2 * Copyright (C) 2011 Gilead Kutnick "Exophase" <exophase@gmail.com>
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public License as
6 * published by the Free Software Foundation; either version 2 of
7 * the License, or (at your option) any later version.
9 * This program is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * General Public License for more details.
17 #define MAX_BLOCKS_PER_ROW 128
19 #include "psx_gpu_offsets.h"
21 #define psx_gpu_b_dx_offset (psx_gpu_b_block_span_offset + 4)
23 #define edge_data_left_x_offset 0
24 #define edge_data_num_blocks_offset 2
25 #define edge_data_right_mask_offset 4
26 #define edge_data_y_offset 6
70 #define gw_by_l gw_bx_l
100 #define uvrg_xxxx2 q5
105 #define yyyy_uvrg0 q6
110 #define yyyy_uvrg1 q7
115 #define yyyy_uvrg2 q8
142 #define uvrg_base q11
188 /* FIXME: users of this should be in psx_gpu instead */
190 #define load_pointer(register, pointer) \
191 movw register, :lower16:pointer; \
192 movt register, :upper16:pointer; \
195 #define load_pointer(register, pointer) \
196 ldr register, =pointer \
200 #define function(name) \
209 function(compute_all_gradients)
210 // First compute the triangle area reciprocal and shift. The division will
211 // happen concurrently with much of the work which follows.
212 @ r12 = psx_gpu->triangle_area
213 ldr r12, [ psx_gpu, #psx_gpu_triangle_area_offset ]
214 stmdb sp!, { r4 - r11, lr }
216 @ load exponent of 62 into upper half of double
218 clz r14, r12 @ r14 = shift
220 movt r4, #((62 + 1023) << 4)
221 mov r12, r12, lsl r14 @ r12 = triangle_area_normalized
223 @ load area normalized into lower half of double
225 vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n
227 movt r4, #((1022 + 31) << 4)
230 add r4, r4, r12, lsr #11
233 vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n
235 // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) =
236 // ( d0 * d1 ) - ( d2 * d3 ) =
237 // ( m0 ) - ( m1 ) = gradient
239 // This is split to do 12 elements at a time over three sets: a, b, and c.
240 // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so
241 // two of the slots are unused.
243 // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as
246 // First type is: uvrg bxxx xxxx
247 // Second type is: yyyy ybyy uvrg
248 // Since x_a and y_c are the same the same variable is used for both.
250 vld1.u32 { v0 }, [ v_a, : 128 ] @ v0 = { uvrg0, b0, x0, y0 }
251 ldrsh x0, [ v_a, #8 ] @ load x0
253 vld1.u32 { v1 }, [ v_b, : 128 ] @ v1 = { uvrg1, b1, x1, y1}
254 ldrh x1, [ v_b, #8 ] @ load x1
256 vld1.u32 { v2 }, [ v_c, : 128 ] @ v2 = { uvrg2, b2, x2, y2 }
257 ldrh x2, [ v_c, #8 ] @ load x2
259 vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- }
260 ldrh y0, [ v_a, #10 ] @ load y0
262 vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- }
263 ldrh y1, [ v_b, #10 ] @ load y1
265 vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- }
266 ldrh y2, [ v_c, #10 ] @ load y2
268 vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 }
269 vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 }
271 orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 }
272 pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 }
274 vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 }
275 vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 }
277 vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 }
278 vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 }
280 ldrb b2, [ v_c, #4 ] @ load b2
281 orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 }
283 ldrb b1, [ v_b, #4 ] @ load b1
284 orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 }
286 vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 }
287 vsub.s16 d0_ab, x1_ab, x0_ab
289 ldrb b0, [ v_a, #4 ] @ load b0
290 orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 }
292 vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 }
293 vsub.s16 d2_ab, x2_ab, x1_ab
295 vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 }
296 vsub.s16 d1_ab, y2_ab, y1_ab
298 orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 }
299 ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 }
301 ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 }
302 ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 }
304 vsub.s16 d3_ab, y1_ab, y0_ab
305 smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) -
306 @ ((x2 - X1) * (b1 - b0))
307 vmull.s16 ga_uvrg_x, d0_a, d1_a
308 smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) -
309 @ ((b2 - b1) * (y1 - y0))
310 vmlsl.s16 ga_uvrg_x, d2_a, d3_a
311 movs gs_bx, ga_bx, asr #31
313 vmull.s16 ga_uvrg_y, d0_b, d1_b
314 rsbmi ga_bx, ga_bx, #0
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 mov b_base, b0, lsl #16
322 rsbmi ga_by, ga_by, #0
323 vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0
325 @ r12 = psx_gpu->triangle_winding_offset
326 ldrb r12, [ psx_gpu, #psx_gpu_triangle_winding_offset ]
327 vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0
329 add b_base, b_base, #0x8000
330 rsb r12, r12, #0 @ r12 = -(triangle->winding)
332 vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w }
333 sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS)
335 vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16
336 vdup.u32 r_shift, r14 @ r_shift = { shift, shift, shift, shift }
338 vorr.u32 uvrg_base, #0x8000
339 vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x)
341 vmov area_r_s, s0 @ area_r_s = triangle_reciprocal
342 vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y)
344 vmull.u32 gw_rg_x, ga_rg_x, d0[0]
345 vmull.u32 gw_uv_x, ga_uv_x, d0[0]
346 vmull.u32 gw_rg_y, ga_rg_y, d0[0]
347 vmull.u32 gw_uv_y, ga_uv_y, d0[0]
349 vshl.u64 gw_rg_x, gw_rg_x, r_shift
350 vshl.u64 gw_uv_x, gw_uv_x, r_shift
351 vshl.u64 gw_rg_y, gw_rg_y, r_shift
352 vshl.u64 gw_uv_y, gw_uv_y, r_shift
354 veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask
355 vmovn.u64 g_uv_x, gw_uv_x
357 veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask
358 vmovn.u64 g_rg_x, gw_rg_x
360 veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
361 vmovn.u64 g_uv_y, gw_uv_y
363 vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
364 vmovn.u64 g_rg_y, gw_rg_y
366 veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
367 mov ga_bx, ga_bx, lsl #13
369 vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
370 mov ga_by, ga_by, lsl #13
373 umull gw_bx_l, gw_bx_h, ga_bx, area_r_s
375 vshl.u32 g_uvrg_x, g_uvrg_x, #4
376 vshl.u32 g_uvrg_y, g_uvrg_y, #4
378 umull gw_by_l, gw_by_h, ga_by, area_r_s
379 vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0]
381 eor gs_bx, gs_bx, r12
382 vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1
384 veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0
385 eor gs_by, gs_by, r12
387 rsb r11, r14, #0 @ r11 = negative shift for scalar lsr
388 add store_a, psx_gpu, #psx_gpu_uvrg_offset
390 sub r11, r11, #(32 - 13)
392 add store_b, store_a, #16
395 vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1
396 vst1.u32 { uvrg_base }, [ store_a, : 128 ], store_inc
398 vst1.u32 { uvrg_dx1 }, [ store_b, : 128 ], store_inc
399 mov g_bx, gw_bx_h, lsr r11
401 vst1.u32 { g_uvrg_y }, [ store_a, : 128 ], store_inc
402 mov g_by, gw_by_h, lsr r11
404 vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \
405 [ store_b, : 128 ], store_inc
406 eor g_bx, g_bx, gs_bx
408 vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \
409 [ store_b, : 128 ], store_inc
410 sub g_bx, g_bx, gs_bx
413 eor g_by, g_by, gs_by
415 mls b_base, g_bx, x0, b_base
416 sub g_by, g_by, gs_by
421 add g_bx2, g_bx, g_bx
422 add g_bx3, g_bx, g_bx2
424 stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by }
426 ldmia sp!, { r4 - r11, pc }
443 #define height_minor_a r7
444 #define height_minor_b r8
445 #define height_major r9
448 #define reciprocal_table_ptr r10
450 #define edge_alt_low r4
451 #define edge_alt_high r5
452 #define edge_dx_dy_alt r6
453 #define edge_shift_alt r10
455 #define edge_dx_dy_alt_low r4
456 #define edge_dx_dy_alt_high r5
458 #define span_edge_data r4
459 #define span_uvrg_offset r5
460 #define span_b_offset r6
468 #define alternate_x q0
469 #define alternate_dx_dy q1
470 #define alternate_x_32 q2
472 #define alternate_x_low d0
473 #define alternate_x_high d1
474 #define alternate_dx_dy_low d2
475 #define alternate_dx_dy_high d3
476 #define alternate_x_32_low d4
477 #define alternate_x_32_high d5
481 #define left_dx_dy q5
482 #define right_dx_dy q6
484 #define right_edge q8
486 #define left_x_low d6
487 #define left_x_high d7
488 #define right_x_low d8
489 #define right_x_high d9
490 #define left_dx_dy_low d10
491 #define left_dx_dy_high d11
492 #define right_dx_dy_low d12
493 #define right_dx_dy_high d13
494 #define left_edge_low d14
495 #define left_edge_high d15
496 #define right_edge_low d16
497 #define right_edge_high d17
499 #define y_mid_point d18
502 #define left_right_x_16 q11
503 #define span_shifts_y q12
506 #define span_shifts d24
511 #define left_right_x_16_low d22
512 #define left_right_x_16_high d23
517 #define alternate_x_16 d4
520 #define v_clip_low d6
522 #define right_x_32 q10
523 #define left_x_32 q11
524 #define alternate_select d24
526 #define right_x_32_low d20
527 #define right_x_32_high d21
528 #define left_x_32_low d22
529 #define left_x_32_high d23
532 #define edges_dx_dy d2
533 #define edge_shifts d3
534 #define edge_shifts_64 q2
536 #define edges_xy_left d0
537 #define edges_xy_right d1
539 #define height_reciprocals d6
547 #define heights_b d12
548 #define edges_dx_dy_64 q10
550 #define edges_dx_dy_64_left d20
551 #define edges_dx_dy_64_right d21
554 #define setup_spans_prologue() \
555 stmdb sp!, { r4 - r11, lr }; \
557 ldrsh x_a, [ v_a, #8 ]; \
558 ldrsh x_b, [ v_b, #8 ]; \
559 ldrsh x_c, [ v_c, #8 ]; \
560 ldrsh y_a, [ v_a, #10 ]; \
561 ldrsh y_b, [ v_b, #10 ]; \
562 ldrsh y_c, [ v_c, #10 ]; \
564 add temp, psx_gpu, #psx_gpu_uvrg_offset; \
565 vld1.32 { uvrg }, [ temp ]; \
566 add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \
567 vld1.32 { uvrg_dy }, [ temp ]; \
568 load_pointer(reciprocal_table_ptr, reciprocal_table); \
570 vmov.u32 c_0x01, #0x01 \
572 #define setup_spans_load_b() \
573 ldr b, [ psx_gpu, #psx_gpu_b_offset ]; \
574 ldr b_dy, [ psx_gpu, #psx_gpu_b_dy_offset ] \
576 #define setup_spans_prologue_b() \
577 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
578 add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \
580 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
581 vmov.u16 c_0x0004, #0x0004; \
583 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
584 vmov.u16 c_0x0001, #0x0001; \
586 vld1.u16 { left_edge_low[], left_edge_high[] }, [ temp ]; \
587 add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \
589 vld1.u16 { right_edge_low[], right_edge_high[] }, [ temp ]; \
590 vadd.u16 right_edge, right_edge, c_0x0001; \
592 vmov.u16 c_0x0007, #0x0007; \
593 vmvn.u16 c_0xFFFE, #0x0001 \
596 #define compute_edge_delta_x2() \
597 ldr temp, [ reciprocal_table_ptr, height, lsl #2 ]; \
599 vdup.u32 heights, height; \
600 vsub.u32 widths, x_ends, x_starts; \
602 vdup.u32 edge_shifts, temp; \
603 vsub.u32 heights_b, heights, c_0x01; \
604 vshr.u32 height_reciprocals, edge_shifts, #10; \
606 vmla.s32 heights_b, x_starts, heights; \
607 vbic.u16 edge_shifts, #0xE0; \
608 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
609 vmull.s32 edges_xy, heights_b, height_reciprocals \
612 #define height_reciprocal_alt r11
613 #define height_b_alt r12
615 #define compute_edge_delta_x3(start_c, height_a, height_b) \
616 vmov.u32 heights, height_a, height_b; \
617 ldr temp, [ reciprocal_table_ptr, height_a, lsl #2 ]; \
618 vmov.u32 edge_shifts[0], temp; \
619 ldr temp, [ reciprocal_table_ptr, height_b, lsl #2 ]; \
620 vmov.u32 edge_shifts[1], temp; \
621 ldr edge_shift_alt, [ reciprocal_table_ptr, height_minor_b, lsl #2 ]; \
623 vsub.u32 widths, x_ends, x_starts; \
624 sub width_alt, x_c, start_c; \
626 vsub.u32 heights_b, heights, c_0x01; \
627 sub height_b_alt, height_minor_b, #1; \
629 vshr.u32 height_reciprocals, edge_shifts, #10; \
630 lsr height_reciprocal_alt, edge_shift_alt, #10; \
632 vmla.s32 heights_b, x_starts, heights; \
633 mla height_b_alt, height_minor_b, start_c, height_b_alt; \
635 vbic.u16 edge_shifts, #0xE0; \
636 and edge_shift_alt, edge_shift_alt, #0x1F; \
638 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
639 mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \
641 vmull.s32 edges_xy, heights_b, height_reciprocals; \
642 smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \
645 #define setup_spans_adjust_y_up() \
646 vsub.u32 y_x4, y_x4, c_0x0004 \
648 #define setup_spans_adjust_y_down() \
649 vadd.u32 y_x4, y_x4, c_0x0004 \
651 #define setup_spans_adjust_interpolants_up() \
652 vsub.u32 uvrg, uvrg, uvrg_dy; \
655 #define setup_spans_adjust_interpolants_down() \
656 vadd.u32 uvrg, uvrg, uvrg_dy; \
660 #define setup_spans_clip_interpolants_increment() \
661 mla b, b_dy, clip, b; \
662 vmla.s32 uvrg, uvrg_dy, v_clip \
664 #define setup_spans_clip_interpolants_decrement() \
665 mls b, b_dy, clip, b; \
666 vmls.s32 uvrg, uvrg_dy, v_clip \
668 #define setup_spans_clip_alternate_yes() \
669 smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \
671 #define setup_spans_clip_alternate_no() \
673 #define setup_spans_clip(direction, alternate_active) \
674 vdup.u32 v_clip, clip; \
675 setup_spans_clip_alternate_##alternate_active(); \
676 setup_spans_clip_interpolants_##direction(); \
677 vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \
680 #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \
681 vmovl.s32 edge_shifts_64, edge_shifts; \
682 vmovl.s32 edges_dx_dy_64, edges_dx_dy; \
684 vshl.s64 edges_xy, edges_xy, edge_shifts_64; \
685 vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \
687 vmov left_x_low, edges_xy_##left_index; \
688 vmov right_x_low, edges_xy_##right_index; \
690 vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \
691 vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \
692 vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \
693 vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \
695 vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \
696 vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \
698 vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \
699 vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \
702 #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \
703 setup_spans_adjust_edges_alternate_no(left_index, right_index); \
705 vdup.u16 y_mid_point, y_b; \
706 rsb temp, edge_shift_alt, #32; \
708 lsl edge_alt_high, edge_alt_high, edge_shift_alt; \
709 orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \
710 lsl edge_alt_low, edge_alt_low, edge_shift_alt; \
711 vmov alternate_x_low, edge_alt_low, edge_alt_high; \
713 asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \
714 lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \
715 vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \
716 vmov alternate_dx_dy_high, alternate_dx_dy_low; \
718 vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \
719 vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \
722 #define setup_spans_y_select_up() \
723 vclt.s16 alternate_select, y_x4, y_mid_point \
725 #define setup_spans_y_select_down() \
726 vcgt.s16 alternate_select, y_x4, y_mid_point \
729 #define setup_spans_alternate_select_left() \
730 vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \
732 #define setup_spans_alternate_select_right() \
733 vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \
736 #define setup_spans_set_x4_alternate_yes(alternate, direction) \
737 vshrn.s64 alternate_x_32_low, alternate_x, #32; \
738 vshrn.s64 left_x_32_low, left_x, #32; \
739 vshrn.s64 right_x_32_low, right_x, #32; \
741 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
742 vadd.u64 left_x, left_x, left_dx_dy; \
743 vadd.u64 right_x, right_x, right_dx_dy; \
745 vshrn.s64 alternate_x_32_high, alternate_x, #32; \
746 vshrn.s64 left_x_32_high, left_x, #32; \
747 vshrn.s64 right_x_32_high, right_x, #32; \
749 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
750 vadd.u64 left_x, left_x, left_dx_dy; \
751 vadd.u64 right_x, right_x, right_dx_dy; \
753 vmovn.u32 alternate_x_16, alternate_x_32; \
754 setup_spans_y_select_##direction(); \
755 vmovn.u32 left_right_x_16_low, left_x_32; \
757 vmovn.u32 left_right_x_16_high, right_x_32; \
758 setup_spans_alternate_select_##alternate(); \
760 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
761 str b, [ span_b_offset ], #4; \
762 setup_spans_adjust_interpolants_##direction(); \
764 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
766 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
767 str b, [ span_b_offset ], #4; \
768 setup_spans_adjust_interpolants_##direction(); \
770 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
772 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
773 str b, [ span_b_offset ], #4; \
774 setup_spans_adjust_interpolants_##direction(); \
776 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
777 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
778 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
780 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
781 str b, [ span_b_offset ], #4; \
782 setup_spans_adjust_interpolants_##direction(); \
784 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
785 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
787 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
789 setup_spans_adjust_y_##direction() \
792 #define setup_spans_set_x4_alternate_no(alternate, direction) \
793 vshrn.s64 left_x_32_low, left_x, #32; \
794 vshrn.s64 right_x_32_low, right_x, #32; \
796 vadd.u64 left_x, left_x, left_dx_dy; \
797 vadd.u64 right_x, right_x, right_dx_dy; \
799 vshrn.s64 left_x_32_high, left_x, #32; \
800 vshrn.s64 right_x_32_high, right_x, #32; \
802 vadd.u64 left_x, left_x, left_dx_dy; \
803 vadd.u64 right_x, right_x, right_dx_dy; \
805 vmovn.u32 left_right_x_16_low, left_x_32; \
806 vmovn.u32 left_right_x_16_high, right_x_32; \
808 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
809 str b, [ span_b_offset ], #4; \
810 setup_spans_adjust_interpolants_##direction(); \
812 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
814 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
815 str b, [ span_b_offset ], #4; \
816 setup_spans_adjust_interpolants_##direction(); \
818 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
820 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
821 str b, [ span_b_offset ], #4; \
822 setup_spans_adjust_interpolants_##direction(); \
824 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
825 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
826 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
828 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
829 str b, [ span_b_offset ], #4; \
830 setup_spans_adjust_interpolants_##direction(); \
832 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
833 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
835 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
837 setup_spans_adjust_y_##direction() \
840 #define edge_adjust_low r11
841 #define edge_adjust_high r12
843 #define setup_spans_alternate_adjust_yes() \
844 smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \
845 subs edge_alt_low, edge_alt_low, edge_adjust_low; \
846 sbc edge_alt_high, edge_alt_high, edge_adjust_high \
848 #define setup_spans_alternate_adjust_no() \
851 #define setup_spans_down(left_index, right_index, alternate, alternate_active) \
852 setup_spans_alternate_adjust_##alternate_active(); \
853 setup_spans_load_b(); \
855 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
856 subs y_c, y_c, temp; \
857 subgt height, height, y_c; \
858 addgt height, height, #1; \
860 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
861 subs clip, temp, y_a; \
864 sub height, height, clip; \
865 add y_a, y_a, clip; \
866 setup_spans_clip(increment, alternate_active); \
872 orr temp, y_a, y_a, lsl #16; \
873 add temp, temp, #(1 << 16); \
875 add y_a, y_a, #(2 << 16); \
876 vmov.u32 y_x4, temp, y_a; \
878 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
880 setup_spans_prologue_b(); \
882 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
885 setup_spans_set_x4_alternate_##alternate_active(alternate, down); \
886 subs height, height, #4; \
892 #define setup_spans_alternate_pre_increment_yes() \
893 adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \
894 adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \
896 #define setup_spans_alternate_pre_increment_no() \
899 #define setup_spans_up_decrement_yes() \
900 suble height, height, #1 \
902 #define setup_spans_up_decrement_no() \
905 #define setup_spans_up(left_index, right_index, alternate, alternate_active) \
906 setup_spans_alternate_adjust_##alternate_active(); \
907 setup_spans_load_b(); \
910 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
911 subs temp, temp, y_c; \
912 subgt height, height, temp; \
913 setup_spans_up_decrement_##alternate_active(); \
915 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
916 subs clip, y_a, temp; \
919 sub height, height, clip; \
920 sub y_a, y_a, clip; \
921 setup_spans_clip(decrement, alternate_active); \
927 orr temp, y_a, y_a, lsl #16; \
928 sub temp, temp, #(1 << 16); \
930 sub y_a, y_a, #(2 << 16); \
931 vmov.u32 y_x4, temp, y_a; \
933 vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \
935 setup_spans_alternate_pre_increment_##alternate_active(); \
936 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
938 setup_spans_adjust_interpolants_up(); \
939 setup_spans_prologue_b(); \
941 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
944 setup_spans_set_x4_alternate_##alternate_active(alternate, up); \
945 subs height, height, #4; \
951 #define setup_spans_epilogue() \
952 ldmia sp!, { r4 - r11, pc } \
955 #define setup_spans_up_up(minor, major) \
956 setup_spans_prologue(); \
957 sub height_minor_a, y_a, y_b; \
958 sub height_minor_b, y_b, y_c; \
959 sub height, y_a, y_c; \
961 vdup.u32 x_starts, x_a; \
962 vmov.u32 x_ends, x_c, x_b; \
964 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
965 setup_spans_up(major, minor, minor, yes); \
966 setup_spans_epilogue() \
968 function(setup_spans_up_left)
969 setup_spans_up_up(left, right)
971 function(setup_spans_up_right)
972 setup_spans_up_up(right, left)
976 #define setup_spans_down_down(minor, major) \
977 setup_spans_prologue(); \
978 sub height_minor_a, y_b, y_a; \
979 sub height_minor_b, y_c, y_b; \
980 sub height, y_c, y_a; \
982 vdup.u32 x_starts, x_a; \
983 vmov.u32 x_ends, x_c, x_b; \
985 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
986 setup_spans_down(major, minor, minor, yes); \
987 setup_spans_epilogue() \
989 function(setup_spans_down_left)
990 setup_spans_down_down(left, right)
992 function(setup_spans_down_right)
993 setup_spans_down_down(right, left)
996 #define setup_spans_up_flat() \
997 sub height, y_a, y_c; \
999 compute_edge_delta_x2(); \
1000 setup_spans_up(left, right, none, no); \
1001 setup_spans_epilogue() \
1003 function(setup_spans_up_a)
1004 setup_spans_prologue()
1006 vmov.u32 x_starts, x_a, x_b
1007 vdup.u32 x_ends, x_c
1009 setup_spans_up_flat()
1011 function(setup_spans_up_b)
1012 setup_spans_prologue()
1014 vdup.u32 x_starts, x_a
1015 vmov.u32 x_ends, x_b, x_c
1017 setup_spans_up_flat()
1019 #define setup_spans_down_flat() \
1020 sub height, y_c, y_a; \
1022 compute_edge_delta_x2(); \
1023 setup_spans_down(left, right, none, no); \
1024 setup_spans_epilogue() \
1026 function(setup_spans_down_a)
1027 setup_spans_prologue()
1029 vmov.u32 x_starts, x_a, x_b
1030 vdup.u32 x_ends, x_c
1032 setup_spans_down_flat()
1034 function(setup_spans_down_b)
1035 setup_spans_prologue()
1037 vdup.u32 x_starts, x_a
1038 vmov.u32 x_ends, x_b, x_c
1040 setup_spans_down_flat()
1045 #define edges_xy_b q11
1046 #define edges_dx_dy_b d26
1047 #define edge_shifts_b d27
1048 #define edges_dx_dy_and_shifts_b q13
1049 #define height_increment d20
1051 #define edges_dx_dy_and_shifts q1
1053 #define edges_xy_b_left d22
1054 #define edges_xy_b_right d23
1056 #define setup_spans_up_down_load_edge_set_b() \
1057 vmov edges_xy, edges_xy_b; \
1058 vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \
1061 function(setup_spans_up_down)
1062 setup_spans_prologue()
1064 // s32 middle_y = y_a;
1065 sub height_minor_a, y_a, y_b
1066 sub height_minor_b, y_c, y_a
1067 sub height_major, y_c, y_b
1069 vmov.u32 x_starts, x_a, x_c
1070 vdup.u32 x_ends, x_b
1072 compute_edge_delta_x3(x_a, height_minor_a, height_major)
1075 vmov.u32 height_increment, temp, height_minor_b
1076 vmlal.s32 edges_xy, edges_dx_dy, height_increment
1078 vmov edges_xy_b_left, edge_alt_low, edge_alt_high
1079 vmov edges_xy_b_right, edges_xy_right
1081 vmov edge_shifts_b, edge_shifts
1082 vmov.u32 edge_shifts_b[0], edge_shift_alt
1084 vneg.s32 edges_dx_dy_b, edges_dx_dy
1085 vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt
1089 setup_spans_load_b()
1092 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1093 subs temp, temp, y_b
1094 subgt height_minor_a, height_minor_a, temp
1096 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1097 subs clip, y_a, temp
1100 sub height_minor_a, height_minor_a, clip
1102 setup_spans_clip(decrement, no)
1105 cmp height_minor_a, #0
1108 orr temp, y_a, y_a, lsl #16
1109 sub temp, temp, #(1 << 16)
1111 sub y_a, y_a, #(2 << 16)
1112 vmov.u32 y_x4, temp, y_a
1114 vaddw.s32 edges_xy, edges_xy, edges_dx_dy
1116 strh height_minor_a, [ psx_gpu, #psx_gpu_num_spans_offset ]
1118 setup_spans_adjust_edges_alternate_no(left, right);
1119 setup_spans_adjust_interpolants_up()
1120 setup_spans_up_down_load_edge_set_b()
1122 setup_spans_prologue_b()
1126 setup_spans_set_x4_alternate_no(none, up)
1127 subs height_minor_a, height_minor_a, #4
1130 add span_edge_data, span_edge_data, height_minor_a, lsl #3
1131 add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4
1132 add span_b_offset, span_b_offset, height_minor_a, lsl #2
1135 add temp, psx_gpu, #psx_gpu_uvrg_offset
1136 vld1.32 { uvrg }, [ temp ]
1139 setup_spans_load_b()
1141 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1143 subgt height_minor_b, height_minor_b, y_c
1144 addgt height_minor_b, height_minor_b, #1
1146 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1147 subs clip, temp, y_a
1150 sub height_minor_b, height_minor_b, clip
1152 setup_spans_clip(increment, no)
1155 cmp height_minor_b, #0
1158 orr temp, y_a, y_a, lsl #16
1159 add temp, temp, #(1 << 16)
1161 add y_a, y_a, #(2 << 16)
1162 vmov.u32 y_x4, temp, y_a
1164 setup_spans_adjust_edges_alternate_no(left, right)
1166 ldrh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1167 add temp, temp, height_minor_b
1168 strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1171 setup_spans_set_x4_alternate_no(none, down)
1172 subs height_minor_b, height_minor_b, #4
1176 setup_spans_epilogue()
1179 setup_spans_up_down_load_edge_set_b()
1180 setup_spans_prologue_b()
1185 #undef span_uvrg_offset
1186 #undef span_edge_data
1187 #undef span_b_offset
1192 #define num_spans r1
1193 #define span_uvrg_offset r2
1194 #define span_edge_data r3
1195 #define span_b_offset r4
1197 #define span_num_blocks r6
1201 #define dither_offset_ptr r10
1202 #define block_ptr_a r11
1204 #define num_blocks r14
1206 #define uvrg_dx_ptr r2
1207 #define texture_mask_ptr r3
1208 #define dither_shift r8
1209 #define dither_row r10
1214 #define block_ptr_b r10
1216 #define block_span_ptr r10
1217 #define right_mask r8
1236 #define b_whole_8 d14
1237 #define fb_mask_ptrs d15
1250 #define u_whole_low d16
1251 #define u_whole_high d17
1252 #define v_whole_low d18
1253 #define v_whole_high d19
1254 #define r_whole_low d20
1255 #define r_whole_high d21
1256 #define g_whole_low d22
1257 #define g_whole_high d23
1258 #define b_whole_low d24
1259 #define b_whole_high d25
1264 #define u_whole_8 d26
1265 #define v_whole_8 d27
1266 #define u_whole_8b d24
1267 #define r_whole_8 d24
1268 #define g_whole_8 d25
1270 #define uv_whole_8 q13
1271 #define uv_whole_8b q14
1273 #define dither_offsets q14
1274 #define texture_mask q15
1275 #define texture_mask_u d30
1276 #define texture_mask_v d31
1278 #define dither_offsets_short d28
1282 #define block_span q10
1287 #define draw_mask q1
1288 #define draw_mask_edge q13
1289 #define test_mask q0
1295 #define setup_blocks_texture_swizzled() \
1296 vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \
1297 vsli.u8 u_whole_8, v_whole_8, #4; \
1298 vsri.u8 v_whole_8, u_whole_8b, #4 \
1300 #define setup_blocks_texture_unswizzled() \
1303 #define setup_blocks_shaded_textured_builder(swizzling) \
1306 function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \
1307 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1308 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1310 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1311 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1313 cmp num_spans, #0; \
1316 stmdb sp!, { r4 - r11, r14 }; \
1317 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1319 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
1320 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1322 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1323 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1325 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1326 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1328 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
1329 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1331 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1334 vmov.u8 fb_mask_ptrs, #0; \
1336 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1337 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1339 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1340 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
1342 cmp span_num_blocks, #0; \
1345 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1346 add num_blocks, span_num_blocks, num_blocks; \
1348 cmp num_blocks, #MAX_BLOCKS; \
1352 ldr b, [ span_b_offset ]; \
1353 add fb_ptr, fb_ptr, y, lsl #11; \
1355 vdup.u32 v_left_x, left_x; \
1358 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1359 add fb_ptr, fb_ptr, left_x, lsl #1; \
1361 mla b, b_dx, left_x, b; \
1362 and dither_shift, left_x, #0x03; \
1364 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1365 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1367 mov dither_shift, dither_shift, lsl #3; \
1368 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1371 subs span_num_blocks, span_num_blocks, #1; \
1373 mov dither_row, dither_row, ror dither_shift; \
1374 mov b_dx4, b_dx, lsl #2; \
1376 vdup.u32 dither_offsets_short, dither_row; \
1377 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1379 vdup.u32 b_block, b; \
1380 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1382 vdup.u32 u_block, uv[0]; \
1383 mov b_dx8, b_dx, lsl #3; \
1385 vdup.u32 v_block, uv[1]; \
1386 vdup.u32 r_block, rg[0]; \
1387 vdup.u32 g_block, rg[1]; \
1389 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1391 vadd.u32 u_block, u_block, block_span; \
1392 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1394 vadd.u32 v_block, v_block, block_span; \
1395 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1397 vadd.u32 r_block, r_block, block_span; \
1398 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1400 vadd.u32 g_block, g_block, block_span; \
1401 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
1403 vadd.u32 b_block, b_block, block_span; \
1404 add block_ptr_b, block_ptr_a, #16; \
1406 vshrn.u32 u_whole_low, u_block, #16; \
1407 vshrn.u32 v_whole_low, v_block, #16; \
1408 vshrn.u32 r_whole_low, r_block, #16; \
1409 vshrn.u32 g_whole_low, g_block, #16; \
1411 vdup.u32 dx4, uv_dx4[0]; \
1412 vshrn.u32 b_whole_low, b_block, #16; \
1414 vaddhn.u32 u_whole_high, u_block, dx4; \
1415 vdup.u32 dx4, uv_dx4[1]; \
1417 vaddhn.u32 v_whole_high, v_block, dx4; \
1418 vdup.u32 dx4, rg_dx4[0]; \
1420 vaddhn.u32 r_whole_high, r_block, dx4; \
1421 vdup.u32 dx4, rg_dx4[1]; \
1423 vaddhn.u32 g_whole_high, g_block, dx4; \
1424 vdup.u32 dx4, b_dx4; \
1426 vaddhn.u32 b_whole_high, b_block, dx4; \
1427 vdup.u32 dx8, uv_dx8[0]; \
1429 vadd.u32 u_block, u_block, dx8; \
1430 vdup.u32 dx8, uv_dx8[1]; \
1432 vadd.u32 v_block, v_block, dx8; \
1433 vdup.u32 dx8, rg_dx8[0]; \
1435 vadd.u32 r_block, r_block, dx8; \
1436 vdup.u32 dx8, rg_dx8[1]; \
1438 vadd.u32 g_block, g_block, dx8; \
1439 vdup.u32 dx8, b_dx8; \
1441 vadd.u32 b_block, b_block, dx8; \
1442 vmovn.u16 u_whole_8, u_whole; \
1444 vmovn.u16 v_whole_8, v_whole; \
1446 vmovn.u16 b_whole_8, b_whole; \
1448 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1450 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1451 setup_blocks_texture_##swizzling(); \
1453 vmovn.u16 r_whole_8, r_whole; \
1457 vmovn.u16 g_whole_8, g_whole; \
1458 vshrn.u32 u_whole_low, u_block, #16; \
1460 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1461 vshrn.u32 v_whole_low, v_block, #16; \
1463 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1464 vshrn.u32 r_whole_low, r_block, #16; \
1466 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1467 vshrn.u32 g_whole_low, g_block, #16; \
1469 vdup.u32 dx4, uv_dx4[0]; \
1470 vshrn.u32 b_whole_low, b_block, #16; \
1472 vaddhn.u32 u_whole_high, u_block, dx4; \
1473 vdup.u32 dx4, uv_dx4[1]; \
1475 vaddhn.u32 v_whole_high, v_block, dx4; \
1476 vdup.u32 dx4, rg_dx4[0]; \
1478 vaddhn.u32 r_whole_high, r_block, dx4; \
1479 vdup.u32 dx4, rg_dx4[1]; \
1481 vaddhn.u32 g_whole_high, g_block, dx4; \
1482 vdup.u32 dx4, b_dx4; \
1484 vaddhn.u32 b_whole_high, b_block, dx4; \
1485 vdup.u32 dx8, uv_dx8[0]; \
1487 vadd.u32 u_block, u_block, dx8; \
1488 vdup.u32 dx8, uv_dx8[1]; \
1490 vadd.u32 v_block, v_block, dx8; \
1491 vdup.u32 dx8, rg_dx8[0]; \
1493 vadd.u32 r_block, r_block, dx8; \
1494 vdup.u32 dx8, rg_dx8[1]; \
1496 vadd.u32 g_block, g_block, dx8; \
1497 vdup.u32 dx8, b_dx8; \
1499 vadd.u32 b_block, b_block, dx8; \
1500 vmovn.u16 u_whole_8, u_whole; \
1502 add fb_ptr, fb_ptr, #16; \
1503 vmovn.u16 v_whole_8, v_whole; \
1505 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1506 vmovn.u16 b_whole_8, b_whole; \
1510 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1511 subs span_num_blocks, span_num_blocks, #1; \
1513 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1514 setup_blocks_texture_##swizzling(); \
1516 vmovn.u16 r_whole_8, r_whole; \
1520 vmovn.u16 g_whole_8, g_whole; \
1521 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1523 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1524 vdup.u8 draw_mask, right_mask; \
1526 vmov.u32 fb_mask_ptrs[0], right_mask; \
1527 vtst.u16 draw_mask, draw_mask, test_mask; \
1528 vzip.u8 u_whole_8, v_whole_8; \
1530 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1531 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1532 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1533 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1534 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1537 add span_uvrg_offset, span_uvrg_offset, #16; \
1538 add span_b_offset, span_b_offset, #4; \
1540 add span_edge_data, span_edge_data, #8; \
1541 subs num_spans, num_spans, #1; \
1543 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1546 ldmia sp!, { r4 - r11, pc }; \
1549 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1550 vpush { texture_mask }; \
1551 vpush { uvrg_dx4 }; \
1553 stmdb sp!, { r0 - r3, r12, r14 }; \
1554 bl flush_render_block_buffer; \
1555 ldmia sp!, { r0 - r3, r12, r14 }; \
1557 vpop { uvrg_dx4 }; \
1558 vpop { texture_mask }; \
1560 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1561 vmov.u8 fb_mask_ptrs, #0; \
1563 mov num_blocks, span_num_blocks; \
1564 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1568 setup_blocks_shaded_textured_builder(swizzled)
1569 setup_blocks_shaded_textured_builder(unswizzled)
1572 #define setup_blocks_unshaded_textured_builder(swizzling) \
1575 function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \
1576 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1577 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1579 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1580 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1582 cmp num_spans, #0; \
1585 stmdb sp!, { r4 - r11, r14 }; \
1586 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1588 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1590 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1591 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1593 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1594 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1596 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1598 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1601 vmov.u8 fb_mask_ptrs, #0; \
1603 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1604 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1606 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1607 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
1609 cmp span_num_blocks, #0; \
1612 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1613 add num_blocks, span_num_blocks, num_blocks; \
1615 cmp num_blocks, #MAX_BLOCKS; \
1619 add fb_ptr, fb_ptr, y, lsl #11; \
1621 vdup.u32 v_left_x, left_x; \
1624 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1625 add fb_ptr, fb_ptr, left_x, lsl #1; \
1627 and dither_shift, left_x, #0x03; \
1629 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1630 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1632 mov dither_shift, dither_shift, lsl #3; \
1633 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1636 subs span_num_blocks, span_num_blocks, #1; \
1638 mov dither_row, dither_row, ror dither_shift; \
1640 vdup.u32 dither_offsets_short, dither_row; \
1641 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1643 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1645 vdup.u32 u_block, uv[0]; \
1647 vdup.u32 v_block, uv[1]; \
1648 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1650 vadd.u32 u_block, u_block, block_span; \
1651 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1653 vadd.u32 v_block, v_block, block_span; \
1654 add block_ptr_b, block_ptr_a, #16; \
1656 vshrn.u32 u_whole_low, u_block, #16; \
1657 vshrn.u32 v_whole_low, v_block, #16; \
1659 vdup.u32 dx4, uv_dx4[0]; \
1661 vaddhn.u32 u_whole_high, u_block, dx4; \
1662 vdup.u32 dx4, uv_dx4[1]; \
1664 vaddhn.u32 v_whole_high, v_block, dx4; \
1665 vdup.u32 dx8, uv_dx8[0]; \
1667 vadd.u32 u_block, u_block, dx8; \
1668 vdup.u32 dx8, uv_dx8[1]; \
1670 vadd.u32 v_block, v_block, dx8; \
1671 vmovn.u16 u_whole_8, u_whole; \
1673 vmovn.u16 v_whole_8, v_whole; \
1676 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1678 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1679 setup_blocks_texture_##swizzling(); \
1684 vshrn.u32 u_whole_low, u_block, #16; \
1686 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1687 vshrn.u32 v_whole_low, v_block, #16; \
1689 add block_ptr_b, block_ptr_b, #32; \
1690 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1692 vdup.u32 dx4, uv_dx4[0]; \
1693 vaddhn.u32 u_whole_high, u_block, dx4; \
1694 vdup.u32 dx4, uv_dx4[1]; \
1696 vaddhn.u32 v_whole_high, v_block, dx4; \
1697 vdup.u32 dx8, uv_dx8[0]; \
1699 vadd.u32 u_block, u_block, dx8; \
1700 vdup.u32 dx8, uv_dx8[1]; \
1702 vadd.u32 v_block, v_block, dx8; \
1703 vmovn.u16 u_whole_8, u_whole; \
1705 add fb_ptr, fb_ptr, #16; \
1706 vmovn.u16 v_whole_8, v_whole; \
1708 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1711 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1712 subs span_num_blocks, span_num_blocks, #1; \
1714 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1715 setup_blocks_texture_##swizzling(); \
1720 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1722 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1723 vdup.u8 draw_mask, right_mask; \
1725 vmov.u32 fb_mask_ptrs[0], right_mask; \
1726 vtst.u16 draw_mask, draw_mask, test_mask; \
1727 vzip.u8 u_whole_8, v_whole_8; \
1729 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1730 add block_ptr_b, block_ptr_b, #32; \
1731 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1732 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1733 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1736 add span_uvrg_offset, span_uvrg_offset, #16; \
1737 add span_edge_data, span_edge_data, #8; \
1738 subs num_spans, num_spans, #1; \
1740 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1743 ldmia sp!, { r4 - r11, pc }; \
1746 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1747 vpush { texture_mask }; \
1748 vpush { uvrg_dx4 }; \
1750 stmdb sp!, { r0 - r3, r12, r14 }; \
1751 bl flush_render_block_buffer; \
1752 ldmia sp!, { r0 - r3, r12, r14 }; \
1754 vpop { uvrg_dx4 }; \
1755 vpop { texture_mask }; \
1757 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1758 vmov.u8 fb_mask_ptrs, #0; \
1760 mov num_blocks, span_num_blocks; \
1761 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1765 setup_blocks_unshaded_textured_builder(swizzled)
1766 setup_blocks_unshaded_textured_builder(unswizzled)
1771 function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect)
1772 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1773 veor.u32 draw_mask, draw_mask, draw_mask
1778 stmdb sp!, { r4 - r11, r14 }
1779 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1781 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1783 ubfx color_r, color, #3, #5
1784 ubfx color_g, color, #11, #5
1785 ubfx color_b, color, #19, #5
1787 orr color, color_r, color_b, lsl #10
1788 orr color, color, color_g, lsl #5
1790 vdup.u16 colors, color
1792 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1793 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1795 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1796 add block_ptr_a, block_ptr_a, num_blocks, lsl #6
1799 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1800 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1802 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
1804 cmp span_num_blocks, #0
1807 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1808 add num_blocks, span_num_blocks, num_blocks
1810 cmp num_blocks, #MAX_BLOCKS
1814 add fb_ptr, fb_ptr, y, lsl #11
1817 add fb_ptr, fb_ptr, left_x, lsl #1
1820 subs span_num_blocks, span_num_blocks, #1
1822 add block_ptr_b, block_ptr_a, #16
1825 vmov.u32 fb_mask_ptrs[1], fb_ptr
1829 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_32
1830 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1831 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1833 add fb_ptr, fb_ptr, #16
1834 add block_ptr_b, block_ptr_b, #32
1838 vmov.u32 fb_mask_ptrs[1], fb_ptr
1839 subs span_num_blocks, span_num_blocks, #1
1844 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1846 vdup.u8 draw_mask_edge, right_mask
1847 vtst.u16 draw_mask_edge, draw_mask_edge, test_mask
1849 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1850 vst1.u32 { draw_mask_edge }, [ block_ptr_a, :128 ], c_32
1851 add block_ptr_b, block_ptr_b, #32
1852 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1855 add span_edge_data, span_edge_data, #8
1856 subs num_spans, num_spans, #1
1858 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1861 ldmia sp!, { r4 - r11, pc }
1866 stmdb sp!, { r0 - r3, r12, r14 }
1867 bl flush_render_block_buffer
1868 ldmia sp!, { r0 - r3, r12, r14 }
1872 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1873 veor.u32 draw_mask, draw_mask, draw_mask
1875 mov num_blocks, span_num_blocks
1876 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1880 #define mask_msb_scalar r14
1882 #define msb_mask q15
1884 #define pixels_low d16
1886 #define msb_mask_low d30
1887 #define msb_mask_high d31
1892 function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct)
1893 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1898 stmdb sp!, { r4 - r11, r14 }
1900 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1902 ubfx color_r, color, #3, #5
1903 ubfx color_g, color, #11, #5
1905 ldrh mask_msb_scalar, [ psx_gpu, #psx_gpu_mask_msb_offset ]
1906 ubfx color_b, color, #19, #5
1908 orr color, color_r, color_b, lsl #10
1909 orr color, color, color_g, lsl #5
1910 orr color, color, mask_msb_scalar
1912 vdup.u16 colors, color
1914 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1915 orr color, color, lsl #16
1919 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1920 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1922 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
1924 cmp span_num_blocks, #0
1927 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1929 add fb_ptr, fb_ptr, y, lsl #11
1930 subs span_num_blocks, span_num_blocks, #1
1932 add fb_ptr, fb_ptr, left_x, lsl #1
1936 vst1.u32 { colors }, [ fb_ptr ]!
1937 subs span_num_blocks, span_num_blocks, #1
1942 ldrb right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1944 cmp right_mask, #0x0
1947 tst right_mask, #0xF
1948 streq color, [ fb_ptr ], #4
1949 moveq right_mask, right_mask, lsr #4
1950 streq color, [ fb_ptr ], #4
1952 tst right_mask, #0x3
1953 streq color, [ fb_ptr ], #4
1954 moveq right_mask, right_mask, lsr #2
1956 tst right_mask, #0x1
1957 streqh color, [ fb_ptr ]
1960 add span_edge_data, span_edge_data, #8
1961 subs num_spans, num_spans, #1
1964 ldmia sp!, { r4 - r11, pc }
1967 vst1.u32 { colors }, [ fb_ptr ]
1974 #define rg_dx_ptr r2
1992 #undef dither_offsets
2012 #define r_whole_low d6
2013 #define r_whole_high d7
2014 #define g_whole_low d8
2015 #define g_whole_high d9
2016 #define b_whole_low d10
2017 #define b_whole_high d11
2019 #define gb_whole_8 q6
2021 #define g_whole_8 d12
2022 #define b_whole_8 d13
2024 #define r_whole_8 d14
2036 #define block_span q5
2044 #define d128_0x7 q13
2048 #define dither_offsets q14
2049 #define draw_mask q15
2051 #define dither_offsets_low d28
2054 #define test_mask q10
2057 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2058 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2059 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2061 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2062 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2063 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2065 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2067 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2070 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2073 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2074 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2075 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2077 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2079 cmp num_spans, #0; \
2082 stmdb sp!, { r4 - r11, r14 }; \
2083 vshl.u32 rg_dx4, rg_dx, #2; \
2085 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2086 vshl.u32 rg_dx8, rg_dx, #3; \
2088 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2090 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2091 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2093 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2094 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2096 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2097 vmov.u8 d64_1, #1; \
2099 vmov.u8 d128_4, #4; \
2100 vmov.u8 d64_128, #128; \
2102 vmov.u8 d128_0x7, #0x7; \
2105 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2106 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2108 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2109 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
2111 cmp span_num_blocks, #0; \
2114 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2115 add num_blocks, span_num_blocks, num_blocks; \
2117 cmp num_blocks, #MAX_BLOCKS; \
2121 ldr b, [ span_b_offset ]; \
2122 add fb_ptr, fb_ptr, y, lsl #11; \
2124 vdup.u32 v_left_x, left_x; \
2127 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2128 add fb_ptr, fb_ptr, left_x, lsl #1; \
2130 mla b, b_dx, left_x, b; \
2131 and dither_shift, left_x, #0x03; \
2133 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2134 vshr.u32 rg_dx, rg_dx4, #2; \
2136 mov dither_shift, dither_shift, lsl #3; \
2137 vmla.u32 rg, rg_dx, v_left_x; \
2140 subs span_num_blocks, span_num_blocks, #1; \
2142 mov dither_row, dither_row, ror dither_shift; \
2143 mov b_dx4, b_dx, lsl #2; \
2145 vdup.u32 dither_offsets, dither_row; \
2146 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2148 vdup.u32 b_block, b; \
2149 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2151 mov b_dx8, b_dx, lsl #3; \
2152 vdup.u32 r_block, rg[0]; \
2153 vdup.u32 g_block, rg[1]; \
2155 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2157 vadd.u32 r_block, r_block, block_span; \
2158 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2160 vadd.u32 g_block, g_block, block_span; \
2161 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2163 vadd.u32 b_block, b_block, block_span; \
2164 add block_ptr_b, block_ptr_a, #16; \
2166 vshrn.u32 r_whole_low, r_block, #16; \
2167 vshrn.u32 g_whole_low, g_block, #16; \
2168 vshrn.u32 b_whole_low, b_block, #16; \
2169 vdup.u32 dx4, rg_dx4[0]; \
2171 vaddhn.u32 r_whole_high, r_block, dx4; \
2172 vdup.u32 dx4, rg_dx4[1]; \
2174 vaddhn.u32 g_whole_high, g_block, dx4; \
2175 vdup.u32 dx4, b_dx4; \
2177 vaddhn.u32 b_whole_high, b_block, dx4; \
2178 vdup.u32 dx8, rg_dx8[0]; \
2180 vadd.u32 r_block, r_block, dx8; \
2181 vdup.u32 dx8, rg_dx8[1]; \
2183 vadd.u32 g_block, g_block, dx8; \
2184 vdup.u32 dx8, b_dx8; \
2186 vadd.u32 b_block, b_block, dx8; \
2188 vmovn.u16 r_whole_8, r_whole; \
2189 vmovn.u16 g_whole_8, g_whole; \
2190 vmovn.u16 b_whole_8, b_whole; \
2193 veor.u32 draw_mask, draw_mask, draw_mask; \
2196 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2197 vshrn.u32 r_whole_low, r_block, #16; \
2199 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2200 vshrn.u32 g_whole_low, g_block, #16; \
2202 vshrn.u32 b_whole_low, b_block, #16; \
2203 str fb_ptr, [ block_ptr_a, #44 ]; \
2205 vdup.u32 dx4, rg_dx4[0]; \
2206 vshr.u8 r_whole_8, r_whole_8, #3; \
2207 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2209 vaddhn.u32 r_whole_high, r_block, dx4; \
2210 vdup.u32 dx4, rg_dx4[1]; \
2212 vaddhn.u32 g_whole_high, g_block, dx4; \
2213 vdup.u32 dx4, b_dx4; \
2215 vaddhn.u32 b_whole_high, b_block, dx4; \
2216 vdup.u32 dx8, rg_dx8[0]; \
2218 vmull.u8 pixels, r_whole_8, d64_1; \
2219 vmlal.u8 pixels, g_whole_8, d64_4; \
2220 vmlal.u8 pixels, b_whole_8, d64_128; \
2222 vadd.u32 r_block, r_block, dx8; \
2223 vdup.u32 dx8, rg_dx8[1]; \
2225 vadd.u32 g_block, g_block, dx8; \
2226 vdup.u32 dx8, b_dx8; \
2228 vadd.u32 b_block, b_block, dx8; \
2229 add fb_ptr, fb_ptr, #16; \
2231 vmovn.u16 r_whole_8, r_whole; \
2232 vmovn.u16 g_whole_8, g_whole; \
2233 vmovn.u16 b_whole_8, b_whole; \
2235 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2236 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2240 subs span_num_blocks, span_num_blocks, #1; \
2244 str fb_ptr, [ block_ptr_a, #44 ]; \
2245 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2247 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2248 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2250 vshr.u8 r_whole_8, r_whole_8, #3; \
2251 vdup.u8 draw_mask, right_mask; \
2253 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2254 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
2256 vtst.u16 draw_mask, draw_mask, test_mask; \
2258 vmull.u8 pixels, r_whole_8, d64_1; \
2259 vmlal.u8 pixels, g_whole_8, d64_4; \
2260 vmlal.u8 pixels, b_whole_8, d64_128; \
2262 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2263 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2266 add span_uvrg_offset, span_uvrg_offset, #16; \
2267 add span_b_offset, span_b_offset, #4; \
2269 add span_edge_data, span_edge_data, #8; \
2270 subs num_spans, num_spans, #1; \
2272 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2275 ldmia sp!, { r4 - r11, pc }; \
2278 /* TODO: Load from psx_gpu instead of saving/restoring these */\
2281 stmdb sp!, { r0 - r3, r12, r14 }; \
2282 bl flush_render_block_buffer; \
2283 ldmia sp!, { r0 - r3, r12, r14 }; \
2287 vmov.u8 d64_1, #1; \
2288 vmov.u8 d128_4, #4; \
2289 vmov.u8 d64_128, #128; \
2290 vmov.u8 d128_0x7, #0x7; \
2292 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2294 mov num_blocks, span_num_blocks; \
2295 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2299 setup_blocks_shaded_untextured_indirect_builder(undithered)
2300 setup_blocks_shaded_untextured_indirect_builder(dithered)
2305 #define mask_msb_ptr r14
2307 #define draw_mask q0
2308 #define pixels_low d16
2309 #define pixels_high d17
2313 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2316 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2317 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2318 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2320 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2322 cmp num_spans, #0; \
2325 stmdb sp!, { r4 - r11, r14 }; \
2326 vshl.u32 rg_dx4, rg_dx, #2; \
2328 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2329 vshl.u32 rg_dx8, rg_dx, #3; \
2331 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2332 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2334 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2335 vmov.u8 d64_1, #1; \
2337 vmov.u8 d128_4, #4; \
2338 vmov.u8 d64_128, #128; \
2340 vmov.u8 d128_0x7, #0x7; \
2341 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2342 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
2345 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2346 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2348 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2349 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
2351 cmp span_num_blocks, #0; \
2354 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2355 add fb_ptr, fb_ptr, y, lsl #11; \
2357 ldr b, [ span_b_offset ]; \
2358 vdup.u32 v_left_x, left_x; \
2361 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2362 add fb_ptr, fb_ptr, left_x, lsl #1; \
2364 mla b, b_dx, left_x, b; \
2365 and dither_shift, left_x, #0x03; \
2367 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2368 vshr.u32 rg_dx, rg_dx4, #2; \
2370 mov dither_shift, dither_shift, lsl #3; \
2371 vmla.u32 rg, rg_dx, v_left_x; \
2373 subs span_num_blocks, span_num_blocks, #1; \
2375 mov dither_row, dither_row, ror dither_shift; \
2376 mov b_dx4, b_dx, lsl #2; \
2378 vdup.u32 dither_offsets, dither_row; \
2379 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2381 vdup.u32 b_block, b; \
2382 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2384 mov b_dx8, b_dx, lsl #3; \
2385 vdup.u32 r_block, rg[0]; \
2386 vdup.u32 g_block, rg[1]; \
2388 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2390 vadd.u32 r_block, r_block, block_span; \
2391 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2393 vadd.u32 g_block, g_block, block_span; \
2394 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2396 vadd.u32 b_block, b_block, block_span; \
2397 add block_ptr_b, block_ptr_a, #16; \
2399 vshrn.u32 r_whole_low, r_block, #16; \
2400 vshrn.u32 g_whole_low, g_block, #16; \
2401 vshrn.u32 b_whole_low, b_block, #16; \
2402 vdup.u32 dx4, rg_dx4[0]; \
2404 vaddhn.u32 r_whole_high, r_block, dx4; \
2405 vdup.u32 dx4, rg_dx4[1]; \
2407 vaddhn.u32 g_whole_high, g_block, dx4; \
2408 vdup.u32 dx4, b_dx4; \
2410 vaddhn.u32 b_whole_high, b_block, dx4; \
2411 vdup.u32 dx8, rg_dx8[0]; \
2413 vadd.u32 r_block, r_block, dx8; \
2414 vdup.u32 dx8, rg_dx8[1]; \
2416 vadd.u32 g_block, g_block, dx8; \
2417 vdup.u32 dx8, b_dx8; \
2419 vadd.u32 b_block, b_block, dx8; \
2421 vmovn.u16 r_whole_8, r_whole; \
2422 vmovn.u16 g_whole_8, g_whole; \
2423 vmovn.u16 b_whole_8, b_whole; \
2428 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2429 vshrn.u32 r_whole_low, r_block, #16; \
2431 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2432 vshrn.u32 g_whole_low, g_block, #16; \
2434 vshrn.u32 b_whole_low, b_block, #16; \
2436 vdup.u32 dx4, rg_dx4[0]; \
2437 vshr.u8 r_whole_8, r_whole_8, #3; \
2438 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2440 vaddhn.u32 r_whole_high, r_block, dx4; \
2441 vdup.u32 dx4, rg_dx4[1]; \
2443 vmov pixels, msb_mask; \
2444 vaddhn.u32 g_whole_high, g_block, dx4; \
2445 vdup.u32 dx4, b_dx4; \
2447 vaddhn.u32 b_whole_high, b_block, dx4; \
2448 vdup.u32 dx8, rg_dx8[0]; \
2450 vmlal.u8 pixels, r_whole_8, d64_1; \
2451 vmlal.u8 pixels, g_whole_8, d64_4; \
2452 vmlal.u8 pixels, b_whole_8, d64_128; \
2454 vadd.u32 r_block, r_block, dx8; \
2455 vdup.u32 dx8, rg_dx8[1]; \
2457 vadd.u32 g_block, g_block, dx8; \
2458 vdup.u32 dx8, b_dx8; \
2460 vadd.u32 b_block, b_block, dx8; \
2462 vmovn.u16 r_whole_8, r_whole; \
2463 vmovn.u16 g_whole_8, g_whole; \
2464 vmovn.u16 b_whole_8, b_whole; \
2466 vst1.u32 { pixels }, [ fb_ptr ]!; \
2467 subs span_num_blocks, span_num_blocks, #1; \
2471 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2473 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2474 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2476 vshr.u8 r_whole_8, r_whole_8, #3; \
2477 rbit right_mask, right_mask; \
2478 vmov pixels, msb_mask; \
2479 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2480 clz right_mask, right_mask; \
2482 vmlal.u8 pixels, r_whole_8, d64_1; \
2483 vmlal.u8 pixels, g_whole_8, d64_4; \
2484 vmlal.u8 pixels, b_whole_8, d64_128; \
2486 ldr pc, [ pc, right_mask, lsl #2 ]; \
2499 vst1.u16 { pixels_low[0] }, [ fb_ptr ]; \
2503 vst1.u32 { pixels_low[0] }, [ fb_ptr ]; \
2507 vst1.u32 { pixels_low[0] }, [ fb_ptr ]!; \
2508 vst1.u16 { pixels_low[2] }, [ fb_ptr ]; \
2512 vst1.u32 { pixels_low }, [ fb_ptr ]; \
2516 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2517 vst1.u16 { pixels_high[0] }, [ fb_ptr ]; \
2521 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2522 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2526 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2527 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2528 vst1.u16 { pixels_high[2] }, [ fb_ptr ]; \
2532 vst1.u32 { pixels }, [ fb_ptr ]; \
2536 add span_uvrg_offset, span_uvrg_offset, #16; \
2537 add span_b_offset, span_b_offset, #4; \
2539 add span_edge_data, span_edge_data, #8; \
2540 subs num_spans, num_spans, #1; \
2544 ldmia sp!, { r4 - r11, pc } \
2546 setup_blocks_shaded_untextured_direct_builder(undithered)
2547 setup_blocks_shaded_untextured_direct_builder(dithered)
2556 #define block_ptr r1
2557 #define num_blocks r2
2570 #define texture_ptr r11
2584 #define pixels_d r10
2588 #define clut_ptr r12
2589 #define current_texture_mask r5
2590 #define dirty_textures_mask r6
2594 #define clut_low_a d2
2595 #define clut_low_b d3
2596 #define clut_high_a d4
2597 #define clut_high_b d5
2602 #define texels_low d6
2603 #define texels_high d7
2607 function(texture_blocks_untextured)
2613 function(texture_blocks_4bpp)
2614 stmdb sp!, { r3 - r11, r14 }
2615 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2617 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2618 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2620 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2621 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]
2623 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2624 vuzp.u8 clut_a, clut_b
2626 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
2627 tst dirty_textures_mask, current_texture_mask
2633 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2635 uxtah uv_0, texture_ptr, uv_01
2636 uxtah uv_1, texture_ptr, uv_01, ror #16
2638 uxtah uv_2, texture_ptr, uv_23
2639 uxtah uv_3, texture_ptr, uv_23, ror #16
2641 uxtah uv_4, texture_ptr, uv_45
2642 ldrb pixel_0, [ uv_0 ]
2644 uxtah uv_5, texture_ptr, uv_45, ror #16
2645 ldrb pixel_1, [ uv_1 ]
2647 uxtah uv_6, texture_ptr, uv_67
2648 ldrb pixel_2, [ uv_2 ]
2650 uxtah uv_7, texture_ptr, uv_67, ror #16
2651 ldrb pixel_3, [ uv_3 ]
2653 ldrb pixel_4, [ uv_4 ]
2654 subs num_blocks, num_blocks, #1
2656 ldrb pixel_5, [ uv_5 ]
2657 orr pixels_a, pixel_0, pixel_1, lsl #8
2659 ldrb pixel_6, [ uv_6 ]
2660 orr pixels_b, pixel_4, pixel_5, lsl #8
2662 ldrb pixel_7, [ uv_7 ]
2663 orr pixels_a, pixels_a, pixel_2, lsl #16
2665 orr pixels_b, pixels_b, pixel_6, lsl #16
2666 orr pixels_a, pixels_a, pixel_3, lsl #24
2668 orr pixels_b, pixels_b, pixel_7, lsl #24
2669 vmov.u32 texels, pixels_a, pixels_b
2671 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2672 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2674 vst2.u8 { texels_low, texels_high }, [ block_ptr, :128 ], c_64
2677 ldmia sp!, { r3 - r11, pc }
2680 stmdb sp!, { r1 - r2 }
2681 bl update_texture_4bpp_cache
2684 ldmia sp!, { r1 - r2 }
2690 function(texture_blocks_8bpp)
2691 stmdb sp!, { r3 - r11, r14 }
2692 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2694 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2695 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2697 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2698 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2700 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]
2701 tst dirty_textures_mask, current_texture_mask
2707 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2709 uxtah uv_0, texture_ptr, uv_01
2710 uxtah uv_1, texture_ptr, uv_01, ror #16
2712 uxtah uv_2, texture_ptr, uv_23
2713 uxtah uv_3, texture_ptr, uv_23, ror #16
2715 uxtah uv_4, texture_ptr, uv_45
2716 ldrb pixel_0, [ uv_0 ]
2718 uxtah uv_5, texture_ptr, uv_45, ror #16
2719 ldrb pixel_1, [ uv_1 ]
2721 uxtah uv_6, texture_ptr, uv_67
2722 ldrb pixel_2, [ uv_2 ]
2724 uxtah uv_7, texture_ptr, uv_67, ror #16
2725 ldrb pixel_3, [ uv_3 ]
2727 ldrb pixel_4, [ uv_4 ]
2728 add pixel_0, pixel_0, pixel_0
2730 ldrb pixel_5, [ uv_5 ]
2731 add pixel_1, pixel_1, pixel_1
2733 ldrb pixel_6, [ uv_6 ]
2734 add pixel_2, pixel_2, pixel_2
2736 ldrb pixel_7, [ uv_7 ]
2737 add pixel_3, pixel_3, pixel_3
2739 ldrh pixel_0, [ clut_ptr, pixel_0 ]
2740 add pixel_4, pixel_4, pixel_4
2742 ldrh pixel_1, [ clut_ptr, pixel_1 ]
2743 add pixel_5, pixel_5, pixel_5
2745 ldrh pixel_2, [ clut_ptr, pixel_2 ]
2746 add pixel_6, pixel_6, pixel_6
2748 ldrh pixel_3, [ clut_ptr, pixel_3 ]
2749 add pixel_7, pixel_7, pixel_7
2751 ldrh pixel_4, [ clut_ptr, pixel_4 ]
2752 orr pixels_a, pixel_0, pixel_1, lsl #16
2754 ldrh pixel_5, [ clut_ptr, pixel_5 ]
2755 orr pixels_c, pixel_2, pixel_3, lsl #16
2757 ldrh pixel_6, [ clut_ptr, pixel_6 ]
2758 subs num_blocks, num_blocks, #1
2760 ldrh pixel_7, [ clut_ptr, pixel_7 ]
2761 orr pixels_b, pixel_4, pixel_5, lsl #16
2763 orr pixels_d, pixel_6, pixel_7, lsl #16
2764 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2766 add block_ptr, block_ptr, #64
2769 ldmia sp!, { r3 - r11, pc }
2772 stmdb sp!, { r1 - r2, r12 }
2774 bl update_texture_8bpp_cache
2776 ldmia sp!, { r1 - r2, r12 }
2806 #define block_ptr r1
2807 #define num_blocks r2
2851 #define texture_ptr r12
2856 function(texture_blocks_16bpp)
2857 stmdb sp!, { r3 - r11, r14 }
2858 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2860 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2861 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2864 ldrh uv_0, [ block_ptr ]
2865 subs num_blocks, num_blocks, #1
2867 ldrh uv_1, [ block_ptr, #2 ]
2869 and v_0, uv_0, #0xFF00
2870 and v_1, uv_1, #0xFF00
2872 and u_0, uv_0, #0xFF
2873 and u_1, uv_1, #0xFF
2875 add uv_0, u_0, v_0, lsl #2
2876 ldrh uv_2, [ block_ptr, #4 ]
2878 add uv_1, u_1, v_1, lsl #2
2879 ldrh uv_3, [ block_ptr, #6 ]
2881 add uv_0, uv_0, uv_0
2882 add uv_1, uv_1, uv_1
2884 and v_2, uv_2, #0xFF00
2885 and v_3, uv_3, #0xFF00
2887 and u_2, uv_2, #0xFF
2888 and u_3, uv_3, #0xFF
2890 add uv_2, u_2, v_2, lsl #2
2891 ldrh uv_4, [ block_ptr, #8 ]
2893 add uv_3, u_3, v_3, lsl #2
2894 ldrh uv_5, [ block_ptr, #10 ]
2896 add uv_2, uv_2, uv_2
2897 add uv_3, uv_3, uv_3
2899 and v_4, uv_4, #0xFF00
2900 and v_5, uv_5, #0xFF00
2902 and u_4, uv_4, #0xFF
2903 and u_5, uv_5, #0xFF
2905 add uv_4, u_4, v_4, lsl #2
2906 ldrh uv_6, [ block_ptr, #12 ]
2908 add uv_5, u_5, v_5, lsl #2
2909 ldrh uv_7, [ block_ptr, #14 ]
2911 add uv_4, uv_4, uv_4
2912 ldrh pixel_0, [ texture_ptr, uv_0 ]
2914 add uv_5, uv_5, uv_5
2915 ldrh pixel_1, [ texture_ptr, uv_1 ]
2917 and v_6, uv_6, #0xFF00
2918 ldrh pixel_2, [ texture_ptr, uv_2 ]
2920 and v_7, uv_7, #0xFF00
2921 ldrh pixel_3, [ texture_ptr, uv_3 ]
2923 and u_6, uv_6, #0xFF
2924 ldrh pixel_4, [ texture_ptr, uv_4 ]
2926 and u_7, uv_7, #0xFF
2927 ldrh pixel_5, [ texture_ptr, uv_5 ]
2929 add uv_6, u_6, v_6, lsl #2
2930 add uv_7, u_7, v_7, lsl #2
2932 add uv_6, uv_6, uv_6
2933 add uv_7, uv_7, uv_7
2935 orr pixels_a, pixel_0, pixel_1, lsl #16
2936 orr pixels_b, pixel_2, pixel_3, lsl #16
2938 ldrh pixel_6, [ texture_ptr, uv_6 ]
2939 orr pixels_c, pixel_4, pixel_5, lsl #16
2941 ldrh pixel_7, [ texture_ptr, uv_7 ]
2942 orr pixels_d, pixel_6, pixel_7, lsl #16
2944 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
2945 add block_ptr, block_ptr, #64
2949 ldmia sp!, { r3 - r11, pc }
2964 #undef msb_mask_high
2972 #define num_blocks r1
2973 #define color_ptr r2
2974 #define colors_scalar r2
2975 #define colors_scalar_compare r3
2976 #define mask_msb_ptr r2
2978 #define block_ptr_load_a r0
2979 #define block_ptr_store r3
2980 #define block_ptr_load_b r12
2985 #define draw_mask_bits_scalar r5
2987 #define d128_0x07 q0
2988 #define d128_0x1F q1
2989 #define d128_0x8000 q2
2990 #define test_mask q3
2992 #define colors_rg q5
2993 #define colors_b_dm_bits q6
2994 #define texels_rg q7
2997 #define pixels_b q10
2999 #define zero_mask q4
3000 #define draw_mask q12
3001 #define msb_mask q13
3003 #define fb_pixels q8
3005 #define pixels_gb_low q9
3007 #define colors_r d10
3008 #define colors_g d11
3009 #define colors_b d12
3010 #define draw_mask_bits d13
3011 #define texels_r d14
3012 #define texels_g d15
3013 #define pixels_r_low d16
3014 #define pixels_g_low d18
3015 #define pixels_b_low d19
3016 #define msb_mask_low d26
3017 #define msb_mask_high d27
3022 #define texels_b d31
3024 #define shade_blocks_textured_modulated_prologue_indirect() \
3026 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3028 #define shade_blocks_textured_modulated_prologue_direct() \
3029 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3030 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ] \
3033 #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \
3035 #define shade_blocks_textured_false_modulation_check_undithered(target) \
3036 ldr colors_scalar, [ psx_gpu, #psx_gpu_triangle_color_offset ]; \
3037 movw colors_scalar_compare, #0x8080; \
3039 movt colors_scalar_compare, #0x80; \
3040 cmp colors_scalar, colors_scalar_compare; \
3041 beq shade_blocks_textured_unmodulated_##target \
3043 #define shade_blocks_textured_false_modulation_check_dithered(target) \
3045 #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \
3046 shade_blocks_textured_false_modulation_check_##dithering(target); \
3047 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3048 vld1.u32 { colors_r[] }, [ color_ptr, :32 ]; \
3049 vdup.u8 colors_g, colors_r[1]; \
3050 vdup.u8 colors_b, colors_r[2]; \
3051 vdup.u8 colors_r, colors_r[0] \
3054 #define shade_blocks_textured_modulated_load_dithered(target) \
3055 vld1.u32 { target }, [ block_ptr_load_b, :128 ] \
3057 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3058 vld1.u32 { target }, [ block_ptr_load_b, :128 ], c_32 \
3060 #define shade_blocks_textured_modulated_load_undithered(target) \
3062 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3063 add block_ptr_load_b, block_ptr_load_b, #32 \
3065 #define shade_blocks_textured_modulate_dithered(channel) \
3066 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3068 #define shade_blocks_textured_modulate_undithered(channel) \
3069 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3072 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3073 vst1.u32 { draw_mask }, [ block_ptr_store, :128 ]! \
3075 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3076 ldr fb_ptr, [ block_ptr_load_b, #(offset - 64) ]; \
3077 vld1.u32 { fb_pixels }, [ fb_ptr ]; \
3078 vbit.u16 pixels, fb_pixels, draw_mask \
3080 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3081 vst1.u32 { pixels }, [ block_ptr_store, :128 ], c_48 \
3083 #define shade_blocks_textured_modulated_store_pixels_direct() \
3084 vst1.u32 { pixels }, [ fb_ptr ] \
3087 #define shade_blocks_textured_modulated_load_rg_shaded() \
3088 vld1.u32 { colors_r, colors_g }, [ block_ptr_load_b, :128 ], c_32 \
3090 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3091 add block_ptr_load_b, block_ptr_load_b, #32 \
3093 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3094 vld1.u32 { colors_b, draw_mask_bits }, [ block_ptr_load_a, :128 ], c_32 \
3096 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3097 ldr draw_mask_bits_scalar, [ block_ptr_load_a, #8 ]; \
3098 add block_ptr_load_a, block_ptr_load_a, #32 \
3100 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3101 vdup.u16 draw_mask, draw_mask_bits[0] \
3103 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3104 vdup.u16 draw_mask, draw_mask_bits_scalar \
3107 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3109 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3110 vorr.u16 pixels, pixels, msb_mask \
3113 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3116 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3117 shade_blocks_textured_modulated_prologue_##shading(dithering, target); \
3118 stmdb sp!, { r4 - r5, lr }; \
3119 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3121 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
3123 shade_blocks_textured_modulated_prologue_##target(); \
3125 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3128 add block_ptr_load_b, block_ptr_load_a, #16; \
3129 vmov.u8 d64_1, #1; \
3130 vmov.u8 d64_4, #4; \
3131 vmov.u8 d64_128, #128; \
3133 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3134 vmov.u8 d128_0x07, #0x07; \
3136 shade_blocks_textured_modulated_load_rg_##shading(); \
3137 vmov.u8 d128_0x1F, #0x1F; \
3139 shade_blocks_textured_modulated_load_bdm_##shading(); \
3140 vmov.u16 d128_0x8000, #0x8000; \
3142 vmovn.u16 texels_r, texels; \
3143 vshrn.u16 texels_g, texels, #5; \
3145 vshrn.u16 texels_b, texels, #7; \
3146 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3148 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3149 vtst.u16 draw_mask, draw_mask, test_mask; \
3151 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3152 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3154 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3155 vshr.u8 texels_b, texels_b, #3; \
3157 shade_blocks_textured_modulate_##dithering(r); \
3158 shade_blocks_textured_modulate_##dithering(g); \
3159 shade_blocks_textured_modulate_##dithering(b); \
3161 vand.u16 pixels, texels, d128_0x8000; \
3162 vceq.u16 zero_mask, texels, #0; \
3164 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3165 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3166 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3168 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3169 vorr.u16 draw_mask, draw_mask, zero_mask; \
3170 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3171 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3173 subs num_blocks, num_blocks, #1; \
3179 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3180 shade_blocks_textured_modulated_load_rg_##shading(); \
3181 vshrn.u16 texels_g, texels, #5; \
3183 shade_blocks_textured_modulated_load_bdm_##shading(); \
3184 vshrn.u16 texels_b, texels, #7; \
3186 vmovn.u16 texels_r, texels; \
3187 vmlal.u8 pixels, pixels_r_low, d64_1; \
3189 vmlal.u8 pixels, pixels_g_low, d64_4; \
3190 vmlal.u8 pixels, pixels_b_low, d64_128; \
3191 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3193 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3194 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3196 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3197 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3199 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3200 vtst.u16 draw_mask, draw_mask, test_mask; \
3202 shade_blocks_textured_modulated_store_pixels_##target(); \
3203 vshr.u8 texels_b, texels_b, #3; \
3205 shade_blocks_textured_modulate_##dithering(r); \
3206 shade_blocks_textured_modulate_##dithering(g); \
3207 shade_blocks_textured_modulate_##dithering(b); \
3209 vand.u16 pixels, texels, d128_0x8000; \
3210 vceq.u16 zero_mask, texels, #0; \
3212 subs num_blocks, num_blocks, #1; \
3214 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3215 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3216 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3218 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3219 vorr.u16 draw_mask, draw_mask, zero_mask; \
3220 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3221 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3226 vmlal.u8 pixels, pixels_r_low, d64_1; \
3227 vmlal.u8 pixels, pixels_g_low, d64_4; \
3228 vmlal.u8 pixels, pixels_b_low, d64_128; \
3230 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3231 shade_blocks_textured_modulated_store_pixels_##target(); \
3233 ldmia sp!, { r4 - r5, pc } \
3236 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3237 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3238 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3239 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3241 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3242 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3243 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3244 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3262 #undef msb_mask_high
3265 #define num_blocks r1
3266 #define mask_msb_ptr r2
3267 #define color_ptr r3
3269 #define block_ptr_load r0
3270 #define draw_mask_store_ptr r3
3271 #define draw_mask_bits_ptr r12
3272 #define draw_mask_ptr r12
3273 #define pixel_store_ptr r14
3275 #define fb_ptr_cmp r4
3278 #define fb_ptr_next r14
3282 #define test_mask q0
3284 #define draw_mask q2
3285 #define zero_mask q3
3286 #define draw_mask_combined q4
3287 #define fb_pixels q5
3288 #define fb_pixels_next q6
3291 #define draw_mask_low d4
3292 #define draw_mask_high d5
3293 #define msb_mask_low d14
3294 #define msb_mask_high d15
3297 function(shade_blocks_textured_unmodulated_indirect)
3298 str r14, [ sp, #-4 ]
3299 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3301 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3302 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3304 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3305 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3308 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3310 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3311 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3312 [ draw_mask_bits_ptr, :16 ], c_64
3313 vceq.u16 zero_mask, pixels, #0
3315 vtst.u16 draw_mask, draw_mask, test_mask
3316 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3318 subs num_blocks, num_blocks, #1
3322 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3323 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3325 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3326 [ draw_mask_bits_ptr, :16 ], c_64
3327 vceq.u16 zero_mask, pixels, #0
3329 vtst.u16 draw_mask, draw_mask, test_mask
3330 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3332 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3333 subs num_blocks, num_blocks, #1
3338 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3339 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3346 function(shade_blocks_textured_unmodulated_direct)
3347 stmdb sp!, { r4, r14 }
3348 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3350 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3351 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3353 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3356 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3357 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3359 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3360 [ draw_mask_bits_ptr, :16 ], c_64
3361 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3363 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3364 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3365 vceq.u16 zero_mask, pixels, #0
3366 vtst.u16 draw_mask, draw_mask, test_mask
3368 subs num_blocks, num_blocks, #1
3372 mov fb_ptr, fb_ptr_next
3373 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3375 vorr.u16 pixels, pixels, msb_mask
3377 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3378 vmov fb_pixels, fb_pixels_next
3380 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3381 [ draw_mask_bits_ptr, :16 ], c_64
3382 vbif.u16 fb_pixels, pixels, draw_mask_combined
3384 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3386 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3387 add fb_ptr_cmp, fb_ptr_cmp, #14
3391 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3392 vceq.u16 zero_mask, pixels, #0
3394 vst1.u16 { fb_pixels }, [ fb_ptr ]
3395 vtst.u16 draw_mask, draw_mask, test_mask
3398 subs num_blocks, num_blocks, #1
3402 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3403 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3405 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3407 ldmia sp!, { r4, pc }
3410 vst1.u16 { fb_pixels }, [ fb_ptr ]
3411 vceq.u16 zero_mask, pixels, #0
3413 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3414 vtst.u16 draw_mask, draw_mask, test_mask
3419 function(shade_blocks_unshaded_untextured_indirect)
3424 function(shade_blocks_unshaded_untextured_direct)
3425 stmdb sp!, { r4, r14 }
3426 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3428 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3429 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3431 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3432 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3434 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3435 vld1.u16 { pixels }, [ color_ptr, :128 ]
3438 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3440 vorr.u16 pixels, pixels, msb_mask
3441 subs num_blocks, num_blocks, #1
3443 ldr fb_ptr_next, [ block_ptr_load ], #64
3445 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3449 vmov fb_pixels, fb_pixels_next
3450 mov fb_ptr, fb_ptr_next
3451 ldr fb_ptr_next, [ block_ptr_load ], #64
3453 vbif.u16 fb_pixels, pixels, draw_mask
3454 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3456 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3457 add fb_ptr_cmp, fb_ptr_cmp, #14
3461 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3462 vst1.u16 { fb_pixels }, [ fb_ptr ]
3465 subs num_blocks, num_blocks, #1
3469 vbif.u16 fb_pixels_next, pixels, draw_mask
3470 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3472 ldmia sp!, { r4, pc }
3475 vst1.u16 { fb_pixels }, [ fb_ptr ]
3476 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3480 #undef draw_mask_ptr
3487 #define num_blocks r1
3488 #define msb_mask_ptr r2
3489 #define pixel_ptr r3
3490 #define draw_mask_ptr r0
3493 #define fb_ptr_next r14
3494 #define fb_ptr_cmp r4
3502 #undef msb_mask_high
3503 #undef draw_mask_next
3506 #undef fb_pixels_next
3509 #define draw_mask q1
3511 #define fb_pixels q3
3512 #define blend_pixels q4
3513 #define pixels_no_msb q5
3514 #define blend_mask q6
3515 #define fb_pixels_no_msb q7
3516 #define d128_0x8000 q8
3517 #define d128_0x0421 q9
3518 #define fb_pixels_next q10
3519 #define blend_pixels_next q11
3520 #define pixels_next q12
3521 #define draw_mask_next q13
3522 #define write_mask q14
3524 #define pixels_rb q5
3525 #define pixels_mg q7
3527 #define d128_0x7C1F q8
3528 #define d128_0x03E0 q9
3529 #define fb_pixels_rb q10
3530 #define fb_pixels_g q11
3531 #define fb_pixels_masked q11
3532 #define d128_0x83E0 q15
3533 #define pixels_fourth q7
3534 #define d128_0x1C07 q12
3535 #define d128_0x00E0 q13
3536 #define d128_0x80E0 q13
3538 #define msb_mask_low d0
3539 #define msb_mask_high d1
3541 #define blend_blocks_average_set_blend_mask_textured(source) \
3542 vclt.s16 blend_mask, source, #0 \
3544 #define blend_blocks_average_set_stp_bit_textured() \
3545 vorr.u16 blend_pixels, #0x8000 \
3547 #define blend_blocks_average_combine_textured(source) \
3548 vbif.u16 blend_pixels, source, blend_mask \
3550 #define blend_blocks_average_set_blend_mask_untextured(source) \
3552 #define blend_blocks_average_set_stp_bit_untextured() \
3554 #define blend_blocks_average_combine_untextured(source) \
3556 #define blend_blocks_average_mask_set_on() \
3557 vclt.s16 write_mask, fb_pixels_next, #0 \
3559 #define blend_blocks_average_mask_copy_on() \
3560 vorr.u16 draw_mask, draw_mask_next, write_mask \
3562 #define blend_blocks_average_mask_copy_b_on() \
3563 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3565 #define blend_blocks_average_mask_set_off() \
3567 #define blend_blocks_average_mask_copy_off() \
3568 vmov draw_mask, draw_mask_next \
3570 #define blend_blocks_average_mask_copy_b_off() \
3572 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3575 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3576 stmdb sp!, { r4, r14 }; \
3577 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3578 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3580 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3581 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3583 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3586 vmov.u16 d128_0x8000, #0x8000; \
3587 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3588 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3590 vmov.u16 d128_0x0421, #0x0400; \
3591 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3593 vorr.u16 d128_0x0421, #0x0021; \
3594 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3596 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3597 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3598 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3599 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3600 blend_blocks_average_mask_set_##mask_evaluate(); \
3601 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3603 subs num_blocks, num_blocks, #1; \
3607 mov fb_ptr, fb_ptr_next; \
3608 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3610 vmov pixels, pixels_next; \
3611 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3613 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3615 blend_blocks_average_mask_copy_##mask_evaluate(); \
3616 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3618 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3619 blend_blocks_average_set_stp_bit_##texturing(); \
3620 vmov fb_pixels, fb_pixels_next; \
3621 blend_blocks_average_combine_##texturing(pixels); \
3623 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3624 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3625 cmp fb_ptr_cmp, #28; \
3628 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3629 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3631 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3632 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3634 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3635 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3637 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3638 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3639 blend_blocks_average_mask_set_##mask_evaluate(); \
3640 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3643 subs num_blocks, num_blocks, #1; \
3647 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3648 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3650 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3651 blend_blocks_average_set_stp_bit_##texturing(); \
3652 blend_blocks_average_combine_##texturing(pixels_next); \
3654 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3655 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3656 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3658 ldmia sp!, { r4, pc }; \
3661 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3662 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3663 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3665 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3666 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3667 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3668 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3669 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3670 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3674 blend_blocks_average_builder(textured, off)
3675 blend_blocks_average_builder(untextured, off)
3676 blend_blocks_average_builder(textured, on)
3677 blend_blocks_average_builder(untextured, on)
3680 #define blend_blocks_add_mask_set_on() \
3681 vclt.s16 write_mask, fb_pixels, #0 \
3683 #define blend_blocks_add_mask_copy_on() \
3684 vorr.u16 draw_mask, draw_mask, write_mask \
3686 #define blend_blocks_add_mask_set_off() \
3688 #define blend_blocks_add_mask_copy_off() \
3691 #define blend_blocks_add_textured_builder(mask_evaluate) \
3694 function(blend_blocks_textured_add_##mask_evaluate) \
3695 stmdb sp!, { r4, r14 }; \
3696 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3697 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3699 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3700 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3702 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3705 vmov.u16 d128_0x7C1F, #0x7C00; \
3706 vmov.u16 d128_0x03E0, #0x0300; \
3707 vmov.u16 d128_0x83E0, #0x8000; \
3708 vorr.u16 d128_0x03E0, #0x00E0; \
3709 vorr.u16 d128_0x7C1F, #0x001F; \
3710 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3712 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3713 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3714 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3715 vclt.s16 blend_mask, pixels, #0; \
3716 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3717 blend_blocks_add_mask_set_##mask_evaluate(); \
3718 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3720 blend_blocks_add_mask_copy_##mask_evaluate(); \
3721 vorr.u16 pixels, pixels, msb_mask; \
3722 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3723 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3724 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3725 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3726 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3727 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3728 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3729 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3731 subs num_blocks, num_blocks, #1; \
3735 mov fb_ptr, fb_ptr_next; \
3737 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3739 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3740 vclt.s16 blend_mask, pixels, #0; \
3742 vorr.u16 pixels, pixels, msb_mask; \
3743 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3744 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3746 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3747 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3749 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3750 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3751 cmp fb_ptr_cmp, #28; \
3754 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3755 blend_blocks_add_mask_set_##mask_evaluate(); \
3756 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3757 blend_blocks_add_mask_copy_##mask_evaluate(); \
3758 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3759 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3760 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3763 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3764 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3765 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3766 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3767 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3769 subs num_blocks, num_blocks, #1; \
3773 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3774 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3775 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3777 ldmia sp!, { r4, pc }; \
3780 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3781 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3783 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3784 blend_blocks_add_mask_set_##mask_evaluate(); \
3785 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3786 blend_blocks_add_mask_copy_##mask_evaluate(); \
3787 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3791 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3794 function(blend_blocks_untextured_add_##mask_evaluate) \
3795 stmdb sp!, { r4, r14 }; \
3796 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3797 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3799 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3800 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3802 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3805 vmov.u16 d128_0x7C1F, #0x7C00; \
3806 vmov.u16 d128_0x03E0, #0x0300; \
3807 vorr.u16 d128_0x7C1F, #0x001F; \
3808 vorr.u16 d128_0x03E0, #0x00E0; \
3810 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3811 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3812 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3813 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3814 blend_blocks_add_mask_set_##mask_evaluate(); \
3815 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3817 blend_blocks_add_mask_copy_##mask_evaluate(); \
3818 vand.u16 pixels_g, pixels, d128_0x03E0; \
3819 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3820 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3821 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3822 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3823 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3824 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3826 subs num_blocks, num_blocks, #1; \
3830 mov fb_ptr, fb_ptr_next; \
3832 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3834 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3836 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3837 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3838 vand.u16 pixels_g, pixels, d128_0x03E0; \
3840 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3841 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3843 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3844 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3845 cmp fb_ptr_cmp, #28; \
3848 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3849 blend_blocks_add_mask_set_##mask_evaluate(); \
3850 blend_blocks_add_mask_copy_##mask_evaluate(); \
3851 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3852 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3853 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3856 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3857 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3858 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3859 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3860 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3862 subs num_blocks, num_blocks, #1; \
3866 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3867 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3868 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3869 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3871 ldmia sp!, { r4, pc }; \
3874 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3875 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3877 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3878 blend_blocks_add_mask_set_##mask_evaluate(); \
3879 blend_blocks_add_mask_copy_##mask_evaluate(); \
3880 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3884 blend_blocks_add_textured_builder(off)
3885 blend_blocks_add_textured_builder(on)
3886 blend_blocks_add_untextured_builder(off)
3887 blend_blocks_add_untextured_builder(on)
3889 #define blend_blocks_subtract_set_blend_mask_textured() \
3890 vclt.s16 blend_mask, pixels_next, #0 \
3892 #define blend_blocks_subtract_combine_textured() \
3893 vbif.u16 blend_pixels, pixels, blend_mask \
3895 #define blend_blocks_subtract_set_stb_textured() \
3896 vorr.u16 blend_pixels, #0x8000 \
3898 #define blend_blocks_subtract_msb_mask_textured() \
3899 vorr.u16 pixels, pixels_next, msb_mask \
3901 #define blend_blocks_subtract_set_blend_mask_untextured() \
3903 #define blend_blocks_subtract_combine_untextured() \
3905 #define blend_blocks_subtract_set_stb_untextured() \
3906 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3908 #define blend_blocks_subtract_msb_mask_untextured() \
3911 #define blend_blocks_subtract_mask_set_on() \
3912 vclt.s16 write_mask, fb_pixels, #0 \
3914 #define blend_blocks_subtract_mask_copy_on() \
3915 vorr.u16 draw_mask, draw_mask_next, write_mask \
3917 #define blend_blocks_subtract_mask_set_off() \
3919 #define blend_blocks_subtract_mask_copy_off() \
3920 vmov draw_mask, draw_mask_next \
3923 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
3926 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
3927 stmdb sp!, { r4, r14 }; \
3928 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3929 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3931 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3932 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3934 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3937 vmov.u16 d128_0x7C1F, #0x7C00; \
3938 vmov.u16 d128_0x03E0, #0x0300; \
3939 vorr.u16 d128_0x7C1F, #0x001F; \
3940 vorr.u16 d128_0x03E0, #0x00E0; \
3942 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3943 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3944 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3945 blend_blocks_subtract_set_blend_mask_##texturing(); \
3946 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3947 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3948 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3950 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3951 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3952 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3953 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3954 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3956 subs num_blocks, num_blocks, #1; \
3960 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
3961 mov fb_ptr, fb_ptr_next; \
3962 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3964 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3965 blend_blocks_subtract_msb_mask_##texturing(); \
3967 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3968 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3969 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3970 blend_blocks_subtract_set_stb_##texturing(); \
3971 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3972 blend_blocks_subtract_combine_##texturing(); \
3973 blend_blocks_subtract_set_blend_mask_##texturing(); \
3974 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3976 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3977 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3978 cmp fb_ptr_cmp, #28; \
3981 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3982 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3983 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3984 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3985 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3986 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3987 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3990 subs num_blocks, num_blocks, #1; \
3994 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
3996 blend_blocks_subtract_msb_mask_##texturing(); \
3997 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3998 blend_blocks_subtract_set_stb_##texturing(); \
3999 blend_blocks_subtract_combine_##texturing(); \
4000 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4001 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4003 ldmia sp!, { r4, pc }; \
4006 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4007 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4008 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4009 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4010 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4011 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4012 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4016 blend_blocks_subtract_builder(textured, off)
4017 blend_blocks_subtract_builder(textured, on)
4018 blend_blocks_subtract_builder(untextured, off)
4019 blend_blocks_subtract_builder(untextured, on)
4022 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4025 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4026 stmdb sp!, { r4, r14 }; \
4027 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4028 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4030 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4031 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4033 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4036 vmov.u16 d128_0x7C1F, #0x7C00; \
4037 vmov.u16 d128_0x03E0, #0x0300; \
4038 vmov.u16 d128_0x1C07, #0x1C00; \
4039 vmov.u16 d128_0x00E0, #0x00E0; \
4040 vorr.u16 d128_0x7C1F, #0x001F; \
4041 vorr.u16 d128_0x03E0, #0x00E0; \
4042 vorr.u16 d128_0x1C07, #0x0007; \
4044 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4045 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4046 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4047 vclt.s16 blend_mask, pixels, #0; \
4048 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4049 blend_blocks_add_mask_set_##mask_evaluate(); \
4050 vshr.s16 pixels_fourth, pixels, #2; \
4051 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4053 blend_blocks_add_mask_copy_##mask_evaluate(); \
4054 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4055 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4056 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4057 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4058 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4059 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4060 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4062 subs num_blocks, num_blocks, #1; \
4066 mov fb_ptr, fb_ptr_next; \
4067 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4069 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4070 vbif.u16 blend_pixels, pixels, blend_mask; \
4072 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4073 vclt.s16 blend_mask, pixels, #0; \
4074 vshr.s16 pixels_fourth, pixels, #2; \
4075 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4076 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4078 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4079 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4081 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4082 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4083 cmp fb_ptr_cmp, #28; \
4086 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4087 blend_blocks_add_mask_set_##mask_evaluate(); \
4088 blend_blocks_add_mask_copy_##mask_evaluate(); \
4089 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4090 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4091 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4094 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4095 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4096 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4097 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4098 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4100 subs num_blocks, num_blocks, #1; \
4104 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4105 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4106 vbif.u16 blend_pixels, pixels, blend_mask; \
4107 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4108 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4110 ldmia sp!, { r4, pc }; \
4113 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4114 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4116 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4117 blend_blocks_add_mask_set_##mask_evaluate(); \
4118 blend_blocks_add_mask_copy_##mask_evaluate(); \
4119 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4124 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4127 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4128 stmdb sp!, { r4, r14 }; \
4129 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4130 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4132 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4133 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4135 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4138 vmov.u16 d128_0x7C1F, #0x7C00; \
4139 vmov.u16 d128_0x03E0, #0x0300; \
4140 vmov.u16 d128_0x1C07, #0x1C00; \
4141 vmov.u16 d128_0x00E0, #0x00E0; \
4142 vorr.u16 d128_0x7C1F, #0x001F; \
4143 vorr.u16 d128_0x03E0, #0x00E0; \
4144 vorr.u16 d128_0x1C07, #0x0007; \
4146 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4147 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4148 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4149 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4150 blend_blocks_add_mask_set_##mask_evaluate(); \
4151 vshr.s16 pixels_fourth, pixels, #2; \
4152 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4154 blend_blocks_add_mask_copy_##mask_evaluate(); \
4155 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4156 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4157 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4158 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4159 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4160 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4161 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4163 subs num_blocks, num_blocks, #1; \
4167 mov fb_ptr, fb_ptr_next; \
4168 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4170 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4172 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4173 vshr.s16 pixels_fourth, pixels, #2; \
4174 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4175 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4177 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4178 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4180 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4181 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4182 cmp fb_ptr_cmp, #28; \
4185 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4186 blend_blocks_add_mask_set_##mask_evaluate(); \
4187 blend_blocks_add_mask_copy_##mask_evaluate(); \
4188 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4189 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4190 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4193 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4194 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4195 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4196 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4197 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4199 subs num_blocks, num_blocks, #1; \
4203 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4204 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4205 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4206 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4208 ldmia sp!, { r4, pc }; \
4211 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4212 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4214 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4215 blend_blocks_add_mask_set_##mask_evaluate(); \
4216 blend_blocks_add_mask_copy_##mask_evaluate(); \
4217 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4221 blend_blocks_add_fourth_textured_builder(off)
4222 blend_blocks_add_fourth_textured_builder(on)
4223 blend_blocks_add_fourth_untextured_builder(off)
4224 blend_blocks_add_fourth_untextured_builder(on)
4226 // TODO: Optimize this more. Need a scene that actually uses it for
4231 function(blend_blocks_textured_unblended_on)
4232 stmdb sp!, { r4, r14 }
4233 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4234 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
4236 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4237 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
4239 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4242 ldr fb_ptr, [ pixel_ptr, #28 ]
4243 vld1.u16 { fb_pixels }, [ fb_ptr ]
4244 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4245 vclt.s16 write_mask, fb_pixels, #0
4246 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4248 subs num_blocks, num_blocks, #1
4252 vorr.u16 draw_mask, draw_mask, write_mask
4253 vbif.u16 fb_pixels, pixels, draw_mask
4254 vst1.u16 { fb_pixels }, [ fb_ptr ]
4256 ldr fb_ptr, [ pixel_ptr, #28 ]
4257 vld1.u16 { fb_pixels }, [ fb_ptr ]
4258 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4259 vclt.s16 write_mask, fb_pixels, #0
4260 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4262 subs num_blocks, num_blocks, #1
4266 vorr.u16 draw_mask, draw_mask, write_mask
4267 vbif.u16 fb_pixels, pixels, draw_mask
4268 vst1.u16 { fb_pixels }, [ fb_ptr ]
4270 ldmia sp!, { r4, pc }
4273 function(blend_blocks_textured_unblended_off)
4283 vld1.u32 { u_whole_8, v_whole_8 }, [ r1, :128 ], r3
4303 #define num_width r12
4313 function(render_block_fill_body)
4314 vdup.u16 colors_a, color
4317 vmov colors_b, colors_a
4318 sub pitch, pitch, width, lsl #1
4320 mov num_width, width
4323 vst1.u32 { colors_a, colors_b }, [ vram_ptr, :256 ]!
4325 subs num_width, num_width, #16
4328 add vram_ptr, vram_ptr, pitch
4329 mov num_width, width
4331 subs height, height, #1
4345 #undef dirty_textures_mask
4347 #undef current_texture_mask
4358 #define offset_u_right r10
4359 #define width_rounded r11
4360 #define height_rounded r12
4362 #define texture_offset_base r1
4363 #define tile_width r2
4364 #define tile_height r3
4365 #define num_blocks r4
4367 #define sub_tile_height r6
4369 #define texture_mask r8
4370 #define column_data r9
4371 #define texture_offset r10
4372 #define tiles_remaining r11
4373 #define fb_ptr_advance_column r12
4374 #define texture_block_ptr r14
4376 #define texture_page_ptr r3
4377 #define left_block_mask r4
4378 #define right_block_mask r5
4379 #define texture_mask_rev r10
4380 #define control_mask r11
4382 #define dirty_textures_mask r4
4384 #define current_texture_mask r6
4398 #define draw_masks_fb_ptrs q1
4400 #define draw_mask_fb_ptr_left d2
4401 #define draw_mask_fb_ptr_right d3
4403 #define clut_low_a d4
4404 #define clut_low_b d5
4405 #define clut_high_a d6
4406 #define clut_high_b d7
4408 #define block_masks d8
4409 #define block_masks_shifted d9
4414 #define texels_low d10
4415 #define texels_high d11
4418 setup_sprite_flush_blocks_single:
4421 stmdb sp!, { r0 - r3, r12, r14 }
4422 bl flush_render_block_buffer
4423 ldmia sp!, { r0 - r3, r12, r14 }
4427 add block, psx_gpu, #psx_gpu_blocks_offset
4429 mov num_blocks, sub_tile_height
4433 setup_sprite_flush_blocks_double:
4436 stmdb sp!, { r0 - r3, r12, r14 }
4437 bl flush_render_block_buffer
4438 ldmia sp!, { r0 - r3, r12, r14 }
4442 add block, psx_gpu, #psx_gpu_blocks_offset
4444 mov num_blocks, sub_tile_height, lsl #1
4448 setup_sprite_update_texture_4bpp_cache:
4449 stmdb sp!, { r0 - r3, r14 }
4450 bl update_texture_4bpp_cache
4451 ldmia sp!, { r0 - r3, pc }
4454 setup_sprite_update_texture_8bpp_cache:
4455 stmdb sp!, { r0 - r3, r14 }
4456 bl update_texture_8bpp_cache
4457 ldmia sp!, { r0 - r3, pc }
4460 #define setup_sprite_tiled_initialize_4bpp() \
4461 ldr dirty_textures_mask, \
4462 [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]; \
4463 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4465 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4466 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4468 tst current_texture_mask, dirty_textures_mask; \
4469 vuzp.u8 clut_a, clut_b; \
4471 blne setup_sprite_update_texture_4bpp_cache \
4473 #define setup_sprite_tiled_initialize_8bpp() \
4474 ldr dirty_textures_mask, \
4475 [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]; \
4476 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4478 tst current_texture_mask, dirty_textures_mask; \
4479 blne setup_sprite_update_texture_8bpp_cache \
4482 #define setup_sprite_tile_setup_block_no(side, offset, texture_mode) \
4484 #define setup_sprite_block_count_single() \
4487 #define setup_sprite_block_count_double() \
4488 sub_tile_height, lsl #1 \
4490 #define setup_sprite_tile_add_blocks(type) \
4491 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4492 cmp num_blocks, #MAX_BLOCKS; \
4494 blgt setup_sprite_flush_blocks_##type \
4497 #define setup_sprite_tile_full_4bpp(edge) \
4498 setup_sprite_tile_add_blocks(double); \
4501 and texture_block_ptr, texture_offset, texture_mask; \
4502 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4505 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4506 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4508 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4509 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4511 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4512 add texture_block_ptr, texture_offset, #8; \
4514 and texture_block_ptr, texture_block_ptr, texture_mask; \
4515 add block, block, #40; \
4517 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4518 add fb_ptr, fb_ptr, #16; \
4520 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4521 add block, block, #24; \
4523 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4524 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4527 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4528 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4530 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4531 add block, block, #40; \
4533 add texture_offset, texture_offset, #0x10; \
4534 add fb_ptr, fb_ptr, #(2048 - 16); \
4536 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4537 add block, block, #24; \
4539 subs sub_tile_height, sub_tile_height, #1; \
4542 add texture_offset, texture_offset, #0xF00; \
4543 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4546 #define setup_sprite_tile_half_4bpp(edge) \
4547 setup_sprite_tile_add_blocks(single); \
4550 and texture_block_ptr, texture_offset, texture_mask; \
4551 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4554 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4555 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4557 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4558 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4560 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4561 add block, block, #40; \
4563 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4564 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4566 add block, block, #24; \
4567 add texture_offset, texture_offset, #0x10; \
4569 add fb_ptr, fb_ptr, #2048; \
4570 subs sub_tile_height, sub_tile_height, #1; \
4574 add texture_offset, texture_offset, #0xF00; \
4575 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4578 #define setup_sprite_tile_full_8bpp(edge) \
4579 setup_sprite_tile_add_blocks(double); \
4580 add block, block, #16; \
4583 and texture_block_ptr, texture_offset, texture_mask; \
4584 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4587 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4588 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4590 add texture_block_ptr, texture_offset, #8; \
4591 vst1.u32 { texels }, [ block, :64 ]; \
4593 and texture_block_ptr, texture_block_ptr, texture_mask; \
4594 add block, block, #24; \
4596 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4598 add fb_ptr, fb_ptr, #16; \
4599 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4601 add block, block, #40; \
4602 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4605 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4606 vst1.u32 { texels }, [ block, :64 ]; \
4607 add block, block, #24; \
4609 add texture_offset, texture_offset, #0x10; \
4610 add fb_ptr, fb_ptr, #(2048 - 16); \
4612 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4613 add block, block, #40; \
4615 subs sub_tile_height, sub_tile_height, #1; \
4618 sub block, block, #16; \
4619 add texture_offset, texture_offset, #0xF00; \
4620 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4623 #define setup_sprite_tile_half_8bpp(edge) \
4624 setup_sprite_tile_add_blocks(single); \
4625 add block, block, #16; \
4628 and texture_block_ptr, texture_offset, texture_mask; \
4629 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4632 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4633 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4635 vst1.u32 { texels }, [ block, :64 ]; \
4636 add block, block, #24; \
4638 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4639 add block, block, #40; \
4641 add texture_offset, texture_offset, #0x10; \
4642 add fb_ptr, fb_ptr, #2048; \
4644 subs sub_tile_height, sub_tile_height, #1; \
4647 sub block, block, #16; \
4648 add texture_offset, texture_offset, #0xF00; \
4649 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4652 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4653 add texture_offset, texture_offset_base, #8; \
4654 add fb_ptr, fb_ptr, #16 \
4656 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4657 mov texture_offset, texture_offset_base \
4659 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4660 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4662 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4663 mov texture_offset, texture_offset_base \
4665 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4666 sub fb_ptr, fb_ptr, #16 \
4668 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4670 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4671 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4673 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4676 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode) \
4677 mov sub_tile_height, column_data; \
4678 setup_sprite_tile_column_edge_pre_adjust_##edge_mode(edge); \
4679 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4680 setup_sprite_tile_column_edge_post_adjust_##edge_mode(edge) \
4682 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode) \
4683 and sub_tile_height, column_data, #0xFF; \
4684 mov tiles_remaining, column_data, lsr #16; \
4685 setup_sprite_tile_column_edge_pre_adjust_##edge_mode(edge); \
4686 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4688 subs tiles_remaining, tiles_remaining, #1; \
4692 mov sub_tile_height, #16; \
4693 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4694 subs tiles_remaining, tiles_remaining, #1; \
4698 uxtb sub_tile_height, column_data, ror #8; \
4699 setup_sprite_tile_##edge_mode##_##texture_mode(edge); \
4700 setup_sprite_tile_column_edge_post_adjust_##edge_mode(edge) \
4703 #define setup_sprite_column_data_single() \
4704 mov column_data, height; \
4705 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] \
4707 #define setup_sprite_column_data_multi() \
4708 and height_rounded, height_rounded, #0xF; \
4709 rsb column_data, offset_v, #16; \
4711 add height_rounded, height_rounded, #1; \
4712 sub tile_height, tile_height, #1; \
4714 orr column_data, column_data, tile_height, lsl #16; \
4715 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]; \
4717 orr column_data, column_data, height_rounded, lsl #8 \
4719 #define setup_sprite_tile_column_width_single(texture_mode, multi_height, \
4721 setup_sprite_##texture_mode##_single_##multi_height##_##edge_mode##_##edge: \
4722 setup_sprite_column_data_##multi_height(); \
4723 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4724 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4725 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4726 vdup.u8 draw_mask_fb_ptr_right, block_masks[1]; \
4728 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, \
4730 ldmia sp!, { r4 - r11, pc } \
4732 #define setup_sprite_tiled_advance_column() \
4733 add texture_offset_base, texture_offset_base, #0x100; \
4734 tst texture_offset_base, #0xF00; \
4735 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4737 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4739 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode: \
4740 setup_sprite_column_data_##multi_height(); \
4741 mov fb_ptr_advance_column, #32; \
4743 sub fb_ptr_advance_column, height, lsl #11; \
4744 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4746 vdup.u8 draw_mask_fb_ptr_right, block_masks[1]; \
4747 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm); \
4749 subs tile_width, tile_width, #2; \
4750 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4752 vmov.u8 draw_masks_fb_ptrs, #0; \
4756 setup_sprite_tiled_advance_column(); \
4757 setup_sprite_tile_column_height_##multi_height(full, none, tm); \
4758 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4759 subs tile_width, tile_width, #1; \
4763 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4764 vdup.u8 draw_mask_fb_ptr_right, block_masks[5]; \
4766 setup_sprite_tiled_advance_column(); \
4767 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm); \
4768 ldmia sp!, { r4 - r11, pc } \
4776 // [ sp + 4 ]: width
4777 // [ sp + 8 ]: height
4778 // [ sp + 12 ]: color (unused)
4780 #define setup_sprite_tiled_builder(texture_mode) \
4782 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full); \
4783 setup_sprite_tile_column_width_single(texture_mode, multi, full, none); \
4784 setup_sprite_tile_column_width_multi(texture_mode, single, full, full); \
4785 setup_sprite_tile_column_width_single(texture_mode, single, full, none); \
4786 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full); \
4787 setup_sprite_tile_column_width_single(texture_mode, multi, half, right); \
4788 setup_sprite_tile_column_width_multi(texture_mode, single, half, full); \
4789 setup_sprite_tile_column_width_single(texture_mode, single, half, right); \
4790 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half); \
4791 setup_sprite_tile_column_width_single(texture_mode, multi, half, left); \
4792 setup_sprite_tile_column_width_multi(texture_mode, single, full, half); \
4793 setup_sprite_tile_column_width_single(texture_mode, single, half, left); \
4794 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half); \
4795 setup_sprite_tile_column_width_multi(texture_mode, single, half, half); \
4799 function(setup_sprite_##texture_mode) \
4800 stmdb sp!, { r4 - r11, r14 }; \
4801 setup_sprite_tiled_initialize_##texture_mode(); \
4803 ldr v, [ sp, #36 ]; \
4804 and offset_u, u, #0xF; \
4806 ldr width, [ sp, #40 ]; \
4807 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
4809 ldr height, [ sp, #44 ]; \
4810 add fb_ptr, fb_ptr, y, lsl #11; \
4812 add fb_ptr, fb_ptr, x, lsl #1; \
4813 and offset_v, v, #0xF; \
4815 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4816 add width_rounded, offset_u, width; \
4818 add height_rounded, offset_v, height; \
4819 add width_rounded, width_rounded, #15; \
4821 add height_rounded, height_rounded, #15; \
4822 mov tile_width, width_rounded, lsr #4; \
4824 /* texture_offset_base = VH-VL-00-00 */\
4825 mov texture_offset_base, v, lsl #8; \
4826 and offset_u_right, width_rounded, #0xF; \
4828 /* texture_offset_base = VH-UH-UL-00 */\
4829 bfi texture_offset_base, u, #4, #8; \
4830 movw right_block_mask, #0xFFFE; \
4832 /* texture_offset_base = VH-UH-VL-00 */\
4833 bfi texture_offset_base, v, #4, #4; \
4834 movw left_block_mask, #0xFFFF; \
4836 mov tile_height, height_rounded, lsr #4; \
4837 mvn left_block_mask, left_block_mask, lsl offset_u; \
4839 /* texture_mask = HH-HL-WH-WL */\
4840 ldrh texture_mask, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]; \
4841 mov right_block_mask, right_block_mask, lsl offset_u_right; \
4843 /* texture_mask_rev = WH-WL-HH-HL */\
4844 rev16 texture_mask_rev, texture_mask; \
4845 vmov block_masks, left_block_mask, right_block_mask; \
4847 /* texture_mask = HH-HL-HL-WL */\
4848 bfi texture_mask, texture_mask_rev, #4, #4; \
4849 /* texture_mask_rev = 00-00-00-WH */\
4850 mov texture_mask_rev, texture_mask_rev, lsr #12; \
4852 /* texture_mask = HH-WH-HL-WL */\
4853 bfi texture_mask, texture_mask_rev, #8, #4; \
4854 and left_block_mask, left_block_mask, #0xFF; \
4856 mov control_mask, #0; \
4857 cmp left_block_mask, #0xFF; \
4859 uxtb right_block_mask, right_block_mask, ror #8; \
4860 orreq control_mask, control_mask, #0x4; \
4862 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4863 cmp right_block_mask, #0xFF; \
4865 orreq control_mask, control_mask, #0x8; \
4866 cmp tile_width, #1; \
4868 add block, psx_gpu, #psx_gpu_blocks_offset; \
4869 orreq control_mask, control_mask, #0x1; \
4871 cmp tile_height, #1; \
4872 add block, block, num_blocks, lsl #6; \
4874 orreq control_mask, control_mask, #0x2; \
4875 ldr pc, [ pc, control_mask, lsl #2 ]; \
4878 .word setup_sprite_##texture_mode##_multi_multi_full_full; \
4879 .word setup_sprite_##texture_mode##_single_multi_full_none; \
4880 .word setup_sprite_##texture_mode##_multi_single_full_full; \
4881 .word setup_sprite_##texture_mode##_single_single_full_none; \
4882 .word setup_sprite_##texture_mode##_multi_multi_half_full; \
4883 .word setup_sprite_##texture_mode##_single_multi_half_right; \
4884 .word setup_sprite_##texture_mode##_multi_single_half_full; \
4885 .word setup_sprite_##texture_mode##_single_single_half_right; \
4886 .word setup_sprite_##texture_mode##_multi_multi_full_half; \
4887 .word setup_sprite_##texture_mode##_single_multi_half_left; \
4888 .word setup_sprite_##texture_mode##_multi_single_full_half; \
4889 .word setup_sprite_##texture_mode##_single_single_half_left; \
4890 .word setup_sprite_##texture_mode##_multi_multi_half_half; \
4892 .word setup_sprite_##texture_mode##_multi_single_half_half \
4895 setup_sprite_tiled_builder(4bpp);
4896 setup_sprite_tiled_builder(8bpp);
4904 #define block_ptr r0
4905 #define num_blocks r1
4907 #define texel_shift_mask r3
4908 #define block_pixels_a r4
4909 #define block_pixels_b r5
4918 #define texels_01 r6
4919 #define texels_23 r7
4920 #define texels_45 r8
4921 #define texels_67 r9
4923 function(texture_sprite_blocks_8bpp)
4924 stmdb sp!, { r4 - r11, r14 }
4925 movw texel_shift_mask, #(0xFF << 1)
4927 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
4928 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
4930 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
4931 ldr block_pixels_a, [ block_ptr, #16 ]
4934 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
4935 ldr block_pixels_b, [ block_ptr, #20 ]
4937 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
4938 ldrh texel_0, [ clut_ptr, texel_0 ]
4940 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
4941 ldrh texel_1, [ clut_ptr, texel_1 ]
4943 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
4944 ldr block_pixels_a, [ block_ptr, #(64 + 16) ]
4946 ldrh texel_2, [ clut_ptr, texel_2 ]
4947 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
4949 ldrh texel_3, [ clut_ptr, texel_3 ]
4950 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
4952 ldrh texel_4, [ clut_ptr, texel_4 ]
4953 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
4955 ldrh texel_5, [ clut_ptr, texel_5 ]
4956 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
4958 ldrh texel_6, [ clut_ptr, texel_6 ]
4959 orr texels_01, texel_0, texel_1, lsl #16
4961 ldrh texel_7, [ clut_ptr, texel_7 ]
4962 orr texels_23, texel_2, texel_3, lsl #16
4964 orr texels_45, texel_4, texel_5, lsl #16
4965 str texels_01, [ block_ptr, #0 ]
4967 orr texels_67, texel_6, texel_7, lsl #16
4968 str texels_23, [ block_ptr, #4 ]
4970 subs num_blocks, num_blocks, #1
4971 str texels_45, [ block_ptr, #8 ]
4973 str texels_67, [ block_ptr, #12 ]
4974 add block_ptr, block_ptr, #64
4978 ldmia sp!, { r4 - r11, pc }
4981 #undef width_rounded
4984 #undef texture_offset
4993 #define left_offset r8
4994 #define width_rounded r9
4995 #define right_width r10
4996 #define block_width r11
4998 #define texture_offset_base r1
4999 #define texture_mask r2
5000 #define texture_page_ptr r3
5001 #define num_blocks r4
5004 #define texture_offset r8
5005 #define blocks_remaining r9
5006 #define fb_ptr_pitch r12
5007 #define texture_block_ptr r14
5009 #define texture_mask_width r2
5010 #define texture_mask_height r3
5011 #define left_mask_bits r4
5012 #define right_mask_bits r5
5016 #undef block_masks_shifted
5019 #define block_masks d0
5020 #define block_masks_shifted d1
5021 #define draw_mask_fb_ptr d2
5025 setup_sprites_16bpp_flush_single:
5028 stmdb sp!, { r0 - r3, r12, r14 }
5029 bl flush_render_block_buffer
5030 ldmia sp!, { r0 - r3, r12, r14 }
5034 add block, psx_gpu, #psx_gpu_blocks_offset
5039 setup_sprites_16bpp_flush_row:
5042 stmdb sp!, { r0 - r3, r12, r14 }
5043 bl flush_render_block_buffer
5044 ldmia sp!, { r0 - r3, r12, r14 }
5048 add block, psx_gpu, #psx_gpu_blocks_offset
5049 mov num_blocks, block_width
5053 function(setup_sprite_16bpp)
5054 stmdb sp!, { r4 - r11, r14 }
5055 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5058 add fb_ptr, fb_ptr, y, lsl #11
5060 ldr width, [ sp, #40 ]
5061 add fb_ptr, fb_ptr, x, lsl #1
5063 ldr height, [ sp, #44 ]
5064 and left_offset, u, #0x7
5066 add texture_offset_base, u, u
5067 add width_rounded, width, #7
5069 add texture_offset_base, v, lsl #11
5070 mov left_mask_bits, #0xFF
5072 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5073 add width_rounded, width_rounded, left_offset
5075 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5076 sub fb_ptr, fb_ptr, left_offset, lsl #1
5078 add texture_mask, texture_mask_width, texture_mask_width
5079 mov right_mask_bits, #0xFE
5081 and right_width, width_rounded, #0x7
5082 mvn left_mask_bits, left_mask_bits, lsl left_offset
5084 add texture_mask, texture_mask_height, lsl #11
5085 mov block_width, width_rounded, lsr #3
5087 mov right_mask_bits, right_mask_bits, lsl right_width
5088 movw fb_ptr_pitch, #(2048 + 16)
5090 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5091 vmov block_masks, left_mask_bits, right_mask_bits
5093 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5094 add block, psx_gpu, #psx_gpu_blocks_offset
5096 bic texture_offset_base, texture_offset_base, #0xF
5099 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5100 add block, block, num_blocks, lsl #6
5104 vext.32 block_masks_shifted, block_masks, block_masks, #1
5105 vorr.u32 block_masks, block_masks, block_masks_shifted
5106 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5109 add num_blocks, num_blocks, #1
5110 cmp num_blocks, #MAX_BLOCKS
5111 blgt setup_sprites_16bpp_flush_single
5113 and texture_block_ptr, texture_offset_base, texture_mask
5114 subs height, height, #1
5116 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5117 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5119 vst1.u32 { texels }, [ block, :128 ]
5120 add block, block, #40
5122 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5125 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5127 add block, block, #24
5128 add texture_offset_base, texture_offset_base, #2048
5129 add fb_ptr, fb_ptr, #2048
5130 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5133 ldmia sp!, { r4 - r11, pc }
5136 add num_blocks, num_blocks, block_width
5137 mov texture_offset, texture_offset_base
5139 cmp num_blocks, #MAX_BLOCKS
5140 blgt setup_sprites_16bpp_flush_row
5142 add texture_offset_base, texture_offset_base, #2048
5143 and texture_block_ptr, texture_offset, texture_mask
5145 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5146 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5148 vst1.u32 { texels }, [ block, :128 ]
5149 add block, block, #40
5151 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5152 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5155 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5156 subs blocks_remaining, block_width, #2
5158 add texture_offset, texture_offset, #16
5159 add fb_ptr, fb_ptr, #16
5161 vmov.u8 draw_mask_fb_ptr, #0
5163 add block, block, #24
5167 and texture_block_ptr, texture_offset, texture_mask
5168 subs blocks_remaining, blocks_remaining, #1
5170 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5171 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5173 vst1.u32 { texels }, [ block, :128 ]
5174 add block, block, #40
5176 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5179 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5181 add texture_offset, texture_offset, #16
5182 add fb_ptr, fb_ptr, #16
5184 add block, block, #24
5188 and texture_block_ptr, texture_offset, texture_mask
5189 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5191 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5192 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5194 vst1.u32 { texels }, [ block, :128 ]
5195 add block, block, #40
5197 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5198 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5200 add block, block, #24
5201 subs height, height, #1
5203 add fb_ptr, fb_ptr, fb_ptr_pitch
5204 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5208 ldmia sp!, { r4 - r11, pc }
5211 #undef texture_page_ptr
5213 #undef dirty_textures_mask
5214 #undef current_texture_mask
5217 #define current_texture_page r1
5218 #define texture_page_ptr r2
5219 #define vram_ptr_a r3
5220 #define current_texture_page_x r12
5221 #define current_texture_page_y r4
5222 #define dirty_textures_mask r5
5226 #define current_texture_mask r9
5228 #define vram_ptr_b r11
5230 #define texel_block_a d0
5231 #define texel_block_b d1
5232 #define texel_block_expanded_a q1
5233 #define texel_block_expanded_b q2
5234 #define texel_block_expanded_ab q2
5235 #define texel_block_expanded_c q3
5236 #define texel_block_expanded_d q4
5237 #define texel_block_expanded_cd q3
5239 function(update_texture_4bpp_cache)
5240 stmdb sp!, { r4 - r11, r14 }
5243 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5245 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5246 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5248 and current_texture_page_x, current_texture_page, #0xF
5249 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
5251 mov current_texture_page_y, current_texture_page, lsr #4
5252 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5254 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5257 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5258 bic dirty_textures_mask, current_texture_mask
5261 str dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5266 add vram_ptr_b, vram_ptr_a, #2048
5269 vld1.u32 { texel_block_a }, [ vram_ptr_a, :64 ], c_4096
5270 vld1.u32 { texel_block_b }, [ vram_ptr_b, :64 ], c_4096
5272 vmovl.u8 texel_block_expanded_a, texel_block_a
5273 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5274 vmovl.u8 texel_block_expanded_c, texel_block_b
5275 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5277 vbic.u16 texel_block_expanded_a, #0x00F0
5278 vbic.u16 texel_block_expanded_b, #0x00F0
5279 vbic.u16 texel_block_expanded_c, #0x00F0
5280 vbic.u16 texel_block_expanded_d, #0x00F0
5282 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
5283 texel_block_expanded_b
5284 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
5285 texel_block_expanded_d
5287 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
5288 [ texture_page_ptr, :256 ]!
5290 subs sub_y, sub_y, #1
5294 add vram_ptr_a, vram_ptr_a, #8
5295 add vram_ptr_b, vram_ptr_b, #8
5297 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5298 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5300 subs tile_x, tile_x, #1
5304 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5305 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5307 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5308 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5310 subs tile_y, tile_y, #1
5314 ldmia sp!, { r4 - r11, pc }
5317 #undef current_texture_page
5320 #define texture_page r1
5321 #define texture_page_ptr r2
5322 #define vram_ptr_a r3
5323 #define texture_page_x r12
5324 #define texture_page_y r4
5325 #define current_texture_page r5
5330 #define vram_ptr_b r11
5342 function(update_texture_8bpp_cache_slice)
5343 stmdb sp!, { r4 - r11, r14 }
5346 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5347 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5349 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5352 and texture_page_x, texture_page, #0xF
5353 mov texture_page_y, texture_page, lsr #4
5355 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
5358 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
5359 eor current_texture_page, current_texture_page, texture_page
5361 ands current_texture_page, current_texture_page, #0x1
5364 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
5367 add vram_ptr_b, vram_ptr_a, #2048
5370 vld1.u32 { texels_a }, [ vram_ptr_a, :128 ], c_4096
5371 vld1.u32 { texels_b }, [ vram_ptr_b, :128 ], c_4096
5372 vld1.u32 { texels_c }, [ vram_ptr_a, :128 ], c_4096
5373 vld1.u32 { texels_d }, [ vram_ptr_b, :128 ], c_4096
5375 vst1.u32 { texels_a, texels_b }, [ texture_page_ptr, :256 ]!
5376 vst1.u32 { texels_c, texels_d }, [ texture_page_ptr, :256 ]!
5378 subs sub_y, sub_y, #1
5383 add vram_ptr_a, vram_ptr_a, #16
5384 add vram_ptr_b, vram_ptr_b, #16
5386 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5387 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5389 subs tile_x, tile_x, #1
5394 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5395 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5397 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5398 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5400 subs tile_y, tile_y, #1
5401 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
5406 ldmia sp!, { r4 - r11, pc }