fb_ptr, #2048; \
4655 subs sub_tile_height, sub_tile_height, #1; \
4658 sub block, block, #16; \
4659 add texture_offset, texture_offset, #0xF00; \
4660 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4663 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4664 add texture_offset, texture_offset_base, #8; \
4665 add fb_ptr, fb_ptr, #16 \
4667 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4668 mov texture_offset, texture_offset_base \
4670 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4671 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4673 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4674 mov texture_offset, texture_offset_base \
4676 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4677 sub fb_ptr, fb_ptr, #16 \
4679 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4681 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4682 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4684 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4687 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \
4689 mov sub_tile_height, column_data; \
4690 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4691 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4692 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4694 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \
4696 and sub_tile_height, column_data, #0xFF; \
4697 mov tiles_remaining, column_data, lsr #16; \
4698 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4699 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4701 subs tiles_remaining, tiles_remaining, #1; \
4705 mov sub_tile_height, #16; \
4706 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4707 subs tiles_remaining, tiles_remaining, #1; \
4711 uxtb sub_tile_height, column_data, ror #8; \
4712 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4713 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4716 #define setup_sprite_column_data_single() \
4717 mov column_data, height; \
4718 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] \
4720 #define setup_sprite_column_data_multi() \
4721 and height_rounded, height_rounded, #0xF; \
4722 rsb column_data, offset_v, #16; \
4724 add height_rounded, height_rounded, #1; \
4725 sub tile_height, tile_height, #1; \
4727 orr column_data, column_data, tile_height, lsl #16; \
4728 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]; \
4730 orr column_data, column_data, height_rounded, lsl #8 \
4732 #define setup_sprite_setup_left_draw_mask_fb_ptr() \
4733 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4734 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4736 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column() \
4737 mov fb_ptr_advance_column, #32; \
4738 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4740 sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11; \
4741 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4743 #define setup_sprite_setup_right_draw_mask_fb_ptr() \
4744 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4745 vdup.u8 draw_mask_fb_ptr_right, block_masks[5] \
4747 #define setup_sprite_tile_column_width_single(tm, multi_height, edge_mode, \
4749 setup_sprite_##tm##_single_##multi_height##_##edge_mode##_##edge##x4mode: \
4750 setup_sprite_column_data_##multi_height(); \
4751 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4752 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4753 setup_sprite_setup_left_draw_mask_fb_ptr##x4mode(); \
4755 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, tm, x4mode); \
4756 ldmia sp!, { r4 - r11, pc } \
4758 #define setup_sprite_tiled_advance_column() \
4759 add texture_offset_base, texture_offset_base, #0x100; \
4760 tst texture_offset_base, #0xF00; \
4761 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4763 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4764 right_mode, x4mode) \
4765 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode##x4mode:\
4766 setup_sprite_column_data_##multi_height(); \
4768 setup_sprite_setup_left_draw_mask_fb_ptr_advance_column##x4mode(); \
4770 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm, x4mode);\
4772 subs tile_width, tile_width, #2; \
4773 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4777 vmov.u8 draw_masks_fb_ptrs, #0; \
4778 vmov.u8 draw_masks_fb_ptrs2, #0; \
4781 setup_sprite_tiled_advance_column(); \
4782 setup_sprite_tile_column_height_##multi_height(full, none, tm, x4mode); \
4783 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4784 subs tile_width, tile_width, #1; \
4788 setup_sprite_setup_right_draw_mask_fb_ptr##x4mode(); \
4790 setup_sprite_tiled_advance_column(); \
4791 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm, x4mode);\
4792 ldmia sp!, { r4 - r11, pc } \
4795 #define setup_sprite_offset_u_adjust() \
4797 #define setup_sprite_get_left_block_mask() \
4798 and left_block_mask, left_block_mask, #0xFF \
4800 #define setup_sprite_compare_left_block_mask() \
4801 cmp left_block_mask, #0xFF \
4803 #define setup_sprite_get_right_block_mask() \
4804 uxtb right_block_mask, right_block_mask, ror #8 \
4806 #define setup_sprite_compare_right_block_mask() \
4807 cmp right_block_mask, #0xFF \
4812 #define fb_ptr2 column_data
4814 #define setup_sprite_offset_u_adjust_4x() \
4815 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4816 lsl offset_u_right, #1; \
4818 add offset_u_right, #1 \
4820 #define setup_sprite_get_left_block_mask_4x() \
4821 sxth left_block_mask, left_block_mask \
4823 #define setup_sprite_compare_left_block_mask_4x() \
4824 cmp left_block_mask, #0xFFFFFFFF \
4826 #define setup_sprite_get_right_block_mask_4x() \
4827 sxth right_block_mask, right_block_mask, ror #16 \
4829 #define setup_sprite_compare_right_block_mask_4x() \
4830 cmp right_block_mask, #0xFFFFFFFF \
4833 #define widen_texels_16bpp(texels_) \
4834 vmov texels_wide_low, texels_; \
4835 vmov texels_wide_high, texels_; \
4836 vzip.16 texels_wide_low, texels_wide_high \
4838 #define widen_texels_8bpp(texels_) \
4839 vmov texels_wide_low, texels_; \
4840 vmov texels_wide_high, texels_; \
4841 vzip.8 texels_wide_low, texels_wide_high \
4843 #define write_block_16bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4844 vst1.u32 { texels_ }, [ block_, :128 ]; \
4845 add block_, block_, #40; \
4847 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4848 vst1.u32 { draw_mask_fb_ptr_ }, [ block_, :64 ]; \
4849 add block_, block_, #24 \
4851 /* assumes 16-byte offset already added to block_ */
4852 #define write_block_8bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4853 vst1.u32 { texels_ }, [ block_, :64 ]; \
4854 add block_, block_, #24; \
4856 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4857 vst1.u32 { draw_mask_fb_ptr_ }, [ block_, :64 ]; \
4858 add block_, block_, #40 \
4860 #define do_texture_block_16bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4861 draw_mask_fb_ptr_b_) \
4862 widen_texels_16bpp(texels_low); \
4863 add fb_ptr_tmp, fb_ptr, #1024*2; \
4865 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr); \
4867 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4868 widen_texels_16bpp(texels_high); \
4870 add fb_ptr_tmp, fb_ptr, #8*2; \
4871 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4873 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4874 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4876 #define do_texture_block_8bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4877 draw_mask_fb_ptr_b_) \
4878 widen_texels_8bpp(texels); \
4879 add fb_ptr_tmp, fb_ptr, #1024*2; \
4881 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr); \
4882 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4884 add fb_ptr_tmp, fb_ptr, #8*2; \
4885 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4887 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4888 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4891 #define setup_sprite_tiled_initialize_4bpp_4x() \
4892 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4893 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4895 vuzp.u8 clut_a, clut_b \
4897 #define setup_sprite_tiled_initialize_8bpp_4x() \
4900 #define setup_sprite_block_count_single_4x() \
4901 sub_tile_height, lsl #2 \
4903 #define setup_sprite_block_count_double_4x() \
4904 sub_tile_height, lsl #(1+2) \
4906 #define setup_sprite_tile_full_4bpp_4x(edge) \
4907 setup_sprite_tile_add_blocks(double_4x); \
4908 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4911 and texture_block_ptr, texture_offset, texture_mask; \
4914 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4915 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4917 add texture_block_ptr, texture_offset, #8; \
4918 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4920 and texture_block_ptr, texture_block_ptr, texture_mask; \
4921 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4923 vzip.8 texels_low, texels_high; \
4924 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
4925 draw_mask_fb_ptr_left_b); \
4927 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4928 pld [ fb_ptr, #2048 ]; \
4930 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4931 add fb_ptr, fb_ptr, #16*2; \
4933 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4934 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4936 vzip.8 texels_low, texels_high; \
4937 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
4938 draw_mask_fb_ptr_right_b); \
4940 add texture_offset, texture_offset, #0x10; \
4941 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
4943 subs sub_tile_height, sub_tile_height, #1; \
4946 ldr column_data, [sp], #8; /* fb_ptr2 */ \
4947 add texture_offset, texture_offset, #0xF00; \
4948 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4951 #define setup_sprite_tile_half_4bpp_4x(edge) \
4952 setup_sprite_tile_add_blocks(single_4x); \
4953 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4956 and texture_block_ptr, texture_offset, texture_mask; \
4959 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4960 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4962 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4963 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4965 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4966 add texture_offset, texture_offset, #0x10; \
4968 vzip.8 texels_low, texels_high; \
4969 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
4970 draw_mask_fb_ptr_##edge##_b); \
4972 pld [ fb_ptr, #2048 ]; \
4973 add fb_ptr, fb_ptr, #2048 * 2; \
4975 subs sub_tile_height, sub_tile_height, #1; \
4978 ldr column_data, [sp], #8; /* fb_ptr2 */ \
4979 add texture_offset, texture_offset, #0xF00; \
4980 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4983 #define setup_sprite_tile_full_8bpp_4x(edge) \
4984 setup_sprite_tile_add_blocks(double_4x); \
4985 add block, block, #16; \
4986 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4989 and texture_block_ptr, texture_offset, texture_mask; \
4992 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4993 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4995 add texture_block_ptr, texture_offset, #8; \
4996 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
4997 draw_mask_fb_ptr_left_b); \
4999 pld [ fb_ptr, #2048 ]; \
5000 and texture_block_ptr, texture_block_ptr, texture_mask; \
5002 add fb_ptr, fb_ptr, #16*2; \
5003 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5005 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
5007 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
5008 draw_mask_fb_ptr_right_b); \
5010 add texture_offset, texture_offset, #0x10; \
5011 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
5013 subs sub_tile_height, sub_tile_height, #1; \
5016 sub block, block, #16; \
5017 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5018 add texture_offset, texture_offset, #0xF00; \
5019 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
5022 #define setup_sprite_tile_half_8bpp_4x(edge) \
5023 setup_sprite_tile_add_blocks(single_4x); \
5024 add block, block, #16; \
5025 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5028 and texture_block_ptr, texture_offset, texture_mask; \
5031 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5032 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
5034 pld [ fb_ptr, #2048 ]; \
5035 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5036 draw_mask_fb_ptr_##edge##_b); \
5038 add texture_offset, texture_offset, #0x10; \
5039 add fb_ptr, fb_ptr, #2048 * 2; \
5041 subs sub_tile_height, sub_tile_height, #1; \
5044 sub block, block, #16; \
5045 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5046 add texture_offset, texture_offset, #0xF00; \
5047 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
5050 #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \
5051 add texture_offset, texture_offset_base, #8; \
5052 add fb_ptr, fb_ptr, #16 * 2 \
5054 #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \
5055 mov texture_offset, texture_offset_base \
5057 #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \
5058 setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \
5060 #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \
5061 mov texture_offset, texture_offset_base \
5063 #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \
5064 sub fb_ptr, fb_ptr, #16 * 2 \
5066 #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \
5068 #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \
5069 setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \
5071 #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \
5074 #define setup_sprite_setup_left_draw_mask_fb_ptr_4x() \
5075 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5076 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5077 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5078 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5080 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column_4x() \
5081 mov fb_ptr_advance_column, #32 * 2; \
5082 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5083 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5084 sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11 + 1; \
5085 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5086 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5088 #define setup_sprite_setup_right_draw_mask_fb_ptr_4x() \
5089 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[4]; \
5090 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[5]; \
5091 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[6]; \
5092 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[7] \
5100 // [ sp + 4 ]: width
5101 // [ sp + 8 ]: height
5102 // [ sp + 12 ]: color (unused)
5104 #define setup_sprite_tiled_builder(texture_mode, x4mode) \
5106 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \
5108 setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \
5110 setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \
5112 setup_sprite_tile_column_width_single(texture_mode, single, full, none, \
5114 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \
5116 setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \
5118 setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \
5120 setup_sprite_tile_column_width_single(texture_mode, single, half, right, \
5122 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \
5124 setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \
5126 setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \
5128 setup_sprite_tile_column_width_single(texture_mode, single, half, left, \
5130 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \
5132 setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \
5137 function(setup_sprite_##texture_mode##x4mode) \
5138 stmdb sp!, { r4 - r11, r14 }; \
5139 setup_sprite_tiled_initialize_##texture_mode##x4mode(); \
5141 ldr v, [ sp, #36 ]; \
5142 and offset_u, u, #0xF; \
5144 ldr width, [ sp, #40 ]; \
5145 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
5147 ldr height, [ sp, #44 ]; \
5148 add fb_ptr, fb_ptr, y, lsl #11; \
5150 add fb_ptr, fb_ptr, x, lsl #1; \
5151 and offset_v, v, #0xF; \
5153 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
5154 add width_rounded, offset_u, width; \
5156 add height_rounded, offset_v, height; \
5157 add width_rounded, width_rounded, #15; \
5159 add height_rounded, height_rounded, #15; \
5160 mov tile_width, width_rounded, lsr #4; \
5162 /* texture_offset_base = VH-VL-00-00 */\
5163 mov texture_offset_base, v, lsl #8; \
5164 and offset_u_right, width_rounded, #0xF; \
5166 /* texture_offset_base = VH-UH-UL-00 */\
5167 bfi texture_offset_base, u, #4, #8; \
5168 mov right_block_mask, #0xFFFFFFFE; \
5170 setup_sprite_offset_u_adjust##x4mode(); \
5172 /* texture_offset_base = VH-UH-VL-00 */\
5173 bfi texture_offset_base, v, #4, #4; \
5174 mov left_block_mask, #0xFFFFFFFF; \
5176 mov tile_height, height_rounded, lsr #4; \
5177 mvn left_block_mask, left_block_mask, lsl offset_u; \
5179 /* texture_mask = HH-HL-WH-WL */\
5180 ldrh texture_mask, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]; \
5181 mov right_block_mask, right_block_mask, lsl offset_u_right; \
5183 /* texture_mask_rev = WH-WL-HH-HL */\
5184 rev16 texture_mask_rev, texture_mask; \
5185 vmov block_masks, left_block_mask, right_block_mask; \
5187 /* texture_mask = HH-HL-HL-WL */\
5188 bfi texture_mask, texture_mask_rev, #4, #4; \
5189 /* texture_mask_rev = 00-00-00-WH */\
5190 mov texture_mask_rev, texture_mask_rev, lsr #12; \
5192 /* texture_mask = HH-WH-HL-WL */\
5193 bfi texture_mask, texture_mask_rev, #8, #4; \
5194 setup_sprite_get_left_block_mask##x4mode(); \
5196 mov control_mask, #0; \
5197 setup_sprite_compare_left_block_mask##x4mode(); \
5199 setup_sprite_get_right_block_mask##x4mode(); \
5200 orreq control_mask, control_mask, #0x4; \
5202 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
5203 setup_sprite_compare_right_block_mask##x4mode(); \
5205 orreq control_mask, control_mask, #0x8; \
5206 cmp tile_width, #1; \
5208 add block, psx_gpu, #psx_gpu_blocks_offset; \
5209 orreq control_mask, control_mask, #0x1; \
5211 cmp tile_height, #1; \
5212 add block, block, num_blocks, lsl #6; \
5214 orreq control_mask, control_mask, #0x2; \
5215 ldr pc, [ pc, control_mask, lsl #2 ]; \
5218 .word setup_sprite_##texture_mode##_multi_multi_full_full##x4mode; \
5219 .word setup_sprite_##texture_mode##_single_multi_full_none##x4mode; \
5220 .word setup_sprite_##texture_mode##_multi_single_full_full##x4mode; \
5221 .word setup_sprite_##texture_mode##_single_single_full_none##x4mode; \
5222 .word setup_sprite_##texture_mode##_multi_multi_half_full##x4mode; \
5223 .word setup_sprite_##texture_mode##_single_multi_half_right##x4mode; \
5224 .word setup_sprite_##texture_mode##_multi_single_half_full##x4mode; \
5225 .word setup_sprite_##texture_mode##_single_single_half_right##x4mode; \
5226 .word setup_sprite_##texture_mode##_multi_multi_full_half##x4mode; \
5227 .word setup_sprite_##texture_mode##_single_multi_half_left##x4mode; \
5228 .word setup_sprite_##texture_mode##_multi_single_full_half##x4mode; \
5229 .word setup_sprite_##texture_mode##_single_single_half_left##x4mode; \
5230 .word setup_sprite_##texture_mode##_multi_multi_half_half##x4mode; \
5232 .word setup_sprite_##texture_mode##_multi_single_half_half##x4mode; \
5235 setup_sprite_tiled_builder(4bpp,);
5236 setup_sprite_tiled_builder(8bpp,);
5238 #undef draw_mask_fb_ptr_left
5239 #undef draw_mask_fb_ptr_right
5241 setup_sprite_tiled_builder(4bpp, _4x);
5242 setup_sprite_tiled_builder(8bpp, _4x);
5250 #define block_ptr r0
5251 #define num_blocks r1
5253 #define texel_shift_mask r3
5254 #define block_pixels_a r4
5255 #define block_pixels_b r5
5264 #define texels_01 r6
5265 #define texels_23 r7
5266 #define texels_45 r8
5267 #define texels_67 r9
5269 function(texture_sprite_blocks_8bpp)
5270 stmdb sp!, { r4 - r11, r14 }
5271 movw texel_shift_mask, #(0xFF << 1)
5273 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5274 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
5276 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
5277 ldr block_pixels_a, [ block_ptr, #16 ]
5280 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
5281 ldr block_pixels_b, [ block_ptr, #20 ]
5283 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
5284 ldrh texel_0, [ clut_ptr, texel_0 ]
5286 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
5287 ldrh texel_1, [ clut_ptr, texel_1 ]
5289 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
5290 ldr block_pixels_a, [ block_ptr, #(64 + 16) ]
5292 ldrh texel_2, [ clut_ptr, texel_2 ]
5293 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
5295 ldrh texel_3, [ clut_ptr, texel_3 ]
5296 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
5298 ldrh texel_4, [ clut_ptr, texel_4 ]
5299 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
5301 ldrh texel_5, [ clut_ptr, texel_5 ]
5302 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
5304 ldrh texel_6, [ clut_ptr, texel_6 ]
5305 orr texels_01, texel_0, texel_1, lsl #16
5307 ldrh texel_7, [ clut_ptr, texel_7 ]
5308 orr texels_23, texel_2, texel_3, lsl #16
5310 orr texels_45, texel_4, texel_5, lsl #16
5311 str texels_01, [ block_ptr, #0 ]
5313 orr texels_67, texel_6, texel_7, lsl #16
5314 str texels_23, [ block_ptr, #4 ]
5316 subs num_blocks, num_blocks, #1
5317 str texels_45, [ block_ptr, #8 ]
5319 str texels_67, [ block_ptr, #12 ]
5320 add block_ptr, block_ptr, #64
5324 ldmia sp!, { r4 - r11, pc }
5327 #undef width_rounded
5330 #undef texture_offset
5333 #undef texels_wide_low
5334 #undef texels_wide_high
5345 #define left_offset r8
5346 #define width_rounded r9
5347 #define right_width r10
5349 #define block_width r11
5351 #define texture_offset_base r1
5352 #define texture_mask r2
5353 #define texture_page_ptr r3
5354 #define num_blocks r4
5357 #define texture_offset r8
5358 #define blocks_remaining r9
5360 #define fb_ptr_pitch r12
5361 #define texture_block_ptr r14
5363 #define texture_mask_width r2
5364 #define texture_mask_height r3
5365 #define left_mask_bits r4
5366 #define right_mask_bits r5
5370 #undef block_masks_shifted
5373 #define block_masks d0
5374 #define block_masks_shifted d1
5375 #define draw_mask_fb_ptr d2
5378 #define draw_mask_fb_ptr_a d2
5379 #define draw_mask_fb_ptr_b d3
5380 #define texels_low d4
5381 #define texels_high d5
5382 #define texels_wide_low d6
5383 #define texels_wide_high d7
5384 #define texels_wide q3
5387 setup_sprites_16bpp_flush:
5390 stmdb sp!, { r0 - r3, r12, r14 }
5391 bl flush_render_block_buffer
5392 ldmia sp!, { r0 - r3, r12, r14 }
5396 add block, psx_gpu, #psx_gpu_blocks_offset
5397 mov num_blocks, block_width
5401 function(setup_sprite_16bpp)
5402 stmdb sp!, { r4 - r11, r14 }
5403 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5406 add fb_ptr, fb_ptr, y, lsl #11
5408 ldr width, [ sp, #40 ]
5409 add fb_ptr, fb_ptr, x, lsl #1
5411 ldr height, [ sp, #44 ]
5412 and left_offset, u, #0x7
5414 add texture_offset_base, u, u
5415 add width_rounded, width, #7
5417 add texture_offset_base, texture_offset_base, v, lsl #11
5418 mov left_mask_bits, #0xFF
5420 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5421 add width_rounded, width_rounded, left_offset
5423 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5424 sub fb_ptr, fb_ptr, left_offset, lsl #1
5426 add texture_mask, texture_mask_width, texture_mask_width
5427 mov right_mask_bits, #0xFE
5429 and right_width, width_rounded, #0x7
5430 mvn left_mask_bits, left_mask_bits, lsl left_offset
5432 add texture_mask, texture_mask, texture_mask_height, lsl #11
5433 mov block_width, width_rounded, lsr #3
5435 mov right_mask_bits, right_mask_bits, lsl right_width
5436 movw fb_ptr_pitch, #(2048 + 16)
5438 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5439 vmov block_masks, left_mask_bits, right_mask_bits
5441 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5442 add block, psx_gpu, #psx_gpu_blocks_offset
5444 bic texture_offset_base, texture_offset_base, #0xF
5447 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5448 add block, block, num_blocks, lsl #6
5452 vext.32 block_masks_shifted, block_masks, block_masks, #1
5453 vorr.u32 block_masks, block_masks, block_masks_shifted
5454 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5457 add num_blocks, num_blocks, #1
5458 cmp num_blocks, #MAX_BLOCKS
5459 blgt setup_sprites_16bpp_flush
5461 and texture_block_ptr, texture_offset_base, texture_mask
5462 subs height, height, #1
5464 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5465 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5467 vst1.u32 { texels }, [ block, :128 ]
5468 add block, block, #40
5470 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5473 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5475 add block, block, #24
5476 add texture_offset_base, texture_offset_base, #2048
5477 add fb_ptr, fb_ptr, #2048
5478 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5481 ldmia sp!, { r4 - r11, pc }
5484 add num_blocks, num_blocks, block_width
5485 mov texture_offset, texture_offset_base
5487 cmp num_blocks, #MAX_BLOCKS
5488 blgt setup_sprites_16bpp_flush
5490 add texture_offset_base, texture_offset_base, #2048
5491 and texture_block_ptr, texture_offset, texture_mask
5493 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5494 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5496 vst1.u32 { texels }, [ block, :128 ]
5497 add block, block, #40
5499 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5500 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5503 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5504 subs blocks_remaining, block_width, #2
5506 add texture_offset, texture_offset, #16
5507 add fb_ptr, fb_ptr, #16
5509 vmov.u8 draw_mask_fb_ptr, #0
5511 add block, block, #24
5515 and texture_block_ptr, texture_offset, texture_mask
5516 subs blocks_remaining, blocks_remaining, #1
5518 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5519 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5521 vst1.u32 { texels }, [ block, :128 ]
5522 add block, block, #40
5524 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5527 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5529 add texture_offset, texture_offset, #16
5530 add fb_ptr, fb_ptr, #16
5532 add block, block, #24
5536 and texture_block_ptr, texture_offset, texture_mask
5537 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5539 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5540 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5542 vst1.u32 { texels }, [ block, :128 ]
5543 add block, block, #40
5545 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5546 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5548 add block, block, #24
5549 subs height, height, #1
5551 add fb_ptr, fb_ptr, fb_ptr_pitch
5552 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5556 ldmia sp!, { r4 - r11, pc }
5560 // FIXME: duplicate code with normal version :(
5561 #undef draw_mask_fb_ptr
5563 function(setup_sprite_16bpp_4x)
5564 stmdb sp!, { r4 - r11, r14 }
5565 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5568 add fb_ptr, fb_ptr, y, lsl #11
5570 ldr width, [ sp, #40 ]
5571 add fb_ptr, fb_ptr, x, lsl #1
5573 ldr height, [ sp, #44 ]
5574 and left_offset, u, #0x7
5576 add texture_offset_base, u, u
5577 add width_rounded, width, #7
5579 add texture_offset_base, texture_offset_base, v, lsl #11
5580 movw left_mask_bits, #0xFFFF
5582 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5583 add width_rounded, width_rounded, left_offset
5587 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5588 sub fb_ptr, fb_ptr, left_offset, lsl #1
5590 add texture_mask, texture_mask_width, texture_mask_width
5591 movw right_mask_bits, #0xFFFC
5593 and right_width, width_rounded, #0x7
5594 mvn left_mask_bits, left_mask_bits, lsl left_offset
5598 add texture_mask, texture_mask, texture_mask_height, lsl #11
5599 mov block_width, width_rounded, lsr #3
5601 mov right_mask_bits, right_mask_bits, lsl right_width
5602 movw fb_ptr_pitch, #(2048 + 16) * 2
5604 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4+1
5605 vmov block_masks, left_mask_bits, right_mask_bits
5607 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5608 add block, psx_gpu, #psx_gpu_blocks_offset
5610 bic texture_offset_base, texture_offset_base, #0xF
5613 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5614 add block, block, num_blocks, lsl #6
5619 vext.32 block_masks_shifted, block_masks, block_masks, #1
5620 vorr.u32 block_masks, block_masks, block_masks_shifted
5621 vdup.u8 draw_mask_fb_ptr_a, block_masks[0]
5622 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5625 add num_blocks, num_blocks, block_width
5626 cmp num_blocks, #MAX_BLOCKS
5627 blgt setup_sprites_16bpp_flush
5629 and texture_block_ptr, texture_offset_base, texture_mask
5630 subs height, height, #1
5632 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5633 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5635 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5637 add texture_offset_base, texture_offset_base, #2048
5638 add fb_ptr, fb_ptr, #2048*2
5639 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5642 ldmia sp!, { r4 - r11, pc }
5645 add num_blocks, num_blocks, block_width
5646 mov texture_offset, texture_offset_base
5648 vdup.u8 draw_mask_fb_ptr_a, block_masks[0] // left_mask_bits
5649 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5651 cmp num_blocks, #MAX_BLOCKS
5652 blgt setup_sprites_16bpp_flush
5654 add texture_offset_base, texture_offset_base, #2048
5655 and texture_block_ptr, texture_offset, texture_mask
5657 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5658 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5660 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5662 subs blocks_remaining, block_width, #2*4
5663 add texture_offset, texture_offset, #16
5665 vmov.u8 draw_mask_fb_ptr_a, #0
5666 vmov.u8 draw_mask_fb_ptr_b, #0
5668 add fb_ptr, fb_ptr, #16*2
5672 and texture_block_ptr, texture_offset, texture_mask
5673 subs blocks_remaining, blocks_remaining, #4
5675 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5676 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5678 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5679 add texture_offset, texture_offset, #16
5681 add fb_ptr, fb_ptr, #16*2
5685 vdup.u8 draw_mask_fb_ptr_a, block_masks[4] // right_mask_bits
5686 vdup.u8 draw_mask_fb_ptr_b, block_masks[5]
5688 and texture_block_ptr, texture_offset, texture_mask
5689 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5691 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5693 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5694 subs height, height, #1
5696 add fb_ptr, fb_ptr, fb_ptr_pitch
5697 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5701 ldmia sp!, { r4 - r11, pc }
5706 #undef right_mask_bits
5709 #undef blocks_remaining
5719 #define right_width r5
5720 #define right_mask_bits r6
5724 #define fb_ptr_pitch r12
5726 // referenced by setup_sprites_16bpp_flush
5727 #define num_blocks r4
5729 #define block_width r11
5734 #define blocks_remaining r6
5737 #define right_mask q1
5738 #define test_mask q2
5739 #define draw_mask q2
5740 #define draw_mask_bits_fb_ptr d6
5745 function(setup_sprite_untextured)
5746 ldrh r12, [ psx_gpu, #psx_gpu_render_state_offset ]
5747 tst r12, #(RENDER_STATE_MASK_EVALUATE | RENDER_FLAGS_MODULATE_TEXELS \
5748 | RENDER_FLAGS_BLEND)
5749 ldrbeq r12, [ psx_gpu, #psx_gpu_render_mode_offset ]
5750 tsteq r12, #RENDER_INTERLACE_ENABLED
5751 beq setup_sprite_untextured_simple
5753 stmdb sp!, { r4 - r11, r14 }
5755 ldr width, [ sp, #40 ]
5756 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5758 ldr height, [ sp, #44 ]
5759 add fb_ptr, fb_ptr, y, lsl #11
5761 add fb_ptr, fb_ptr, x, lsl #1
5762 sub right_width, width, #1
5764 ldr color, [ sp, #48 ]
5767 add block_width, width, #7
5771 mov right_mask_bits, #0xff
5773 sub fb_ptr_pitch, block_width, #1
5774 lsl right_mask_bits, right_width
5776 lsl fb_ptr_pitch, #3+1
5777 ubfx color_r, color, #3, #5
5779 rsb fb_ptr_pitch, #1024*2
5780 ubfx color_g, color, #11, #5
5782 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
5783 ubfx color_b, color, #19, #5
5785 vdup.u16 right_mask, right_mask_bits
5786 orr color, color_r, color_b, lsl #10
5788 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5789 orr color, color, color_g, lsl #5
5791 vtst.u16 right_mask, right_mask, test_mask
5792 add block, psx_gpu, #psx_gpu_blocks_offset
5794 vdup.u16 colors, color
5795 add block, block, num_blocks, lsl #6
5798 setup_sprite_untextured_height_loop:
5799 add num_blocks, block_width
5800 sub blocks_remaining, block_width, #1
5802 cmp num_blocks, #MAX_BLOCKS
5803 blgt setup_sprites_16bpp_flush
5805 cmp blocks_remaining, #0
5808 vmov.u8 draw_mask, #0 /* zero_mask */
5809 vmov.u8 draw_mask_bits_fb_ptr, #0
5812 vst1.u32 { draw_mask }, [ block, :128 ]!
5813 subs blocks_remaining, #1
5815 vst1.u32 { colors }, [ block, :128 ]
5816 add block, block, #24
5818 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5819 vst1.u32 { draw_mask_bits_fb_ptr }, [ block, :64 ]
5821 add block, block, #24
5826 vst1.u32 { right_mask }, [ block, :128 ]!
5829 vst1.u32 { colors }, [ block, :128 ]
5830 add block, block, #24
5832 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5833 vst1.u32 { draw_mask_bits_fb_ptr }, [ block, :64 ]
5835 add block, block, #24
5836 add fb_ptr, fb_ptr_pitch
5838 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5839 bgt setup_sprite_untextured_height_loop
5841 ldmia sp!, { r4 - r11, pc }
5845 #undef texture_page_ptr
5847 #undef dirty_textures_mask
5848 #undef current_texture_mask
5851 #define current_texture_page r1
5852 #define texture_page_ptr r2
5853 #define vram_ptr_a r3
5854 #define current_texture_page_x r12
5855 #define current_texture_page_y r4
5856 #define dirty_textures_mask r5
5860 #define current_texture_mask r9
5862 #define vram_ptr_b r11
5864 #define texel_block_a d0
5865 #define texel_block_b d1
5866 #define texel_block_expanded_a q1
5867 #define texel_block_expanded_b q2
5868 #define texel_block_expanded_ab q2
5869 #define texel_block_expanded_c q3
5870 #define texel_block_expanded_d q4
5871 #define texel_block_expanded_cd q3
5873 function(update_texture_4bpp_cache)
5874 stmdb sp!, { r4 - r11, r14 }
5877 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5879 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5880 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5882 and current_texture_page_x, current_texture_page, #0xF
5883 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
5885 mov current_texture_page_y, current_texture_page, lsr #4
5886 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5888 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5891 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5892 bic dirty_textures_mask, current_texture_mask
5895 str dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5900 add vram_ptr_b, vram_ptr_a, #2048
5903 vld1.u32 { texel_block_a }, [ vram_ptr_a, :64 ], c_4096
5904 vld1.u32 { texel_block_b }, [ vram_ptr_b, :64 ], c_4096
5906 vmovl.u8 texel_block_expanded_a, texel_block_a
5907 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5908 vmovl.u8 texel_block_expanded_c, texel_block_b
5909 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5911 vbic.u16 texel_block_expanded_a, #0x00F0
5912 vbic.u16 texel_block_expanded_b, #0x00F0
5913 vbic.u16 texel_block_expanded_c, #0x00F0
5914 vbic.u16 texel_block_expanded_d, #0x00F0
5916 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
5917 texel_block_expanded_b
5918 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
5919 texel_block_expanded_d
5921 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
5922 [ texture_page_ptr, :256 ]!
5924 subs sub_y, sub_y, #1
5928 add vram_ptr_a, vram_ptr_a, #8
5929 add vram_ptr_b, vram_ptr_b, #8
5931 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5932 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5934 subs tile_x, tile_x, #1
5938 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5939 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5941 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5942 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5944 subs tile_y, tile_y, #1
5948 ldmia sp!, { r4 - r11, pc }
5951 #undef current_texture_page
5954 #define texture_page r1
5955 #define texture_page_ptr r2
5956 #define vram_ptr_a r3
5957 #define texture_page_x r12
5958 #define texture_page_y r4
5959 #define current_texture_page r5
5964 #define vram_ptr_b r11
5976 function(update_texture_8bpp_cache_slice)
5977 stmdb sp!, { r4 - r11, r14 }
5980 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5981 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5983 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5986 and texture_page_x, texture_page, #0xF
5987 mov texture_page_y, texture_page, lsr #4
5989 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
5992 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
5993 eor current_texture_page, current_texture_page, texture_page
5995 ands current_texture_page, current_texture_page, #0x1
5998 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6001 add vram_ptr_b, vram_ptr_a, #2048
6004 vld1.u32 { texels_a }, [ vram_ptr_a, :128 ], c_4096
6005 vld1.u32 { texels_b }, [ vram_ptr_b, :128 ], c_4096
6006 vld1.u32 { texels_c }, [ vram_ptr_a, :128 ], c_4096
6007 vld1.u32 { texels_d }, [ vram_ptr_b, :128 ], c_4096
6009 vst1.u32 { texels_a, texels_b }, [ texture_page_ptr, :256 ]!
6010 vst1.u32 { texels_c, texels_d }, [ texture_page_ptr, :256 ]!
6012 subs sub_y, sub_y, #1
6017 add vram_ptr_a, vram_ptr_a, #16
6018 add vram_ptr_b, vram_ptr_b, #16
6020 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
6021 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
6023 subs tile_x, tile_x, #1
6028 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
6029 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
6031 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
6032 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
6034 subs tile_y, tile_y, #1
6035 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6040 ldmia sp!, { r4 - r11, pc }
6043 /* void scale2x_tiles8(void *dst, const void *src, int w8, int h) */
6044 function(scale2x_tiles8)
6048 add r12, r0, #1024*2
6052 vld1.u16 { q0 }, [ r1, :128 ]!
6053 vld1.u16 { q2 }, [ r1, :128 ]!
6059 vst1.u16 { q0, q1 }, [ r0, :128 ]!
6060 vst1.u16 { q0, q1 }, [ r12, :128 ]!
6062 vst1.u16 { q2, q3 }, [ r0, :128 ]!
6063 vst1.u16 { q2, q3 }, [ r12, :128 ]!
6070 sub r0, r0, r2, lsl #4+1
6072 add r12, r0, #1024*2
6078 // vim:filetype=armasm
notaz.gp2x.de / pcsx_rearmed.git / blob