2 * Copyright (C) 2011 Gilead Kutnick "Exophase" <exophase@gmail.com>
3 * Copyright (C) 2012 GraÅžvydas Ignotas "notaz" <notasas@gmail.com>
5 * This program is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU General Public License as
7 * published by the Free Software Foundation; either version 2 of
8 * the License, or (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License for more details.
18 #define MAX_BLOCKS_PER_ROW 128
20 #define RENDER_STATE_MASK_EVALUATE 0x20
21 #define RENDER_FLAGS_MODULATE_TEXELS 0x1
22 #define RENDER_FLAGS_BLEND 0x2
23 #define RENDER_INTERLACE_ENABLED 0x1
25 #include "psx_gpu_offsets.h"
27 #define psx_gpu_b_dx_offset (psx_gpu_b_block_span_offset + 4)
29 #define edge_data_left_x_offset 0
30 #define edge_data_num_blocks_offset 2
31 #define edge_data_right_mask_offset 4
32 #define edge_data_y_offset 6
78 #define gw_by_l gw_bx_l
103 #define uvrg_xxxx1 q4
108 #define uvrg_xxxx2 q5
113 #define yyyy_uvrg0 q6
118 #define yyyy_uvrg1 q7
123 #define yyyy_uvrg2 q8
150 #define uvrg_base q11
193 #define uvrgb_phase q13
199 #define function(name) \
201 .type name, %function; \
204 #define JT_OP_REL(table_label, index_reg, temp)
205 #define JT_OP(x...) x
206 #define JTE(start, target) target
210 #define function(name) \
215 #define JT_OP_REL(table_label, index_reg, temp) \
216 adr temp, table_label; \
217 ldr temp, [ temp, index_reg, lsl #2 ]; \
221 #define JTE(start, target) (target - start)
223 #define flush_render_block_buffer _flush_render_block_buffer
224 #define setup_sprite_untextured_simple _setup_sprite_untextured_simple
225 #define update_texture_8bpp_cache _update_texture_8bpp_cache
234 function(compute_all_gradients)
235 // First compute the triangle area reciprocal and shift. The division will
236 // happen concurrently with much of the work which follows.
237 @ r12 = psx_gpu->triangle_area
238 ldr r12, [ psx_gpu, #psx_gpu_triangle_area_offset ]
239 stmdb sp!, { r4 - r11, lr }
241 @ load exponent of 62 into upper half of double
243 clz r14, r12 @ r14 = shift
245 movt r4, #((62 + 1023) << 4)
246 mov r12, r12, lsl r14 @ r12 = triangle_area_normalized
248 @ load area normalized into lower half of double
250 vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n
252 movt r4, #((1022 + 31) << 4)
255 add r4, r4, r12, lsr #11
258 vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n
260 // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) =
261 // ( d0 * d1 ) - ( d2 * d3 ) =
262 // ( m0 ) - ( m1 ) = gradient
264 // This is split to do 12 elements at a time over three sets: a, b, and c.
265 // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so
266 // two of the slots are unused.
268 // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as
271 // First type is: uvrg bxxx xxxx
272 // Second type is: yyyy ybyy uvrg
273 // Since x_a and y_c are the same the same variable is used for both.
275 vld1.u32 { v0 }, [ v_a, : 128 ] @ v0 = { uvrg0, b0, x0, y0 }
276 ldrsh x0, [ v_a, #8 ] @ load x0
278 vld1.u32 { v1 }, [ v_b, : 128 ] @ v1 = { uvrg1, b1, x1, y1}
279 ldrh x1, [ v_b, #8 ] @ load x1
281 vld1.u32 { v2 }, [ v_c, : 128 ] @ v2 = { uvrg2, b2, x2, y2 }
282 ldrh x2, [ v_c, #8 ] @ load x2
284 vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- }
285 ldrh y0, [ v_a, #10 ] @ load y0
287 vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- }
288 ldrh y1, [ v_b, #10 ] @ load y1
290 vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- }
291 ldrh y2, [ v_c, #10 ] @ load y2
293 vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 }
294 vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 }
296 orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 }
297 pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 }
299 vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 }
300 vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 }
302 vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 }
303 vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 }
305 ldrb b2, [ v_c, #4 ] @ load b2
306 orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 }
308 ldrb b1, [ v_b, #4 ] @ load b1
309 orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 }
311 vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 }
312 vsub.s16 d0_ab, x1_ab, x0_ab
314 ldrb b0, [ v_a, #4 ] @ load b0
315 orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 }
317 vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 }
318 vsub.s16 d2_ab, x2_ab, x1_ab
320 vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 }
321 vsub.s16 d1_ab, y2_ab, y1_ab
323 orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 }
324 ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 }
326 ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 }
327 ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 }
329 vsub.s16 d3_ab, y1_ab, y0_ab
330 smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) -
331 @ ((x2 - X1) * (b1 - b0))
332 vmull.s16 ga_uvrg_x, d0_a, d1_a
333 smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) -
334 @ ((b2 - b1) * (y1 - y0))
335 vmlsl.s16 ga_uvrg_x, d2_a, d3_a
336 movs gs_bx, ga_bx, asr #31
338 vmull.s16 ga_uvrg_y, d0_b, d1_b
339 rsbmi ga_bx, ga_bx, #0
341 @ r12 = psx_gpu->uvrgb_phase
342 ldr r12, [ psx_gpu, #psx_gpu_uvrgb_phase_offset ]
344 vmlsl.s16 ga_uvrg_y, d2_b, d3_b
345 movs gs_by, ga_by, asr #31
347 vshr.u64 d0, d30, #22
348 add b_base, r12, b0, lsl #16
350 vdup.u32 uvrgb_phase, r12
352 rsbmi ga_by, ga_by, #0
353 vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0
355 @ r12 = psx_gpu->triangle_winding_offset
356 ldrb r12, [ psx_gpu, #psx_gpu_triangle_winding_offset ]
357 vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0
359 rsb r12, r12, #0 @ r12 = -(triangle->winding)
361 vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w }
362 sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS)
364 vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16
365 vdup.u32 r_shift, r14 @ r_shift = { shift, shift, shift, shift }
367 vadd.u32 uvrg_base, uvrgb_phase
368 vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x)
370 vmov area_r_s, s0 @ area_r_s = triangle_reciprocal
371 vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y)
373 vmull.u32 gw_rg_x, ga_rg_x, d0[0]
374 vmull.u32 gw_uv_x, ga_uv_x, d0[0]
375 vmull.u32 gw_rg_y, ga_rg_y, d0[0]
376 vmull.u32 gw_uv_y, ga_uv_y, d0[0]
378 vshl.u64 gw_rg_x, gw_rg_x, r_shift
379 vshl.u64 gw_uv_x, gw_uv_x, r_shift
380 vshl.u64 gw_rg_y, gw_rg_y, r_shift
381 vshl.u64 gw_uv_y, gw_uv_y, r_shift
383 veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask
384 vmovn.u64 g_uv_x, gw_uv_x
386 veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask
387 vmovn.u64 g_rg_x, gw_rg_x
389 veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
390 vmovn.u64 g_uv_y, gw_uv_y
392 vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
393 vmovn.u64 g_rg_y, gw_rg_y
395 veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
396 mov ga_bx, ga_bx, lsl #13
398 vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
399 mov ga_by, ga_by, lsl #13
402 umull gw_bx_l, gw_bx_h, ga_bx, area_r_s
404 vshl.u32 g_uvrg_x, g_uvrg_x, #4
405 vshl.u32 g_uvrg_y, g_uvrg_y, #4
407 umull gw_by_l, gw_by_h, ga_by, area_r_s
408 vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0]
410 eor gs_bx, gs_bx, r12
411 vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1
413 veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0
414 eor gs_by, gs_by, r12
416 rsb r11, r14, #0 @ r11 = negative shift for scalar lsr
417 add store_a, psx_gpu, #psx_gpu_uvrg_offset
419 sub r11, r11, #(32 - 13)
421 add store_b, store_a, #16
424 vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1
425 vst1.u32 { uvrg_base }, [ store_a, : 128 ], store_inc
427 vst1.u32 { uvrg_dx1 }, [ store_b, : 128 ], store_inc
428 mov g_bx, gw_bx_h, lsr r11
430 vst1.u32 { g_uvrg_y }, [ store_a, : 128 ], store_inc
431 mov g_by, gw_by_h, lsr r11
433 vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \
434 [ store_b, : 128 ], store_inc
435 eor g_bx, g_bx, gs_bx
437 vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \
438 [ store_b, : 128 ], store_inc
439 sub g_bx, g_bx, gs_bx
442 eor g_by, g_by, gs_by
444 mls b_base, g_bx, x0, b_base
445 sub g_by, g_by, gs_by
450 add g_bx2, g_bx, g_bx
451 add g_bx3, g_bx, g_bx2
453 stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by }
455 ldmia sp!, { r4 - r11, pc }
472 #define height_minor_a r7
473 #define height_minor_b r8
474 #define height_major r9
477 #define reciprocal_table_ptr r10
479 #define edge_alt_low r4
480 #define edge_alt_high r5
481 #define edge_dx_dy_alt r6
482 #define edge_shift_alt r10
484 #define edge_dx_dy_alt_low r4
485 #define edge_dx_dy_alt_high r5
487 #define span_edge_data r4
488 #define span_uvrg_offset r5
489 #define span_b_offset r6
497 #define alternate_x q0
498 #define alternate_dx_dy q1
499 #define alternate_x_32 q2
501 #define alternate_x_low d0
502 #define alternate_x_high d1
503 #define alternate_dx_dy_low d2
504 #define alternate_dx_dy_high d3
505 #define alternate_x_32_low d4
506 #define alternate_x_32_high d5
510 #define left_dx_dy q5
511 #define right_dx_dy q6
513 #define right_edge q8
515 #define left_x_low d6
516 #define left_x_high d7
517 #define right_x_low d8
518 #define right_x_high d9
519 #define left_dx_dy_low d10
520 #define left_dx_dy_high d11
521 #define right_dx_dy_low d12
522 #define right_dx_dy_high d13
523 #define left_edge_low d14
524 #define left_edge_high d15
525 #define right_edge_low d16
526 #define right_edge_high d17
528 #define y_mid_point d18
531 #define left_right_x_16 q11
532 #define span_shifts_y q12
535 #define span_shifts d24
540 #define left_right_x_16_low d22
541 #define left_right_x_16_high d23
546 #define alternate_x_16 d4
549 #define v_clip_low d6
551 #define right_x_32 q10
552 #define left_x_32 q11
553 #define alternate_select d24
555 #define right_x_32_low d20
556 #define right_x_32_high d21
557 #define left_x_32_low d22
558 #define left_x_32_high d23
561 #define edges_dx_dy d2
562 #define edge_shifts d3
563 #define edge_shifts_64 q2
565 #define edges_xy_left d0
566 #define edges_xy_right d1
568 #define height_reciprocals d6
576 #define heights_b d12
577 #define edges_dx_dy_64 q10
579 #define edges_dx_dy_64_left d20
580 #define edges_dx_dy_64_right d21
583 #define setup_spans_prologue() \
584 stmdb sp!, { r4 - r11, lr }; \
586 ldrsh x_a, [ v_a, #8 ]; \
587 ldrsh x_b, [ v_b, #8 ]; \
588 ldrsh x_c, [ v_c, #8 ]; \
589 ldrsh y_a, [ v_a, #10 ]; \
590 ldrsh y_b, [ v_b, #10 ]; \
591 ldrsh y_c, [ v_c, #10 ]; \
593 add temp, psx_gpu, #psx_gpu_uvrg_offset; \
594 vld1.32 { uvrg }, [ temp ]; \
595 add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \
596 vld1.32 { uvrg_dy }, [ temp ]; \
597 ldr reciprocal_table_ptr, [ psx_gpu, #psx_gpu_reciprocal_table_ptr_offset ]; \
599 vmov.u32 c_0x01, #0x01 \
601 #define setup_spans_load_b() \
602 ldr b, [ psx_gpu, #psx_gpu_b_offset ]; \
603 ldr b_dy, [ psx_gpu, #psx_gpu_b_dy_offset ] \
605 #define setup_spans_prologue_b() \
606 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
607 add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \
609 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
610 vmov.u16 c_0x0004, #0x0004; \
612 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
613 vmov.u16 c_0x0001, #0x0001; \
615 vld1.u16 { left_edge_low[], left_edge_high[] }, [ temp ]; \
616 add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \
618 vld1.u16 { right_edge_low[], right_edge_high[] }, [ temp ]; \
619 vadd.u16 right_edge, right_edge, c_0x0001; \
621 vmov.u16 c_0x0007, #0x0007; \
622 vmvn.u16 c_0xFFFE, #0x0001 \
625 #define compute_edge_delta_x2() \
626 ldr temp, [ reciprocal_table_ptr, height, lsl #2 ]; \
628 vdup.u32 heights, height; \
629 vsub.u32 widths, x_ends, x_starts; \
631 vdup.u32 edge_shifts, temp; \
632 vsub.u32 heights_b, heights, c_0x01; \
633 vshr.u32 height_reciprocals, edge_shifts, #10; \
635 vmla.s32 heights_b, x_starts, heights; \
636 vbic.u16 edge_shifts, #0xE0; \
637 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
638 vmull.s32 edges_xy, heights_b, height_reciprocals \
641 #define height_reciprocal_alt r11
642 #define height_b_alt r12
644 #define compute_edge_delta_x3(start_c, height_a, height_b) \
645 vmov heights, height_a, height_b; \
646 ldr temp, [ reciprocal_table_ptr, height_a, lsl #2 ]; \
647 vmov.u32 edge_shifts[0], temp; \
648 ldr temp, [ reciprocal_table_ptr, height_b, lsl #2 ]; \
649 vmov.u32 edge_shifts[1], temp; \
650 ldr edge_shift_alt, [ reciprocal_table_ptr, height_minor_b, lsl #2 ]; \
652 vsub.u32 widths, x_ends, x_starts; \
653 sub width_alt, x_c, start_c; \
655 vsub.u32 heights_b, heights, c_0x01; \
656 sub height_b_alt, height_minor_b, #1; \
658 vshr.u32 height_reciprocals, edge_shifts, #10; \
659 lsr height_reciprocal_alt, edge_shift_alt, #10; \
661 vmla.s32 heights_b, x_starts, heights; \
662 mla height_b_alt, height_minor_b, start_c, height_b_alt; \
664 vbic.u16 edge_shifts, #0xE0; \
665 and edge_shift_alt, edge_shift_alt, #0x1F; \
667 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
668 mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \
670 vmull.s32 edges_xy, heights_b, height_reciprocals; \
671 smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \
674 #define setup_spans_adjust_y_up() \
675 vsub.u32 y_x4, y_x4, c_0x0004 \
677 #define setup_spans_adjust_y_down() \
678 vadd.u32 y_x4, y_x4, c_0x0004 \
680 #define setup_spans_adjust_interpolants_up() \
681 vsub.u32 uvrg, uvrg, uvrg_dy; \
684 #define setup_spans_adjust_interpolants_down() \
685 vadd.u32 uvrg, uvrg, uvrg_dy; \
689 #define setup_spans_clip_interpolants_increment() \
690 mla b, b_dy, clip, b; \
691 vmla.s32 uvrg, uvrg_dy, v_clip \
693 #define setup_spans_clip_interpolants_decrement() \
694 mls b, b_dy, clip, b; \
695 vmls.s32 uvrg, uvrg_dy, v_clip \
697 #define setup_spans_clip_alternate_yes() \
698 smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \
700 #define setup_spans_clip_alternate_no() \
702 #define setup_spans_clip(direction, alternate_active) \
703 vdup.u32 v_clip, clip; \
704 setup_spans_clip_alternate_##alternate_active(); \
705 setup_spans_clip_interpolants_##direction(); \
706 vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \
709 #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \
710 vmovl.s32 edge_shifts_64, edge_shifts; \
711 vmovl.s32 edges_dx_dy_64, edges_dx_dy; \
713 vshl.s64 edges_xy, edges_xy, edge_shifts_64; \
714 vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \
716 vmov left_x_low, edges_xy_##left_index; \
717 vmov right_x_low, edges_xy_##right_index; \
719 vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \
720 vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \
721 vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \
722 vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \
724 vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \
725 vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \
727 vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \
728 vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \
731 #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \
732 setup_spans_adjust_edges_alternate_no(left_index, right_index); \
734 vdup.u16 y_mid_point, y_b; \
735 rsb temp, edge_shift_alt, #32; \
737 lsl edge_alt_high, edge_alt_high, edge_shift_alt; \
738 orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \
739 lsl edge_alt_low, edge_alt_low, edge_shift_alt; \
740 vmov alternate_x_low, edge_alt_low, edge_alt_high; \
742 asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \
743 lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \
744 vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \
745 vmov alternate_dx_dy_high, alternate_dx_dy_low; \
747 vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \
748 vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \
751 #define setup_spans_y_select_up() \
752 vclt.s16 alternate_select, y_x4, y_mid_point \
754 #define setup_spans_y_select_down() \
755 vcgt.s16 alternate_select, y_x4, y_mid_point \
758 #define setup_spans_alternate_select_left() \
759 vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \
761 #define setup_spans_alternate_select_right() \
762 vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \
765 #define setup_spans_set_x4_alternate_yes(alternate, direction) \
766 vshrn.s64 alternate_x_32_low, alternate_x, #32; \
767 vshrn.s64 left_x_32_low, left_x, #32; \
768 vshrn.s64 right_x_32_low, right_x, #32; \
770 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
771 vadd.u64 left_x, left_x, left_dx_dy; \
772 vadd.u64 right_x, right_x, right_dx_dy; \
774 vshrn.s64 alternate_x_32_high, alternate_x, #32; \
775 vshrn.s64 left_x_32_high, left_x, #32; \
776 vshrn.s64 right_x_32_high, right_x, #32; \
778 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
779 vadd.u64 left_x, left_x, left_dx_dy; \
780 vadd.u64 right_x, right_x, right_dx_dy; \
782 vmovn.u32 alternate_x_16, alternate_x_32; \
783 setup_spans_y_select_##direction(); \
784 vmovn.u32 left_right_x_16_low, left_x_32; \
786 vmovn.u32 left_right_x_16_high, right_x_32; \
787 setup_spans_alternate_select_##alternate(); \
789 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
790 str b, [ span_b_offset ], #4; \
791 setup_spans_adjust_interpolants_##direction(); \
793 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
795 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
796 str b, [ span_b_offset ], #4; \
797 setup_spans_adjust_interpolants_##direction(); \
799 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
801 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
802 str b, [ span_b_offset ], #4; \
803 setup_spans_adjust_interpolants_##direction(); \
805 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
806 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
807 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
809 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
810 str b, [ span_b_offset ], #4; \
811 setup_spans_adjust_interpolants_##direction(); \
813 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
814 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
816 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
818 setup_spans_adjust_y_##direction() \
821 #define setup_spans_set_x4_alternate_no(alternate, direction) \
822 vshrn.s64 left_x_32_low, left_x, #32; \
823 vshrn.s64 right_x_32_low, right_x, #32; \
825 vadd.u64 left_x, left_x, left_dx_dy; \
826 vadd.u64 right_x, right_x, right_dx_dy; \
828 vshrn.s64 left_x_32_high, left_x, #32; \
829 vshrn.s64 right_x_32_high, right_x, #32; \
831 vadd.u64 left_x, left_x, left_dx_dy; \
832 vadd.u64 right_x, right_x, right_dx_dy; \
834 vmovn.u32 left_right_x_16_low, left_x_32; \
835 vmovn.u32 left_right_x_16_high, right_x_32; \
837 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
838 str b, [ span_b_offset ], #4; \
839 setup_spans_adjust_interpolants_##direction(); \
841 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
843 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
844 str b, [ span_b_offset ], #4; \
845 setup_spans_adjust_interpolants_##direction(); \
847 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
849 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
850 str b, [ span_b_offset ], #4; \
851 setup_spans_adjust_interpolants_##direction(); \
853 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
854 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
855 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
857 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
858 str b, [ span_b_offset ], #4; \
859 setup_spans_adjust_interpolants_##direction(); \
861 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
862 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
864 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
866 setup_spans_adjust_y_##direction() \
869 #define edge_adjust_low r11
870 #define edge_adjust_high r12
872 #define setup_spans_alternate_adjust_yes() \
873 smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \
874 subs edge_alt_low, edge_alt_low, edge_adjust_low; \
875 sbc edge_alt_high, edge_alt_high, edge_adjust_high \
877 #define setup_spans_alternate_adjust_no() \
880 #define setup_spans_down(left_index, right_index, alternate, alternate_active) \
881 setup_spans_alternate_adjust_##alternate_active(); \
882 setup_spans_load_b(); \
884 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
885 subs y_c, y_c, temp; \
886 subgt height, height, y_c; \
887 addgt height, height, #1; \
889 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
890 subs clip, temp, y_a; \
893 sub height, height, clip; \
894 add y_a, y_a, clip; \
895 setup_spans_clip(increment, alternate_active); \
901 orr temp, y_a, y_a, lsl #16; \
902 add temp, temp, #(1 << 16); \
904 add y_a, y_a, #(2 << 16); \
905 vmov y_x4, temp, y_a; \
907 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
909 setup_spans_prologue_b(); \
911 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
914 setup_spans_set_x4_alternate_##alternate_active(alternate, down); \
915 subs height, height, #4; \
921 #define setup_spans_alternate_pre_increment_yes() \
922 adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \
923 adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \
925 #define setup_spans_alternate_pre_increment_no() \
928 #define setup_spans_up_decrement_yes() \
929 suble height, height, #1 \
931 #define setup_spans_up_decrement_no() \
934 #define setup_spans_up(left_index, right_index, alternate, alternate_active) \
935 setup_spans_alternate_adjust_##alternate_active(); \
936 setup_spans_load_b(); \
939 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
940 subs temp, temp, y_c; \
941 subgt height, height, temp; \
942 setup_spans_up_decrement_##alternate_active(); \
944 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
945 subs clip, y_a, temp; \
948 sub height, height, clip; \
949 sub y_a, y_a, clip; \
950 setup_spans_clip(decrement, alternate_active); \
956 orr temp, y_a, y_a, lsl #16; \
957 sub temp, temp, #(1 << 16); \
959 sub y_a, y_a, #(2 << 16); \
960 vmov y_x4, temp, y_a; \
962 vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \
964 setup_spans_alternate_pre_increment_##alternate_active(); \
965 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
967 setup_spans_adjust_interpolants_up(); \
968 setup_spans_prologue_b(); \
970 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
973 setup_spans_set_x4_alternate_##alternate_active(alternate, up); \
974 subs height, height, #4; \
980 #define setup_spans_epilogue() \
981 ldmia sp!, { r4 - r11, pc } \
984 #define setup_spans_up_up(minor, major) \
985 setup_spans_prologue(); \
986 sub height_minor_a, y_a, y_b; \
987 sub height_minor_b, y_b, y_c; \
988 sub height, y_a, y_c; \
990 vdup.u32 x_starts, x_a; \
991 vmov x_ends, x_c, x_b; \
993 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
994 setup_spans_up(major, minor, minor, yes); \
995 setup_spans_epilogue() \
997 function(setup_spans_up_left)
998 setup_spans_up_up(left, right)
1000 function(setup_spans_up_right)
1001 setup_spans_up_up(right, left)
1003 #define setup_spans_down_down(minor, major) \
1004 setup_spans_prologue(); \
1005 sub height_minor_a, y_b, y_a; \
1006 sub height_minor_b, y_c, y_b; \
1007 sub height, y_c, y_a; \
1009 vdup.u32 x_starts, x_a; \
1010 vmov x_ends, x_c, x_b; \
1012 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
1013 setup_spans_down(major, minor, minor, yes); \
1014 setup_spans_epilogue() \
1016 function(setup_spans_down_left)
1017 setup_spans_down_down(left, right)
1019 function(setup_spans_down_right)
1020 setup_spans_down_down(right, left)
1023 #define setup_spans_up_flat() \
1024 sub height, y_a, y_c; \
1026 compute_edge_delta_x2(); \
1027 setup_spans_up(left, right, none, no); \
1028 setup_spans_epilogue() \
1030 function(setup_spans_up_a)
1031 setup_spans_prologue()
1033 vmov x_starts, x_a, x_b
1034 vdup.u32 x_ends, x_c
1036 setup_spans_up_flat()
1038 function(setup_spans_up_b)
1039 setup_spans_prologue()
1041 vdup.u32 x_starts, x_a
1042 vmov x_ends, x_b, x_c
1044 setup_spans_up_flat()
1046 #define setup_spans_down_flat() \
1047 sub height, y_c, y_a; \
1049 compute_edge_delta_x2(); \
1050 setup_spans_down(left, right, none, no); \
1051 setup_spans_epilogue() \
1053 function(setup_spans_down_a)
1054 setup_spans_prologue()
1056 vmov x_starts, x_a, x_b
1057 vdup.u32 x_ends, x_c
1059 setup_spans_down_flat()
1061 function(setup_spans_down_b)
1062 setup_spans_prologue()
1064 vdup.u32 x_starts, x_a
1065 vmov x_ends, x_b, x_c
1067 setup_spans_down_flat()
1072 #define edges_xy_b q11
1073 #define edges_dx_dy_b d26
1074 #define edge_shifts_b d27
1075 #define edges_dx_dy_and_shifts_b q13
1076 #define height_increment d20
1078 #define edges_dx_dy_and_shifts q1
1080 #define edges_xy_b_left d22
1081 #define edges_xy_b_right d23
1083 #define setup_spans_up_down_load_edge_set_b() \
1084 vmov edges_xy, edges_xy_b; \
1085 vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \
1088 function(setup_spans_up_down)
1089 setup_spans_prologue()
1091 // s32 middle_y = y_a;
1092 sub height_minor_a, y_a, y_b
1093 sub height_minor_b, y_c, y_a
1094 sub height_major, y_c, y_b
1096 vmov x_starts, x_a, x_c
1097 vdup.u32 x_ends, x_b
1099 compute_edge_delta_x3(x_a, height_minor_a, height_major)
1102 vmov height_increment, temp, height_minor_b
1103 vmlal.s32 edges_xy, edges_dx_dy, height_increment
1105 vmov edges_xy_b_left, edge_alt_low, edge_alt_high
1106 vmov edges_xy_b_right, edges_xy_right
1108 vmov edge_shifts_b, edge_shifts
1109 vmov.u32 edge_shifts_b[0], edge_shift_alt
1111 vneg.s32 edges_dx_dy_b, edges_dx_dy
1112 vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt
1116 setup_spans_load_b()
1119 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1120 subs temp, temp, y_b
1121 subgt height_minor_a, height_minor_a, temp
1123 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1124 subs clip, y_a, temp
1127 sub height_minor_a, height_minor_a, clip
1129 setup_spans_clip(decrement, no)
1132 cmp height_minor_a, #0
1135 orr temp, y_a, y_a, lsl #16
1136 sub temp, temp, #(1 << 16)
1138 sub y_a, y_a, #(2 << 16)
1139 vmov y_x4, temp, y_a
1141 vaddw.s32 edges_xy, edges_xy, edges_dx_dy
1143 strh height_minor_a, [ psx_gpu, #psx_gpu_num_spans_offset ]
1145 setup_spans_adjust_edges_alternate_no(left, right);
1146 setup_spans_adjust_interpolants_up()
1147 setup_spans_up_down_load_edge_set_b()
1149 setup_spans_prologue_b()
1153 setup_spans_set_x4_alternate_no(none, up)
1154 subs height_minor_a, height_minor_a, #4
1157 add span_edge_data, span_edge_data, height_minor_a, lsl #3
1158 add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4
1159 add span_b_offset, span_b_offset, height_minor_a, lsl #2
1162 add temp, psx_gpu, #psx_gpu_uvrg_offset
1163 vld1.32 { uvrg }, [ temp ]
1166 setup_spans_load_b()
1168 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1170 subgt height_minor_b, height_minor_b, y_c
1171 addgt height_minor_b, height_minor_b, #1
1173 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1174 subs clip, temp, y_a
1177 sub height_minor_b, height_minor_b, clip
1179 setup_spans_clip(increment, no)
1182 cmp height_minor_b, #0
1185 orr temp, y_a, y_a, lsl #16
1186 add temp, temp, #(1 << 16)
1188 add y_a, y_a, #(2 << 16)
1189 vmov y_x4, temp, y_a
1191 setup_spans_adjust_edges_alternate_no(left, right)
1193 ldrh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1194 add temp, temp, height_minor_b
1196 cmp temp, #MAX_SPANS
1199 strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1202 setup_spans_set_x4_alternate_no(none, down)
1203 subs height_minor_b, height_minor_b, #4
1207 setup_spans_epilogue()
1210 setup_spans_up_down_load_edge_set_b()
1211 setup_spans_prologue_b()
1215 // FIXME: overflow corner case
1216 sub temp, temp, height_minor_b
1217 bics height_minor_b, #3
1218 add temp, temp, height_minor_b
1219 strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1223 #undef span_uvrg_offset
1224 #undef span_edge_data
1225 #undef span_b_offset
1230 #define num_spans r1
1231 #define span_uvrg_offset r2
1232 #define span_edge_data r3
1233 #define span_b_offset r4
1235 #define span_num_blocks r6
1239 #define dither_offset_ptr r10
1240 #define block_ptr_a r11
1242 #define num_blocks r14
1244 #define uvrg_dx_ptr r2
1245 #define texture_mask_ptr r3
1246 #define dither_shift r8
1247 #define dither_row r10
1252 #define block_ptr_b r10
1254 #define block_span_ptr r10
1255 #define right_mask r8
1274 #define b_whole_8 d14
1275 #define fb_mask_ptrs d15
1288 #define u_whole_low d16
1289 #define u_whole_high d17
1290 #define v_whole_low d18
1291 #define v_whole_high d19
1292 #define r_whole_low d20
1293 #define r_whole_high d21
1294 #define g_whole_low d22
1295 #define g_whole_high d23
1296 #define b_whole_low d24
1297 #define b_whole_high d25
1302 #define u_whole_8 d26
1303 #define v_whole_8 d27
1304 #define u_whole_8b d24
1305 #define r_whole_8 d24
1306 #define g_whole_8 d25
1308 #define uv_whole_8 q13
1309 #define uv_whole_8b q14
1311 #define dither_offsets q14
1312 #define texture_mask q15
1313 #define texture_mask_u d30
1314 #define texture_mask_v d31
1316 #define dither_offsets_short d28
1320 #define block_span q10
1325 #define draw_mask q1
1326 #define draw_mask_edge q13
1327 #define test_mask q0
1333 #define setup_blocks_texture_swizzled() \
1334 vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \
1335 vsli.u8 u_whole_8, v_whole_8, #4; \
1336 vsri.u8 v_whole_8, u_whole_8b, #4 \
1338 #define setup_blocks_texture_unswizzled() \
1341 #define setup_blocks_shaded_textured_builder(swizzling) \
1344 function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \
1345 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1346 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1348 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1349 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1351 cmp num_spans, #0; \
1354 stmdb sp!, { r4 - r11, r14 }; \
1355 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1357 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
1358 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1360 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1361 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1363 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1364 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1366 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
1367 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1369 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1372 vmov.u8 fb_mask_ptrs, #0; \
1374 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1375 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1377 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1378 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
1380 cmp span_num_blocks, #0; \
1383 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1384 add num_blocks, span_num_blocks, num_blocks; \
1386 cmp num_blocks, #MAX_BLOCKS; \
1390 ldr b, [ span_b_offset ]; \
1391 add fb_ptr, fb_ptr, y, lsl #11; \
1393 vdup.u32 v_left_x, left_x; \
1396 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1397 add fb_ptr, fb_ptr, left_x, lsl #1; \
1399 mla b, b_dx, left_x, b; \
1400 and dither_shift, left_x, #0x03; \
1402 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1403 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1405 mov dither_shift, dither_shift, lsl #3; \
1406 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1409 subs span_num_blocks, span_num_blocks, #1; \
1411 mov dither_row, dither_row, ror dither_shift; \
1412 mov b_dx4, b_dx, lsl #2; \
1414 vdup.u32 dither_offsets_short, dither_row; \
1415 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1417 vdup.u32 b_block, b; \
1418 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1420 vdup.u32 u_block, uv[0]; \
1421 mov b_dx8, b_dx, lsl #3; \
1423 vdup.u32 v_block, uv[1]; \
1424 vdup.u32 r_block, rg[0]; \
1425 vdup.u32 g_block, rg[1]; \
1427 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1429 vadd.u32 u_block, u_block, block_span; \
1430 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1432 vadd.u32 v_block, v_block, block_span; \
1433 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1435 vadd.u32 r_block, r_block, block_span; \
1436 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1438 vadd.u32 g_block, g_block, block_span; \
1439 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
1441 vadd.u32 b_block, b_block, block_span; \
1442 add block_ptr_b, block_ptr_a, #16; \
1444 vshrn.u32 u_whole_low, u_block, #16; \
1445 vshrn.u32 v_whole_low, v_block, #16; \
1446 vshrn.u32 r_whole_low, r_block, #16; \
1447 vshrn.u32 g_whole_low, g_block, #16; \
1449 vdup.u32 dx4, uv_dx4[0]; \
1450 vshrn.u32 b_whole_low, b_block, #16; \
1452 vaddhn.u32 u_whole_high, u_block, dx4; \
1453 vdup.u32 dx4, uv_dx4[1]; \
1455 vaddhn.u32 v_whole_high, v_block, dx4; \
1456 vdup.u32 dx4, rg_dx4[0]; \
1458 vaddhn.u32 r_whole_high, r_block, dx4; \
1459 vdup.u32 dx4, rg_dx4[1]; \
1461 vaddhn.u32 g_whole_high, g_block, dx4; \
1462 vdup.u32 dx4, b_dx4; \
1464 vaddhn.u32 b_whole_high, b_block, dx4; \
1465 vdup.u32 dx8, uv_dx8[0]; \
1467 vadd.u32 u_block, u_block, dx8; \
1468 vdup.u32 dx8, uv_dx8[1]; \
1470 vadd.u32 v_block, v_block, dx8; \
1471 vdup.u32 dx8, rg_dx8[0]; \
1473 vadd.u32 r_block, r_block, dx8; \
1474 vdup.u32 dx8, rg_dx8[1]; \
1476 vadd.u32 g_block, g_block, dx8; \
1477 vdup.u32 dx8, b_dx8; \
1479 vadd.u32 b_block, b_block, dx8; \
1480 vmovn.u16 u_whole_8, u_whole; \
1482 vmovn.u16 v_whole_8, v_whole; \
1484 vmovn.u16 b_whole_8, b_whole; \
1486 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1488 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1489 setup_blocks_texture_##swizzling(); \
1491 vmovn.u16 r_whole_8, r_whole; \
1495 vmovn.u16 g_whole_8, g_whole; \
1496 vshrn.u32 u_whole_low, u_block, #16; \
1498 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1499 vshrn.u32 v_whole_low, v_block, #16; \
1501 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1502 vshrn.u32 r_whole_low, r_block, #16; \
1504 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1505 vshrn.u32 g_whole_low, g_block, #16; \
1507 vdup.u32 dx4, uv_dx4[0]; \
1508 vshrn.u32 b_whole_low, b_block, #16; \
1510 vaddhn.u32 u_whole_high, u_block, dx4; \
1511 vdup.u32 dx4, uv_dx4[1]; \
1513 vaddhn.u32 v_whole_high, v_block, dx4; \
1514 vdup.u32 dx4, rg_dx4[0]; \
1516 vaddhn.u32 r_whole_high, r_block, dx4; \
1517 vdup.u32 dx4, rg_dx4[1]; \
1519 vaddhn.u32 g_whole_high, g_block, dx4; \
1520 vdup.u32 dx4, b_dx4; \
1522 vaddhn.u32 b_whole_high, b_block, dx4; \
1523 vdup.u32 dx8, uv_dx8[0]; \
1525 vadd.u32 u_block, u_block, dx8; \
1526 vdup.u32 dx8, uv_dx8[1]; \
1528 vadd.u32 v_block, v_block, dx8; \
1529 vdup.u32 dx8, rg_dx8[0]; \
1531 vadd.u32 r_block, r_block, dx8; \
1532 vdup.u32 dx8, rg_dx8[1]; \
1534 vadd.u32 g_block, g_block, dx8; \
1535 vdup.u32 dx8, b_dx8; \
1537 vadd.u32 b_block, b_block, dx8; \
1538 vmovn.u16 u_whole_8, u_whole; \
1540 add fb_ptr, fb_ptr, #16; \
1541 vmovn.u16 v_whole_8, v_whole; \
1543 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1544 vmovn.u16 b_whole_8, b_whole; \
1548 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1549 subs span_num_blocks, span_num_blocks, #1; \
1551 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1552 setup_blocks_texture_##swizzling(); \
1554 vmovn.u16 r_whole_8, r_whole; \
1558 vmovn.u16 g_whole_8, g_whole; \
1559 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1561 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1562 vdup.u8 draw_mask, right_mask; \
1564 vmov.u32 fb_mask_ptrs[0], right_mask; \
1565 vtst.u16 draw_mask, draw_mask, test_mask; \
1566 vzip.u8 u_whole_8, v_whole_8; \
1568 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1569 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1570 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1571 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1572 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1575 add span_uvrg_offset, span_uvrg_offset, #16; \
1576 add span_b_offset, span_b_offset, #4; \
1578 add span_edge_data, span_edge_data, #8; \
1579 subs num_spans, num_spans, #1; \
1581 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1584 ldmia sp!, { r4 - r11, pc }; \
1587 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1588 vpush { texture_mask }; \
1589 vpush { uvrg_dx4 }; \
1591 stmdb sp!, { r0 - r3, r12, r14 }; \
1592 bl flush_render_block_buffer; \
1593 ldmia sp!, { r0 - r3, r12, r14 }; \
1595 vpop { uvrg_dx4 }; \
1596 vpop { texture_mask }; \
1598 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1599 vmov.u8 fb_mask_ptrs, #0; \
1601 mov num_blocks, span_num_blocks; \
1602 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1606 setup_blocks_shaded_textured_builder(swizzled)
1607 setup_blocks_shaded_textured_builder(unswizzled)
1610 #define setup_blocks_unshaded_textured_builder(swizzling) \
1613 function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \
1614 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1615 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1617 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1618 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1620 cmp num_spans, #0; \
1623 stmdb sp!, { r4 - r11, r14 }; \
1624 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1626 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1628 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1629 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1631 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1632 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1634 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1636 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1639 vmov.u8 fb_mask_ptrs, #0; \
1641 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1642 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1644 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1645 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
1647 cmp span_num_blocks, #0; \
1650 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1651 add num_blocks, span_num_blocks, num_blocks; \
1653 cmp num_blocks, #MAX_BLOCKS; \
1657 add fb_ptr, fb_ptr, y, lsl #11; \
1659 vdup.u32 v_left_x, left_x; \
1662 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1663 add fb_ptr, fb_ptr, left_x, lsl #1; \
1665 and dither_shift, left_x, #0x03; \
1667 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1668 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1670 mov dither_shift, dither_shift, lsl #3; \
1671 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1674 subs span_num_blocks, span_num_blocks, #1; \
1676 mov dither_row, dither_row, ror dither_shift; \
1678 vdup.u32 dither_offsets_short, dither_row; \
1679 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1681 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1683 vdup.u32 u_block, uv[0]; \
1685 vdup.u32 v_block, uv[1]; \
1686 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1688 vadd.u32 u_block, u_block, block_span; \
1689 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1691 vadd.u32 v_block, v_block, block_span; \
1692 add block_ptr_b, block_ptr_a, #16; \
1694 vshrn.u32 u_whole_low, u_block, #16; \
1695 vshrn.u32 v_whole_low, v_block, #16; \
1697 vdup.u32 dx4, uv_dx4[0]; \
1699 vaddhn.u32 u_whole_high, u_block, dx4; \
1700 vdup.u32 dx4, uv_dx4[1]; \
1702 vaddhn.u32 v_whole_high, v_block, dx4; \
1703 vdup.u32 dx8, uv_dx8[0]; \
1705 vadd.u32 u_block, u_block, dx8; \
1706 vdup.u32 dx8, uv_dx8[1]; \
1708 vadd.u32 v_block, v_block, dx8; \
1709 vmovn.u16 u_whole_8, u_whole; \
1711 vmovn.u16 v_whole_8, v_whole; \
1714 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1716 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1717 setup_blocks_texture_##swizzling(); \
1722 vshrn.u32 u_whole_low, u_block, #16; \
1724 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1725 vshrn.u32 v_whole_low, v_block, #16; \
1727 add block_ptr_b, block_ptr_b, #32; \
1728 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1730 vdup.u32 dx4, uv_dx4[0]; \
1731 vaddhn.u32 u_whole_high, u_block, dx4; \
1732 vdup.u32 dx4, uv_dx4[1]; \
1734 vaddhn.u32 v_whole_high, v_block, dx4; \
1735 vdup.u32 dx8, uv_dx8[0]; \
1737 vadd.u32 u_block, u_block, dx8; \
1738 vdup.u32 dx8, uv_dx8[1]; \
1740 vadd.u32 v_block, v_block, dx8; \
1741 vmovn.u16 u_whole_8, u_whole; \
1743 add fb_ptr, fb_ptr, #16; \
1744 vmovn.u16 v_whole_8, v_whole; \
1746 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1749 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1750 subs span_num_blocks, span_num_blocks, #1; \
1752 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1753 setup_blocks_texture_##swizzling(); \
1758 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1760 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1761 vdup.u8 draw_mask, right_mask; \
1763 vmov.u32 fb_mask_ptrs[0], right_mask; \
1764 vtst.u16 draw_mask, draw_mask, test_mask; \
1765 vzip.u8 u_whole_8, v_whole_8; \
1767 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1768 add block_ptr_b, block_ptr_b, #32; \
1769 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1770 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1771 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1774 add span_uvrg_offset, span_uvrg_offset, #16; \
1775 add span_edge_data, span_edge_data, #8; \
1776 subs num_spans, num_spans, #1; \
1778 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1781 ldmia sp!, { r4 - r11, pc }; \
1784 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1785 vpush { texture_mask }; \
1786 vpush { uvrg_dx4 }; \
1788 stmdb sp!, { r0 - r3, r12, r14 }; \
1789 bl flush_render_block_buffer; \
1790 ldmia sp!, { r0 - r3, r12, r14 }; \
1792 vpop { uvrg_dx4 }; \
1793 vpop { texture_mask }; \
1795 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1796 vmov.u8 fb_mask_ptrs, #0; \
1798 mov num_blocks, span_num_blocks; \
1799 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1803 setup_blocks_unshaded_textured_builder(swizzled)
1804 setup_blocks_unshaded_textured_builder(unswizzled)
1809 function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect)
1810 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1811 veor.u32 draw_mask, draw_mask, draw_mask
1816 stmdb sp!, { r4 - r11, r14 }
1817 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1819 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1821 ubfx color_r, color, #3, #5
1822 ubfx color_g, color, #11, #5
1823 ubfx color_b, color, #19, #5
1825 orr color, color_r, color_b, lsl #10
1826 orr color, color, color_g, lsl #5
1828 vdup.u16 colors, color
1830 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1831 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1833 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1834 add block_ptr_a, block_ptr_a, num_blocks, lsl #6
1837 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1838 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1840 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
1842 cmp span_num_blocks, #0
1845 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1846 add num_blocks, span_num_blocks, num_blocks
1848 cmp num_blocks, #MAX_BLOCKS
1852 add fb_ptr, fb_ptr, y, lsl #11
1855 add fb_ptr, fb_ptr, left_x, lsl #1
1858 subs span_num_blocks, span_num_blocks, #1
1860 add block_ptr_b, block_ptr_a, #16
1863 vmov.u32 fb_mask_ptrs[1], fb_ptr
1867 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_32
1868 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1869 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1871 add fb_ptr, fb_ptr, #16
1872 add block_ptr_b, block_ptr_b, #32
1876 vmov.u32 fb_mask_ptrs[1], fb_ptr
1877 subs span_num_blocks, span_num_blocks, #1
1882 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1884 vdup.u8 draw_mask_edge, right_mask
1885 vtst.u16 draw_mask_edge, draw_mask_edge, test_mask
1887 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1888 vst1.u32 { draw_mask_edge }, [ block_ptr_a, :128 ], c_32
1889 add block_ptr_b, block_ptr_b, #32
1890 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1893 add span_edge_data, span_edge_data, #8
1894 subs num_spans, num_spans, #1
1896 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1899 ldmia sp!, { r4 - r11, pc }
1904 stmdb sp!, { r0 - r3, r12, r14 }
1905 bl flush_render_block_buffer
1906 ldmia sp!, { r0 - r3, r12, r14 }
1910 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1911 veor.u32 draw_mask, draw_mask, draw_mask
1913 mov num_blocks, span_num_blocks
1914 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1918 #define mask_msb_scalar r14
1920 #define msb_mask q15
1922 #define pixels_low d16
1924 #define msb_mask_low d30
1925 #define msb_mask_high d31
1930 function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct)
1931 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1936 stmdb sp!, { r4 - r11, r14 }
1938 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1940 ubfx color_r, color, #3, #5
1941 ubfx color_g, color, #11, #5
1943 ldrh mask_msb_scalar, [ psx_gpu, #psx_gpu_mask_msb_offset ]
1944 ubfx color_b, color, #19, #5
1946 orr color, color_r, color_b, lsl #10
1947 orr color, color, color_g, lsl #5
1948 orr color, color, mask_msb_scalar
1950 vdup.u16 colors, color
1952 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1953 orr color, color, color, lsl #16
1957 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1958 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1960 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
1962 cmp span_num_blocks, #0
1965 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1967 add fb_ptr, fb_ptr, y, lsl #11
1968 subs span_num_blocks, span_num_blocks, #1
1970 add fb_ptr, fb_ptr, left_x, lsl #1
1974 vst1.u32 { colors }, [ fb_ptr ]!
1975 subs span_num_blocks, span_num_blocks, #1
1980 ldrb right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1982 cmp right_mask, #0x0
1985 tst right_mask, #0xF
1986 streq color, [ fb_ptr ], #4
1987 moveq right_mask, right_mask, lsr #4
1988 streq color, [ fb_ptr ], #4
1990 tst right_mask, #0x3
1991 streq color, [ fb_ptr ], #4
1992 moveq right_mask, right_mask, lsr #2
1994 tst right_mask, #0x1
1995 strheq color, [ fb_ptr ]
1998 add span_edge_data, span_edge_data, #8
1999 subs num_spans, num_spans, #1
2002 ldmia sp!, { r4 - r11, pc }
2005 vst1.u32 { colors }, [ fb_ptr ]
2012 #define rg_dx_ptr r2
2030 #undef dither_offsets
2050 #define r_whole_low d6
2051 #define r_whole_high d7
2052 #define g_whole_low d8
2053 #define g_whole_high d9
2054 #define b_whole_low d10
2055 #define b_whole_high d11
2057 #define gb_whole_8 q6
2059 #define g_whole_8 d12
2060 #define b_whole_8 d13
2062 #define r_whole_8 d14
2074 #define block_span q5
2082 #define d128_0x7 q13
2086 #define dither_offsets q14
2087 #define draw_mask q15
2089 #define dither_offsets_low d28
2092 #define test_mask q10
2095 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2096 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2097 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2099 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2100 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2101 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2103 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2105 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2108 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2111 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2112 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2113 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2115 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2117 cmp num_spans, #0; \
2120 stmdb sp!, { r4 - r11, r14 }; \
2121 vshl.u32 rg_dx4, rg_dx, #2; \
2123 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2124 vshl.u32 rg_dx8, rg_dx, #3; \
2126 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2128 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2129 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2131 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2132 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2134 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2135 vmov.u8 d64_1, #1; \
2137 vmov.u8 d128_4, #4; \
2138 vmov.u8 d64_128, #128; \
2140 vmov.u8 d128_0x7, #0x7; \
2143 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2144 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2146 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2147 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
2149 cmp span_num_blocks, #0; \
2152 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2153 add num_blocks, span_num_blocks, num_blocks; \
2155 cmp num_blocks, #MAX_BLOCKS; \
2159 ldr b, [ span_b_offset ]; \
2160 add fb_ptr, fb_ptr, y, lsl #11; \
2162 vdup.u32 v_left_x, left_x; \
2165 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2166 add fb_ptr, fb_ptr, left_x, lsl #1; \
2168 mla b, b_dx, left_x, b; \
2169 and dither_shift, left_x, #0x03; \
2171 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2172 vshr.u32 rg_dx, rg_dx4, #2; \
2174 mov dither_shift, dither_shift, lsl #3; \
2175 vmla.u32 rg, rg_dx, v_left_x; \
2178 subs span_num_blocks, span_num_blocks, #1; \
2180 mov dither_row, dither_row, ror dither_shift; \
2181 mov b_dx4, b_dx, lsl #2; \
2183 vdup.u32 dither_offsets, dither_row; \
2184 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2186 vdup.u32 b_block, b; \
2187 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2189 mov b_dx8, b_dx, lsl #3; \
2190 vdup.u32 r_block, rg[0]; \
2191 vdup.u32 g_block, rg[1]; \
2193 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2195 vadd.u32 r_block, r_block, block_span; \
2196 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2198 vadd.u32 g_block, g_block, block_span; \
2199 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2201 vadd.u32 b_block, b_block, block_span; \
2202 add block_ptr_b, block_ptr_a, #16; \
2204 vshrn.u32 r_whole_low, r_block, #16; \
2205 vshrn.u32 g_whole_low, g_block, #16; \
2206 vshrn.u32 b_whole_low, b_block, #16; \
2207 vdup.u32 dx4, rg_dx4[0]; \
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 vadd.u32 r_block, r_block, dx8; \
2219 vdup.u32 dx8, rg_dx8[1]; \
2221 vadd.u32 g_block, g_block, dx8; \
2222 vdup.u32 dx8, b_dx8; \
2224 vadd.u32 b_block, b_block, dx8; \
2226 vmovn.u16 r_whole_8, r_whole; \
2227 vmovn.u16 g_whole_8, g_whole; \
2228 vmovn.u16 b_whole_8, b_whole; \
2231 veor.u32 draw_mask, draw_mask, draw_mask; \
2234 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2235 vshrn.u32 r_whole_low, r_block, #16; \
2237 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2238 vshrn.u32 g_whole_low, g_block, #16; \
2240 vshrn.u32 b_whole_low, b_block, #16; \
2241 str fb_ptr, [ block_ptr_a, #44 ]; \
2243 vdup.u32 dx4, rg_dx4[0]; \
2244 vshr.u8 r_whole_8, r_whole_8, #3; \
2245 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2247 vaddhn.u32 r_whole_high, r_block, dx4; \
2248 vdup.u32 dx4, rg_dx4[1]; \
2250 vaddhn.u32 g_whole_high, g_block, dx4; \
2251 vdup.u32 dx4, b_dx4; \
2253 vaddhn.u32 b_whole_high, b_block, dx4; \
2254 vdup.u32 dx8, rg_dx8[0]; \
2256 vmull.u8 pixels, r_whole_8, d64_1; \
2257 vmlal.u8 pixels, g_whole_8, d64_4; \
2258 vmlal.u8 pixels, b_whole_8, d64_128; \
2260 vadd.u32 r_block, r_block, dx8; \
2261 vdup.u32 dx8, rg_dx8[1]; \
2263 vadd.u32 g_block, g_block, dx8; \
2264 vdup.u32 dx8, b_dx8; \
2266 vadd.u32 b_block, b_block, dx8; \
2267 add fb_ptr, fb_ptr, #16; \
2269 vmovn.u16 r_whole_8, r_whole; \
2270 vmovn.u16 g_whole_8, g_whole; \
2271 vmovn.u16 b_whole_8, b_whole; \
2273 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2274 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2278 subs span_num_blocks, span_num_blocks, #1; \
2282 str fb_ptr, [ block_ptr_a, #44 ]; \
2283 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2285 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2286 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2288 vshr.u8 r_whole_8, r_whole_8, #3; \
2289 vdup.u8 draw_mask, right_mask; \
2291 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2292 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
2294 vtst.u16 draw_mask, draw_mask, test_mask; \
2296 vmull.u8 pixels, r_whole_8, d64_1; \
2297 vmlal.u8 pixels, g_whole_8, d64_4; \
2298 vmlal.u8 pixels, b_whole_8, d64_128; \
2300 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2301 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2304 add span_uvrg_offset, span_uvrg_offset, #16; \
2305 add span_b_offset, span_b_offset, #4; \
2307 add span_edge_data, span_edge_data, #8; \
2308 subs num_spans, num_spans, #1; \
2310 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2313 ldmia sp!, { r4 - r11, pc }; \
2316 /* TODO: Load from psx_gpu instead of saving/restoring these */\
2319 stmdb sp!, { r0 - r3, r12, r14 }; \
2320 bl flush_render_block_buffer; \
2321 ldmia sp!, { r0 - r3, r12, r14 }; \
2325 vmov.u8 d64_1, #1; \
2326 vmov.u8 d128_4, #4; \
2327 vmov.u8 d64_128, #128; \
2328 vmov.u8 d128_0x7, #0x7; \
2330 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2332 mov num_blocks, span_num_blocks; \
2333 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2337 setup_blocks_shaded_untextured_indirect_builder(undithered)
2338 setup_blocks_shaded_untextured_indirect_builder(dithered)
2343 #define mask_msb_ptr r14
2345 #define draw_mask q0
2346 #define pixels_low d16
2347 #define pixels_high d17
2351 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2354 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2355 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2356 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2358 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2360 cmp num_spans, #0; \
2363 stmdb sp!, { r4 - r11, r14 }; \
2364 vshl.u32 rg_dx4, rg_dx, #2; \
2366 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2367 vshl.u32 rg_dx8, rg_dx, #3; \
2369 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2370 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2372 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2373 vmov.u8 d64_1, #1; \
2375 vmov.u8 d128_4, #4; \
2376 vmov.u8 d64_128, #128; \
2378 vmov.u8 d128_0x7, #0x7; \
2379 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2380 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
2383 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2384 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2386 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2387 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
2389 cmp span_num_blocks, #0; \
2392 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2393 add fb_ptr, fb_ptr, y, lsl #11; \
2395 ldr b, [ span_b_offset ]; \
2396 vdup.u32 v_left_x, left_x; \
2399 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2400 add fb_ptr, fb_ptr, left_x, lsl #1; \
2402 mla b, b_dx, left_x, b; \
2403 and dither_shift, left_x, #0x03; \
2405 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2406 vshr.u32 rg_dx, rg_dx4, #2; \
2408 mov dither_shift, dither_shift, lsl #3; \
2409 vmla.u32 rg, rg_dx, v_left_x; \
2411 subs span_num_blocks, span_num_blocks, #1; \
2413 mov dither_row, dither_row, ror dither_shift; \
2414 mov b_dx4, b_dx, lsl #2; \
2416 vdup.u32 dither_offsets, dither_row; \
2417 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2419 vdup.u32 b_block, b; \
2420 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2422 mov b_dx8, b_dx, lsl #3; \
2423 vdup.u32 r_block, rg[0]; \
2424 vdup.u32 g_block, rg[1]; \
2426 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2428 vadd.u32 r_block, r_block, block_span; \
2429 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2431 vadd.u32 g_block, g_block, block_span; \
2432 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2434 vadd.u32 b_block, b_block, block_span; \
2435 add block_ptr_b, block_ptr_a, #16; \
2437 vshrn.u32 r_whole_low, r_block, #16; \
2438 vshrn.u32 g_whole_low, g_block, #16; \
2439 vshrn.u32 b_whole_low, b_block, #16; \
2440 vdup.u32 dx4, rg_dx4[0]; \
2442 vaddhn.u32 r_whole_high, r_block, dx4; \
2443 vdup.u32 dx4, rg_dx4[1]; \
2445 vaddhn.u32 g_whole_high, g_block, dx4; \
2446 vdup.u32 dx4, b_dx4; \
2448 vaddhn.u32 b_whole_high, b_block, dx4; \
2449 vdup.u32 dx8, rg_dx8[0]; \
2451 vadd.u32 r_block, r_block, dx8; \
2452 vdup.u32 dx8, rg_dx8[1]; \
2454 vadd.u32 g_block, g_block, dx8; \
2455 vdup.u32 dx8, b_dx8; \
2457 vadd.u32 b_block, b_block, dx8; \
2459 vmovn.u16 r_whole_8, r_whole; \
2460 vmovn.u16 g_whole_8, g_whole; \
2461 vmovn.u16 b_whole_8, b_whole; \
2466 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2467 vshrn.u32 r_whole_low, r_block, #16; \
2469 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2470 vshrn.u32 g_whole_low, g_block, #16; \
2472 vshrn.u32 b_whole_low, b_block, #16; \
2474 vdup.u32 dx4, rg_dx4[0]; \
2475 vshr.u8 r_whole_8, r_whole_8, #3; \
2476 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2478 vaddhn.u32 r_whole_high, r_block, dx4; \
2479 vdup.u32 dx4, rg_dx4[1]; \
2481 vmov pixels, msb_mask; \
2482 vaddhn.u32 g_whole_high, g_block, dx4; \
2483 vdup.u32 dx4, b_dx4; \
2485 vaddhn.u32 b_whole_high, b_block, dx4; \
2486 vdup.u32 dx8, rg_dx8[0]; \
2488 vmlal.u8 pixels, r_whole_8, d64_1; \
2489 vmlal.u8 pixels, g_whole_8, d64_4; \
2490 vmlal.u8 pixels, b_whole_8, d64_128; \
2492 vadd.u32 r_block, r_block, dx8; \
2493 vdup.u32 dx8, rg_dx8[1]; \
2495 vadd.u32 g_block, g_block, dx8; \
2496 vdup.u32 dx8, b_dx8; \
2498 vadd.u32 b_block, b_block, dx8; \
2500 vmovn.u16 r_whole_8, r_whole; \
2501 vmovn.u16 g_whole_8, g_whole; \
2502 vmovn.u16 b_whole_8, b_whole; \
2504 vst1.u32 { pixels }, [ fb_ptr ]!; \
2505 subs span_num_blocks, span_num_blocks, #1; \
2509 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2511 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2512 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2514 vshr.u8 r_whole_8, r_whole_8, #3; \
2515 rbit right_mask, right_mask; \
2516 vmov pixels, msb_mask; \
2517 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2518 clz right_mask, right_mask; \
2520 vmlal.u8 pixels, r_whole_8, d64_1; \
2521 vmlal.u8 pixels, g_whole_8, d64_4; \
2522 vmlal.u8 pixels, b_whole_8, d64_128; \
2524 JT_OP_REL(100f, right_mask, temp); \
2525 JT_OP(ldr pc, [ pc, right_mask, lsl #2 ]); \
2529 .word JTE(100b, 4f); \
2530 .word JTE(100b, 5f); \
2531 .word JTE(100b, 6f); \
2532 .word JTE(100b, 7f); \
2533 .word JTE(100b, 8f); \
2534 .word JTE(100b, 9f); \
2535 .word JTE(100b, 10f); \
2536 .word JTE(100b, 11f); \
2539 vst1.u16 { pixels_low[0] }, [ fb_ptr ]; \
2543 vst1.u32 { pixels_low[0] }, [ fb_ptr ]; \
2547 vst1.u32 { pixels_low[0] }, [ fb_ptr ]!; \
2548 vst1.u16 { pixels_low[2] }, [ fb_ptr ]; \
2552 vst1.u32 { pixels_low }, [ fb_ptr ]; \
2556 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2557 vst1.u16 { pixels_high[0] }, [ fb_ptr ]; \
2561 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2562 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2566 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2567 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2568 vst1.u16 { pixels_high[2] }, [ fb_ptr ]; \
2572 vst1.u32 { pixels }, [ fb_ptr ]; \
2576 add span_uvrg_offset, span_uvrg_offset, #16; \
2577 add span_b_offset, span_b_offset, #4; \
2579 add span_edge_data, span_edge_data, #8; \
2580 subs num_spans, num_spans, #1; \
2584 ldmia sp!, { r4 - r11, pc } \
2586 setup_blocks_shaded_untextured_direct_builder(undithered)
2587 setup_blocks_shaded_untextured_direct_builder(dithered)
2596 #define block_ptr r1
2597 #define num_blocks r2
2610 #define texture_ptr r11
2624 #define pixels_d r10
2628 #define clut_ptr r12
2629 #define current_texture_mask r5
2630 #define dirty_textures_mask r6
2634 #define clut_low_a d2
2635 #define clut_low_b d3
2636 #define clut_high_a d4
2637 #define clut_high_b d5
2642 #define texels_low d6
2643 #define texels_high d7
2647 function(texture_blocks_untextured)
2653 function(texture_blocks_4bpp)
2654 stmdb sp!, { r3 - r11, r14 }
2655 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2657 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2658 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2660 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2661 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]
2663 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2664 vuzp.u8 clut_a, clut_b
2666 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
2667 tst dirty_textures_mask, current_texture_mask
2673 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2675 uxtah uv_0, texture_ptr, uv_01
2676 uxtah uv_1, texture_ptr, uv_01, ror #16
2678 uxtah uv_2, texture_ptr, uv_23
2679 uxtah uv_3, texture_ptr, uv_23, ror #16
2681 uxtah uv_4, texture_ptr, uv_45
2682 ldrb pixel_0, [ uv_0 ]
2684 uxtah uv_5, texture_ptr, uv_45, ror #16
2685 ldrb pixel_1, [ uv_1 ]
2687 uxtah uv_6, texture_ptr, uv_67
2688 ldrb pixel_2, [ uv_2 ]
2690 uxtah uv_7, texture_ptr, uv_67, ror #16
2691 ldrb pixel_3, [ uv_3 ]
2693 ldrb pixel_4, [ uv_4 ]
2694 subs num_blocks, num_blocks, #1
2696 ldrb pixel_5, [ uv_5 ]
2697 orr pixels_a, pixel_0, pixel_1, lsl #8
2699 ldrb pixel_6, [ uv_6 ]
2700 orr pixels_b, pixel_4, pixel_5, lsl #8
2702 ldrb pixel_7, [ uv_7 ]
2703 orr pixels_a, pixels_a, pixel_2, lsl #16
2705 orr pixels_b, pixels_b, pixel_6, lsl #16
2706 orr pixels_a, pixels_a, pixel_3, lsl #24
2708 orr pixels_b, pixels_b, pixel_7, lsl #24
2709 vmov texels, pixels_a, pixels_b
2711 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2712 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2714 vst2.u8 { texels_low, texels_high }, [ block_ptr, :128 ], c_64
2717 ldmia sp!, { r3 - r11, pc }
2720 stmdb sp!, { r1 - r2 }
2721 bl update_texture_4bpp_cache
2724 ldmia sp!, { r1 - r2 }
2730 function(texture_blocks_8bpp)
2731 stmdb sp!, { r3 - r11, r14 }
2732 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2734 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2735 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2737 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2738 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2740 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]
2741 tst dirty_textures_mask, current_texture_mask
2747 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2749 uxtah uv_0, texture_ptr, uv_01
2750 uxtah uv_1, texture_ptr, uv_01, ror #16
2752 uxtah uv_2, texture_ptr, uv_23
2753 uxtah uv_3, texture_ptr, uv_23, ror #16
2755 uxtah uv_4, texture_ptr, uv_45
2756 ldrb pixel_0, [ uv_0 ]
2758 uxtah uv_5, texture_ptr, uv_45, ror #16
2759 ldrb pixel_1, [ uv_1 ]
2761 uxtah uv_6, texture_ptr, uv_67
2762 ldrb pixel_2, [ uv_2 ]
2764 uxtah uv_7, texture_ptr, uv_67, ror #16
2765 ldrb pixel_3, [ uv_3 ]
2767 ldrb pixel_4, [ uv_4 ]
2768 add pixel_0, pixel_0, pixel_0
2770 ldrb pixel_5, [ uv_5 ]
2771 add pixel_1, pixel_1, pixel_1
2773 ldrb pixel_6, [ uv_6 ]
2774 add pixel_2, pixel_2, pixel_2
2776 ldrb pixel_7, [ uv_7 ]
2777 add pixel_3, pixel_3, pixel_3
2779 ldrh pixel_0, [ clut_ptr, pixel_0 ]
2780 add pixel_4, pixel_4, pixel_4
2782 ldrh pixel_1, [ clut_ptr, pixel_1 ]
2783 add pixel_5, pixel_5, pixel_5
2785 ldrh pixel_2, [ clut_ptr, pixel_2 ]
2786 add pixel_6, pixel_6, pixel_6
2788 ldrh pixel_3, [ clut_ptr, pixel_3 ]
2789 add pixel_7, pixel_7, pixel_7
2791 ldrh pixel_4, [ clut_ptr, pixel_4 ]
2792 orr pixels_a, pixel_0, pixel_1, lsl #16
2794 ldrh pixel_5, [ clut_ptr, pixel_5 ]
2795 orr pixels_c, pixel_2, pixel_3, lsl #16
2797 ldrh pixel_6, [ clut_ptr, pixel_6 ]
2798 subs num_blocks, num_blocks, #1
2800 ldrh pixel_7, [ clut_ptr, pixel_7 ]
2801 orr pixels_b, pixel_4, pixel_5, lsl #16
2803 orr pixels_d, pixel_6, pixel_7, lsl #16
2804 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2806 add block_ptr, block_ptr, #64
2809 ldmia sp!, { r3 - r11, pc }
2812 stmdb sp!, { r1 - r2, r12 }
2814 bl update_texture_8bpp_cache
2816 ldmia sp!, { r1 - r2, r12 }
2846 #define block_ptr r1
2847 #define num_blocks r2
2891 #define texture_ptr r12
2896 function(texture_blocks_16bpp)
2897 stmdb sp!, { r3 - r11, r14 }
2898 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2900 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2901 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2904 ldrh uv_0, [ block_ptr ]
2905 subs num_blocks, num_blocks, #1
2907 ldrh uv_1, [ block_ptr, #2 ]
2909 and v_0, uv_0, #0xFF00
2910 and v_1, uv_1, #0xFF00
2912 and u_0, uv_0, #0xFF
2913 and u_1, uv_1, #0xFF
2915 add uv_0, u_0, v_0, lsl #2
2916 ldrh uv_2, [ block_ptr, #4 ]
2918 add uv_1, u_1, v_1, lsl #2
2919 ldrh uv_3, [ block_ptr, #6 ]
2921 add uv_0, uv_0, uv_0
2922 add uv_1, uv_1, uv_1
2924 and v_2, uv_2, #0xFF00
2925 and v_3, uv_3, #0xFF00
2927 and u_2, uv_2, #0xFF
2928 and u_3, uv_3, #0xFF
2930 add uv_2, u_2, v_2, lsl #2
2931 ldrh uv_4, [ block_ptr, #8 ]
2933 add uv_3, u_3, v_3, lsl #2
2934 ldrh uv_5, [ block_ptr, #10 ]
2936 add uv_2, uv_2, uv_2
2937 add uv_3, uv_3, uv_3
2939 and v_4, uv_4, #0xFF00
2940 and v_5, uv_5, #0xFF00
2942 and u_4, uv_4, #0xFF
2943 and u_5, uv_5, #0xFF
2945 add uv_4, u_4, v_4, lsl #2
2946 ldrh uv_6, [ block_ptr, #12 ]
2948 add uv_5, u_5, v_5, lsl #2
2949 ldrh uv_7, [ block_ptr, #14 ]
2951 add uv_4, uv_4, uv_4
2952 ldrh pixel_0, [ texture_ptr, uv_0 ]
2954 add uv_5, uv_5, uv_5
2955 ldrh pixel_1, [ texture_ptr, uv_1 ]
2957 and v_6, uv_6, #0xFF00
2958 ldrh pixel_2, [ texture_ptr, uv_2 ]
2960 and v_7, uv_7, #0xFF00
2961 ldrh pixel_3, [ texture_ptr, uv_3 ]
2963 and u_6, uv_6, #0xFF
2964 ldrh pixel_4, [ texture_ptr, uv_4 ]
2966 and u_7, uv_7, #0xFF
2967 ldrh pixel_5, [ texture_ptr, uv_5 ]
2969 add uv_6, u_6, v_6, lsl #2
2970 add uv_7, u_7, v_7, lsl #2
2972 add uv_6, uv_6, uv_6
2973 add uv_7, uv_7, uv_7
2975 orr pixels_a, pixel_0, pixel_1, lsl #16
2976 orr pixels_b, pixel_2, pixel_3, lsl #16
2978 ldrh pixel_6, [ texture_ptr, uv_6 ]
2979 orr pixels_c, pixel_4, pixel_5, lsl #16
2981 ldrh pixel_7, [ texture_ptr, uv_7 ]
2982 orr pixels_d, pixel_6, pixel_7, lsl #16
2984 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
2985 add block_ptr, block_ptr, #64
2989 ldmia sp!, { r3 - r11, pc }
3004 #undef msb_mask_high
3012 #define num_blocks r1
3013 #define color_ptr r2
3014 #define colors_scalar r2
3015 #define colors_scalar_compare r3
3016 #define mask_msb_ptr r2
3018 #define block_ptr_load_a r0
3019 #define block_ptr_store r3
3020 #define block_ptr_load_b r12
3025 #define draw_mask_bits_scalar r5
3027 #define d128_0x07 q0
3028 #define d128_0x1F q1
3029 #define d128_0x8000 q2
3030 #define test_mask q3
3032 #define colors_rg q5
3033 #define colors_b_dm_bits q6
3034 #define texels_rg q7
3037 #define pixels_b q10
3039 #define zero_mask q4
3040 #define draw_mask q12
3041 #define msb_mask q13
3043 #define fb_pixels q8
3045 #define pixels_gb_low q9
3047 #define colors_r d10
3048 #define colors_g d11
3049 #define colors_b d12
3050 #define draw_mask_bits d13
3051 #define texels_r d14
3052 #define texels_g d15
3053 #define pixels_r_low d16
3054 #define pixels_g_low d18
3055 #define pixels_b_low d19
3056 #define msb_mask_low d26
3057 #define msb_mask_high d27
3062 #define texels_b d31
3064 #define shade_blocks_textured_modulated_prologue_indirect() \
3066 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3068 #define shade_blocks_textured_modulated_prologue_direct() \
3069 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3070 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ] \
3073 #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \
3075 #define shade_blocks_textured_false_modulation_check_undithered(target) \
3076 ldr colors_scalar, [ psx_gpu, #psx_gpu_triangle_color_offset ]; \
3077 movw colors_scalar_compare, #0x8080; \
3079 movt colors_scalar_compare, #0x80; \
3080 cmp colors_scalar, colors_scalar_compare; \
3081 beq shade_blocks_textured_unmodulated_##target \
3083 #define shade_blocks_textured_false_modulation_check_dithered(target) \
3085 #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \
3086 shade_blocks_textured_false_modulation_check_##dithering(target); \
3087 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3088 vld1.u32 { colors_r[] }, [ color_ptr, :32 ]; \
3089 vdup.u8 colors_g, colors_r[1]; \
3090 vdup.u8 colors_b, colors_r[2]; \
3091 vdup.u8 colors_r, colors_r[0] \
3094 #define shade_blocks_textured_modulated_load_dithered(target) \
3095 vld1.u32 { target }, [ block_ptr_load_b, :128 ] \
3097 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3098 vld1.u32 { target }, [ block_ptr_load_b, :128 ], c_32 \
3100 #define shade_blocks_textured_modulated_load_undithered(target) \
3102 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3103 add block_ptr_load_b, block_ptr_load_b, #32 \
3105 #define shade_blocks_textured_modulate_dithered(channel) \
3106 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3108 #define shade_blocks_textured_modulate_undithered(channel) \
3109 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3112 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3113 vst1.u32 { draw_mask }, [ block_ptr_store, :128 ]! \
3115 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3116 ldr fb_ptr, [ block_ptr_load_b, #(offset - 64) ]; \
3117 vld1.u32 { fb_pixels }, [ fb_ptr ]; \
3118 vbit.u16 pixels, fb_pixels, draw_mask \
3120 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3121 vst1.u32 { pixels }, [ block_ptr_store, :128 ], c_48 \
3123 #define shade_blocks_textured_modulated_store_pixels_direct() \
3124 vst1.u32 { pixels }, [ fb_ptr ] \
3127 #define shade_blocks_textured_modulated_load_rg_shaded() \
3128 vld1.u32 { colors_r, colors_g }, [ block_ptr_load_b, :128 ], c_32 \
3130 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3131 add block_ptr_load_b, block_ptr_load_b, #32 \
3133 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3134 vld1.u32 { colors_b, draw_mask_bits }, [ block_ptr_load_a, :128 ], c_32 \
3136 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3137 ldr draw_mask_bits_scalar, [ block_ptr_load_a, #8 ]; \
3138 add block_ptr_load_a, block_ptr_load_a, #32 \
3140 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3141 vdup.u16 draw_mask, draw_mask_bits[0] \
3143 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3144 vdup.u16 draw_mask, draw_mask_bits_scalar \
3147 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3149 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3150 vorr.u16 pixels, pixels, msb_mask \
3153 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3156 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3157 shade_blocks_textured_modulated_prologue_##shading(dithering, target); \
3158 stmdb sp!, { r4 - r5, lr }; \
3159 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3161 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
3163 shade_blocks_textured_modulated_prologue_##target(); \
3165 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3168 add block_ptr_load_b, block_ptr_load_a, #16; \
3169 vmov.u8 d64_1, #1; \
3170 vmov.u8 d64_4, #4; \
3171 vmov.u8 d64_128, #128; \
3173 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3174 vmov.u8 d128_0x07, #0x07; \
3176 shade_blocks_textured_modulated_load_rg_##shading(); \
3177 vmov.u8 d128_0x1F, #0x1F; \
3179 shade_blocks_textured_modulated_load_bdm_##shading(); \
3180 vmov.u16 d128_0x8000, #0x8000; \
3182 vmovn.u16 texels_r, texels; \
3183 vshrn.u16 texels_g, texels, #5; \
3185 vshrn.u16 texels_b, texels, #7; \
3186 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3188 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3189 vtst.u16 draw_mask, draw_mask, test_mask; \
3191 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3192 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3194 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3195 vshr.u8 texels_b, texels_b, #3; \
3197 shade_blocks_textured_modulate_##dithering(r); \
3198 shade_blocks_textured_modulate_##dithering(g); \
3199 shade_blocks_textured_modulate_##dithering(b); \
3201 vand.u16 pixels, texels, d128_0x8000; \
3202 vceq.u16 zero_mask, texels, #0; \
3204 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3205 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3206 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3208 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3209 vorr.u16 draw_mask, draw_mask, zero_mask; \
3210 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3211 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3213 subs num_blocks, num_blocks, #1; \
3219 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3220 shade_blocks_textured_modulated_load_rg_##shading(); \
3221 vshrn.u16 texels_g, texels, #5; \
3223 shade_blocks_textured_modulated_load_bdm_##shading(); \
3224 vshrn.u16 texels_b, texels, #7; \
3226 pld [ block_ptr_load_a ]; \
3227 vmovn.u16 texels_r, texels; \
3228 vmlal.u8 pixels, pixels_r_low, d64_1; \
3230 vmlal.u8 pixels, pixels_g_low, d64_4; \
3231 vmlal.u8 pixels, pixels_b_low, d64_128; \
3232 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3234 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3235 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3237 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3238 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3240 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3241 vtst.u16 draw_mask, draw_mask, test_mask; \
3243 shade_blocks_textured_modulated_store_pixels_##target(); \
3244 vshr.u8 texels_b, texels_b, #3; \
3246 shade_blocks_textured_modulate_##dithering(r); \
3247 shade_blocks_textured_modulate_##dithering(g); \
3248 shade_blocks_textured_modulate_##dithering(b); \
3250 vand.u16 pixels, texels, d128_0x8000; \
3251 vceq.u16 zero_mask, texels, #0; \
3253 subs num_blocks, num_blocks, #1; \
3255 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3256 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3257 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3259 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3260 vorr.u16 draw_mask, draw_mask, zero_mask; \
3261 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3262 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3267 vmlal.u8 pixels, pixels_r_low, d64_1; \
3268 vmlal.u8 pixels, pixels_g_low, d64_4; \
3269 vmlal.u8 pixels, pixels_b_low, d64_128; \
3271 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3272 shade_blocks_textured_modulated_store_pixels_##target(); \
3274 ldmia sp!, { r4 - r5, pc } \
3277 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3278 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3279 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3280 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3282 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3283 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3284 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3285 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3303 #undef msb_mask_high
3306 #define num_blocks r1
3307 #define mask_msb_ptr r2
3308 #define color_ptr r3
3310 #define block_ptr_load r0
3311 #define draw_mask_store_ptr r3
3312 #define draw_mask_bits_ptr r12
3313 #define draw_mask_ptr r12
3314 #define pixel_store_ptr r14
3316 #define fb_ptr_cmp r4
3319 #define fb_ptr_next r14
3323 #define test_mask q0
3325 #define draw_mask q2
3326 #define zero_mask q3
3327 #define draw_mask_combined q4
3328 #define fb_pixels q5
3329 #define fb_pixels_next q6
3332 #define draw_mask_low d4
3333 #define draw_mask_high d5
3334 #define msb_mask_low d14
3335 #define msb_mask_high d15
3338 function(shade_blocks_textured_unmodulated_indirect)
3339 str r14, [ sp, #-4 ]
3340 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3342 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3343 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3345 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3346 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3349 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3351 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3352 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3353 [ draw_mask_bits_ptr, :16 ], c_64
3354 vceq.u16 zero_mask, pixels, #0
3356 vtst.u16 draw_mask, draw_mask, test_mask
3357 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3359 subs num_blocks, num_blocks, #1
3363 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3364 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3366 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3367 [ draw_mask_bits_ptr, :16 ], c_64
3368 vceq.u16 zero_mask, pixels, #0
3370 vtst.u16 draw_mask, draw_mask, test_mask
3371 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3373 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3374 subs num_blocks, num_blocks, #1
3379 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3380 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3387 function(shade_blocks_textured_unmodulated_direct)
3388 stmdb sp!, { r4, r14 }
3389 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3391 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3392 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3394 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3397 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3398 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3400 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3401 [ draw_mask_bits_ptr, :16 ], c_64
3402 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3404 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3405 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3406 vceq.u16 zero_mask, pixels, #0
3407 vtst.u16 draw_mask, draw_mask, test_mask
3409 subs num_blocks, num_blocks, #1
3413 mov fb_ptr, fb_ptr_next
3414 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3416 vorr.u16 pixels, pixels, msb_mask
3418 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3419 vmov fb_pixels, fb_pixels_next
3421 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3422 [ draw_mask_bits_ptr, :16 ], c_64
3423 vbif.u16 fb_pixels, pixels, draw_mask_combined
3425 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3426 pld [ fb_ptr_next, #64 ]
3428 add fb_ptr_cmp, fb_ptr_cmp, #14
3429 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3434 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3435 vceq.u16 zero_mask, pixels, #0
3437 vst1.u16 { fb_pixels }, [ fb_ptr ]
3438 vtst.u16 draw_mask, draw_mask, test_mask
3441 subs num_blocks, num_blocks, #1
3445 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3446 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3448 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3450 ldmia sp!, { r4, pc }
3453 vst1.u16 { fb_pixels }, [ fb_ptr ]
3454 vceq.u16 zero_mask, pixels, #0
3456 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3457 vtst.u16 draw_mask, draw_mask, test_mask
3462 function(shade_blocks_unshaded_untextured_indirect)
3467 function(shade_blocks_unshaded_untextured_direct)
3468 stmdb sp!, { r4, r14 }
3469 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3471 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3472 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3474 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3475 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3477 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3478 vld1.u16 { pixels }, [ color_ptr, :128 ]
3481 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3483 vorr.u16 pixels, pixels, msb_mask
3484 subs num_blocks, num_blocks, #1
3486 ldr fb_ptr_next, [ block_ptr_load ], #64
3488 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3492 vmov fb_pixels, fb_pixels_next
3493 mov fb_ptr, fb_ptr_next
3494 ldr fb_ptr_next, [ block_ptr_load ], #64
3496 vbif.u16 fb_pixels, pixels, draw_mask
3497 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3499 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3500 add fb_ptr_cmp, fb_ptr_cmp, #14
3504 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3505 vst1.u16 { fb_pixels }, [ fb_ptr ]
3508 subs num_blocks, num_blocks, #1
3512 vbif.u16 fb_pixels_next, pixels, draw_mask
3513 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3515 ldmia sp!, { r4, pc }
3518 vst1.u16 { fb_pixels }, [ fb_ptr ]
3519 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3523 #undef draw_mask_ptr
3530 #define num_blocks r1
3531 #define msb_mask_ptr r2
3532 #define pixel_ptr r3
3533 #define draw_mask_ptr r0
3536 #define fb_ptr_next r14
3537 #define fb_ptr_cmp r4
3545 #undef msb_mask_high
3546 #undef draw_mask_next
3549 #undef fb_pixels_next
3552 #define draw_mask q1
3554 #define fb_pixels q3
3555 #define blend_pixels q4
3556 #define pixels_no_msb q5
3557 #define blend_mask q6
3558 #define fb_pixels_no_msb q7
3559 #define d128_0x8000 q8
3560 #define d128_0x0421 q9
3561 #define fb_pixels_next q10
3562 #define blend_pixels_next q11
3563 #define pixels_next q12
3564 #define draw_mask_next q13
3565 #define write_mask q14
3567 #define pixels_rb q5
3568 #define pixels_mg q7
3570 #define d128_0x7C1F q8
3571 #define d128_0x03E0 q9
3572 #define fb_pixels_rb q10
3573 #define fb_pixels_g q11
3574 #define fb_pixels_masked q11
3575 #define d128_0x83E0 q15
3576 #define pixels_fourth q7
3577 #define d128_0x1C07 q12
3578 #define d128_0x00E0 q13
3579 #define d128_0x80E0 q13
3581 #define msb_mask_low d0
3582 #define msb_mask_high d1
3584 #define blend_blocks_average_set_blend_mask_textured(source) \
3585 vclt.s16 blend_mask, source, #0 \
3587 #define blend_blocks_average_set_stp_bit_textured() \
3588 vorr.u16 blend_pixels, #0x8000 \
3590 #define blend_blocks_average_combine_textured(source) \
3591 vbif.u16 blend_pixels, source, blend_mask \
3593 #define blend_blocks_average_set_blend_mask_untextured(source) \
3595 #define blend_blocks_average_set_stp_bit_untextured() \
3597 #define blend_blocks_average_combine_untextured(source) \
3599 #define blend_blocks_average_mask_set_on() \
3600 vclt.s16 write_mask, fb_pixels_next, #0 \
3602 #define blend_blocks_average_mask_copy_on() \
3603 vorr.u16 draw_mask, draw_mask_next, write_mask \
3605 #define blend_blocks_average_mask_copy_b_on() \
3606 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3608 #define blend_blocks_average_mask_set_off() \
3610 #define blend_blocks_average_mask_copy_off() \
3611 vmov draw_mask, draw_mask_next \
3613 #define blend_blocks_average_mask_copy_b_off() \
3615 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3618 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3619 stmdb sp!, { r4, r14 }; \
3620 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3621 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3623 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3624 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3626 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3629 vmov.u16 d128_0x8000, #0x8000; \
3630 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3631 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3633 vmov.u16 d128_0x0421, #0x0400; \
3634 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3636 vorr.u16 d128_0x0421, #0x0021; \
3637 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3639 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3640 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3641 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3642 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3643 blend_blocks_average_mask_set_##mask_evaluate(); \
3644 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3646 subs num_blocks, num_blocks, #1; \
3650 mov fb_ptr, fb_ptr_next; \
3651 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3653 vmov pixels, pixels_next; \
3654 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3656 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3658 blend_blocks_average_mask_copy_##mask_evaluate(); \
3659 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3661 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3662 blend_blocks_average_set_stp_bit_##texturing(); \
3663 vmov fb_pixels, fb_pixels_next; \
3664 blend_blocks_average_combine_##texturing(pixels); \
3666 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3667 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3668 cmp fb_ptr_cmp, #28; \
3671 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3672 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3674 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3675 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3677 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3678 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3680 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3681 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3682 blend_blocks_average_mask_set_##mask_evaluate(); \
3683 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3686 subs num_blocks, num_blocks, #1; \
3690 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3691 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3693 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3694 blend_blocks_average_set_stp_bit_##texturing(); \
3695 blend_blocks_average_combine_##texturing(pixels_next); \
3697 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3698 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3699 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3701 ldmia sp!, { r4, pc }; \
3704 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3705 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3706 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3708 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3709 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3710 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3711 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3712 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3713 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3717 blend_blocks_average_builder(textured, off)
3718 blend_blocks_average_builder(untextured, off)
3719 blend_blocks_average_builder(textured, on)
3720 blend_blocks_average_builder(untextured, on)
3723 #define blend_blocks_add_mask_set_on() \
3724 vclt.s16 write_mask, fb_pixels, #0 \
3726 #define blend_blocks_add_mask_copy_on() \
3727 vorr.u16 draw_mask, draw_mask, write_mask \
3729 #define blend_blocks_add_mask_set_off() \
3731 #define blend_blocks_add_mask_copy_off() \
3734 #define blend_blocks_add_textured_builder(mask_evaluate) \
3737 function(blend_blocks_textured_add_##mask_evaluate) \
3738 stmdb sp!, { r4, r14 }; \
3739 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3740 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3742 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3743 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3745 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3748 vmov.u16 d128_0x7C1F, #0x7C00; \
3749 vmov.u16 d128_0x03E0, #0x0300; \
3750 vmov.u16 d128_0x83E0, #0x8000; \
3751 vorr.u16 d128_0x03E0, #0x00E0; \
3752 vorr.u16 d128_0x7C1F, #0x001F; \
3753 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3755 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3756 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3757 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3758 vclt.s16 blend_mask, pixels, #0; \
3759 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3760 blend_blocks_add_mask_set_##mask_evaluate(); \
3761 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3763 blend_blocks_add_mask_copy_##mask_evaluate(); \
3764 vorr.u16 pixels, pixels, msb_mask; \
3765 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3766 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3767 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3768 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3769 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3770 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3771 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3772 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3774 subs num_blocks, num_blocks, #1; \
3778 mov fb_ptr, fb_ptr_next; \
3780 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3782 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3783 vclt.s16 blend_mask, pixels, #0; \
3785 vorr.u16 pixels, pixels, msb_mask; \
3786 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3787 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3789 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3790 pld [ fb_ptr_next, #64 ]; \
3792 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3793 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3795 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3796 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3798 cmp fb_ptr_cmp, #28; \
3801 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3802 blend_blocks_add_mask_set_##mask_evaluate(); \
3803 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3804 blend_blocks_add_mask_copy_##mask_evaluate(); \
3805 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3806 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3807 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3810 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3811 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3812 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3813 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3814 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3816 subs num_blocks, num_blocks, #1; \
3820 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3821 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3822 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3824 ldmia sp!, { r4, pc }; \
3827 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3828 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3830 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3831 blend_blocks_add_mask_set_##mask_evaluate(); \
3832 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3833 blend_blocks_add_mask_copy_##mask_evaluate(); \
3834 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3838 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3841 function(blend_blocks_untextured_add_##mask_evaluate) \
3842 stmdb sp!, { r4, r14 }; \
3843 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3844 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3846 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3847 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3849 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3852 vmov.u16 d128_0x7C1F, #0x7C00; \
3853 vmov.u16 d128_0x03E0, #0x0300; \
3854 vorr.u16 d128_0x7C1F, #0x001F; \
3855 vorr.u16 d128_0x03E0, #0x00E0; \
3857 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3858 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3859 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3860 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3861 blend_blocks_add_mask_set_##mask_evaluate(); \
3862 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3864 blend_blocks_add_mask_copy_##mask_evaluate(); \
3865 vand.u16 pixels_g, pixels, d128_0x03E0; \
3866 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3867 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3868 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3869 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3870 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3871 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3873 subs num_blocks, num_blocks, #1; \
3877 mov fb_ptr, fb_ptr_next; \
3879 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3881 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3883 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3884 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3885 vand.u16 pixels_g, pixels, d128_0x03E0; \
3887 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3888 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3890 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3891 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3892 cmp fb_ptr_cmp, #28; \
3895 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3896 blend_blocks_add_mask_set_##mask_evaluate(); \
3897 blend_blocks_add_mask_copy_##mask_evaluate(); \
3898 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3899 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3900 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3903 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3904 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3905 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3906 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3907 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3909 subs num_blocks, num_blocks, #1; \
3913 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3914 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3915 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3916 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3918 ldmia sp!, { r4, pc }; \
3921 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3922 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3924 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3925 blend_blocks_add_mask_set_##mask_evaluate(); \
3926 blend_blocks_add_mask_copy_##mask_evaluate(); \
3927 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3931 blend_blocks_add_textured_builder(off)
3932 blend_blocks_add_textured_builder(on)
3933 blend_blocks_add_untextured_builder(off)
3934 blend_blocks_add_untextured_builder(on)
3936 #define blend_blocks_subtract_set_blend_mask_textured() \
3937 vclt.s16 blend_mask, pixels_next, #0 \
3939 #define blend_blocks_subtract_combine_textured() \
3940 vbif.u16 blend_pixels, pixels, blend_mask \
3942 #define blend_blocks_subtract_set_stb_textured() \
3943 vorr.u16 blend_pixels, #0x8000 \
3945 #define blend_blocks_subtract_msb_mask_textured() \
3946 vorr.u16 pixels, pixels_next, msb_mask \
3948 #define blend_blocks_subtract_set_blend_mask_untextured() \
3950 #define blend_blocks_subtract_combine_untextured() \
3952 #define blend_blocks_subtract_set_stb_untextured() \
3953 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3955 #define blend_blocks_subtract_msb_mask_untextured() \
3958 #define blend_blocks_subtract_mask_set_on() \
3959 vclt.s16 write_mask, fb_pixels, #0 \
3961 #define blend_blocks_subtract_mask_copy_on() \
3962 vorr.u16 draw_mask, draw_mask_next, write_mask \
3964 #define blend_blocks_subtract_mask_set_off() \
3966 #define blend_blocks_subtract_mask_copy_off() \
3967 vmov draw_mask, draw_mask_next \
3970 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
3973 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
3974 stmdb sp!, { r4, r14 }; \
3975 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3976 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3978 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3979 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3981 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3984 vmov.u16 d128_0x7C1F, #0x7C00; \
3985 vmov.u16 d128_0x03E0, #0x0300; \
3986 vorr.u16 d128_0x7C1F, #0x001F; \
3987 vorr.u16 d128_0x03E0, #0x00E0; \
3989 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3990 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3991 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3992 blend_blocks_subtract_set_blend_mask_##texturing(); \
3993 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3994 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3995 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3997 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3998 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3999 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4000 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4001 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4003 subs num_blocks, num_blocks, #1; \
4007 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4008 mov fb_ptr, fb_ptr_next; \
4009 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4011 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
4012 blend_blocks_subtract_msb_mask_##texturing(); \
4014 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
4015 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4016 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4017 blend_blocks_subtract_set_stb_##texturing(); \
4018 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4019 blend_blocks_subtract_combine_##texturing(); \
4020 blend_blocks_subtract_set_blend_mask_##texturing(); \
4021 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4023 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4024 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4025 cmp fb_ptr_cmp, #28; \
4028 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4029 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4030 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4031 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4032 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4033 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4034 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4037 subs num_blocks, num_blocks, #1; \
4041 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4043 blend_blocks_subtract_msb_mask_##texturing(); \
4044 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4045 blend_blocks_subtract_set_stb_##texturing(); \
4046 blend_blocks_subtract_combine_##texturing(); \
4047 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4048 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4050 ldmia sp!, { r4, pc }; \
4053 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4054 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4055 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4056 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4057 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4058 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4059 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4063 blend_blocks_subtract_builder(textured, off)
4064 blend_blocks_subtract_builder(textured, on)
4065 blend_blocks_subtract_builder(untextured, off)
4066 blend_blocks_subtract_builder(untextured, on)
4069 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4072 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4073 stmdb sp!, { r4, r14 }; \
4074 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4075 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4077 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4078 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4080 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4083 vmov.u16 d128_0x7C1F, #0x7C00; \
4084 vmov.u16 d128_0x03E0, #0x0300; \
4085 vmov.u16 d128_0x1C07, #0x1C00; \
4086 vmov.u16 d128_0x00E0, #0x00E0; \
4087 vorr.u16 d128_0x7C1F, #0x001F; \
4088 vorr.u16 d128_0x03E0, #0x00E0; \
4089 vorr.u16 d128_0x1C07, #0x0007; \
4091 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4092 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4093 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4094 vclt.s16 blend_mask, pixels, #0; \
4095 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4096 blend_blocks_add_mask_set_##mask_evaluate(); \
4097 vshr.s16 pixels_fourth, pixels, #2; \
4098 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4100 blend_blocks_add_mask_copy_##mask_evaluate(); \
4101 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4102 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4103 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4104 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4105 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4106 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4107 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4109 subs num_blocks, num_blocks, #1; \
4113 mov fb_ptr, fb_ptr_next; \
4114 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4116 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4117 vbif.u16 blend_pixels, pixels, blend_mask; \
4119 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4120 vclt.s16 blend_mask, pixels, #0; \
4121 vshr.s16 pixels_fourth, pixels, #2; \
4122 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4123 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4125 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4126 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4128 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4129 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4130 cmp fb_ptr_cmp, #28; \
4133 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4134 blend_blocks_add_mask_set_##mask_evaluate(); \
4135 blend_blocks_add_mask_copy_##mask_evaluate(); \
4136 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4137 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4138 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4141 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4142 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4143 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4144 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4145 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4147 subs num_blocks, num_blocks, #1; \
4151 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4152 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4153 vbif.u16 blend_pixels, pixels, blend_mask; \
4154 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4155 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4157 ldmia sp!, { r4, pc }; \
4160 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4161 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4163 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4164 blend_blocks_add_mask_set_##mask_evaluate(); \
4165 blend_blocks_add_mask_copy_##mask_evaluate(); \
4166 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4171 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4174 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4175 stmdb sp!, { r4, r14 }; \
4176 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4177 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4179 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4180 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4182 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4185 vmov.u16 d128_0x7C1F, #0x7C00; \
4186 vmov.u16 d128_0x03E0, #0x0300; \
4187 vmov.u16 d128_0x1C07, #0x1C00; \
4188 vmov.u16 d128_0x00E0, #0x00E0; \
4189 vorr.u16 d128_0x7C1F, #0x001F; \
4190 vorr.u16 d128_0x03E0, #0x00E0; \
4191 vorr.u16 d128_0x1C07, #0x0007; \
4193 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4194 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4195 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4196 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4197 blend_blocks_add_mask_set_##mask_evaluate(); \
4198 vshr.s16 pixels_fourth, pixels, #2; \
4199 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4201 blend_blocks_add_mask_copy_##mask_evaluate(); \
4202 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4203 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4204 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4205 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4206 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4207 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4208 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4210 subs num_blocks, num_blocks, #1; \
4214 mov fb_ptr, fb_ptr_next; \
4215 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4217 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4219 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4220 vshr.s16 pixels_fourth, pixels, #2; \
4221 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4222 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4224 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4225 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4227 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4228 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4229 cmp fb_ptr_cmp, #28; \
4232 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4233 blend_blocks_add_mask_set_##mask_evaluate(); \
4234 blend_blocks_add_mask_copy_##mask_evaluate(); \
4235 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4236 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4237 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4240 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4241 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4242 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4243 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4244 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4246 subs num_blocks, num_blocks, #1; \
4250 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4251 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4252 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4253 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4255 ldmia sp!, { r4, pc }; \
4258 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4259 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4261 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4262 blend_blocks_add_mask_set_##mask_evaluate(); \
4263 blend_blocks_add_mask_copy_##mask_evaluate(); \
4264 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4268 blend_blocks_add_fourth_textured_builder(off)
4269 blend_blocks_add_fourth_textured_builder(on)
4270 blend_blocks_add_fourth_untextured_builder(off)
4271 blend_blocks_add_fourth_untextured_builder(on)
4273 // TODO: Optimize this more. Need a scene that actually uses it for
4278 function(blend_blocks_textured_unblended_on)
4279 stmdb sp!, { r4, r14 }
4280 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4281 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
4283 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4284 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
4286 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4289 ldr fb_ptr, [ pixel_ptr, #28 ]
4290 vld1.u16 { fb_pixels }, [ fb_ptr ]
4291 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4292 vclt.s16 write_mask, fb_pixels, #0
4293 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4295 subs num_blocks, num_blocks, #1
4299 vorr.u16 pixels, pixels, msb_mask
4300 vorr.u16 draw_mask, draw_mask, write_mask
4301 vbif.u16 fb_pixels, pixels, draw_mask
4302 vst1.u16 { fb_pixels }, [ fb_ptr ]
4304 ldr fb_ptr, [ pixel_ptr, #28 ]
4305 vld1.u16 { fb_pixels }, [ fb_ptr ]
4306 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4307 vclt.s16 write_mask, fb_pixels, #0
4308 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4310 subs num_blocks, num_blocks, #1
4314 vorr.u16 pixels, pixels, msb_mask
4315 vorr.u16 draw_mask, draw_mask, write_mask
4316 vbif.u16 fb_pixels, pixels, draw_mask
4317 vst1.u16 { fb_pixels }, [ fb_ptr ]
4319 ldmia sp!, { r4, pc }
4322 function(blend_blocks_textured_unblended_off)
4332 vld1.u32 { u_whole_8, v_whole_8 }, [ r1, :128 ], r3
4352 #define num_width r12
4362 function(render_block_fill_body)
4363 vdup.u16 colors_a, color
4366 vmov colors_b, colors_a
4367 sub pitch, pitch, width, lsl #1
4369 mov num_width, width
4372 vst1.u32 { colors_a, colors_b }, [ vram_ptr, :256 ]!
4374 subs num_width, num_width, #16
4377 add vram_ptr, vram_ptr, pitch
4378 mov num_width, width
4380 subs height, height, #1
4394 #undef dirty_textures_mask
4396 #undef current_texture_mask
4407 #define offset_u_right r10
4408 #define width_rounded r11
4409 #define height_rounded r12
4411 #define texture_offset_base r1
4412 #define tile_width r2
4413 #define tile_height r3
4414 #define num_blocks r4
4416 #define sub_tile_height r6
4418 #define texture_mask r8
4419 #define column_data r9
4420 #define texture_offset r10
4421 #define tiles_remaining r11
4422 #define fb_ptr_advance_column r12
4423 #define texture_block_ptr r14
4427 #define texture_page_ptr r3
4428 #define left_block_mask r4
4429 #define right_block_mask r5
4430 #define texture_mask_rev r10
4431 #define control_mask r11
4433 #define dirty_textures_mask r4
4435 #define current_texture_mask r6
4449 #define draw_masks_fb_ptrs q1
4451 #define draw_mask_fb_ptr_left d2
4452 #define draw_mask_fb_ptr_right d3
4454 #define draw_mask_fb_ptr_left_a d2
4455 #define draw_mask_fb_ptr_left_b d3
4456 #define draw_mask_fb_ptr_right_a d10
4457 #define draw_mask_fb_ptr_right_b d11
4458 #define draw_masks_fb_ptrs2 q5
4460 #define clut_low_a d4
4461 #define clut_low_b d5
4462 #define clut_high_a d6
4463 #define clut_high_b d7
4465 #define block_masks d8
4466 #define block_masks_shifted d9
4471 #define texels_low d12
4472 #define texels_high d13
4474 #define texels_wide_low d14
4475 #define texels_wide_high d15
4476 #define texels_wide q7
4479 setup_sprite_flush_blocks:
4482 stmdb sp!, { r0 - r3, r12, r14 }
4483 bl flush_render_block_buffer
4484 ldmia sp!, { r0 - r3, r12, r14 }
4488 add block, psx_gpu, #psx_gpu_blocks_offset
4492 setup_sprite_update_texture_4bpp_cache:
4493 stmdb sp!, { r0 - r3, r14 }
4494 bl update_texture_4bpp_cache
4495 ldmia sp!, { r0 - r3, pc }
4498 setup_sprite_update_texture_8bpp_cache:
4499 stmdb sp!, { r0 - r3, r14 }
4500 bl update_texture_8bpp_cache
4501 ldmia sp!, { r0 - r3, pc }
4504 #define setup_sprite_tiled_initialize_4bpp() \
4505 ldr dirty_textures_mask, \
4506 [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]; \
4507 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4509 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4510 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4512 tst current_texture_mask, dirty_textures_mask; \
4513 vuzp.u8 clut_a, clut_b; \
4515 blne setup_sprite_update_texture_4bpp_cache \
4517 #define setup_sprite_tiled_initialize_8bpp() \
4518 ldr dirty_textures_mask, \
4519 [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]; \
4520 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4522 tst current_texture_mask, dirty_textures_mask; \
4523 blne setup_sprite_update_texture_8bpp_cache \
4526 #define setup_sprite_block_count_single() \
4529 #define setup_sprite_block_count_double() \
4530 sub_tile_height, lsl #1 \
4532 #define setup_sprite_tile_add_blocks(type) \
4533 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4534 cmp num_blocks, #MAX_BLOCKS; \
4536 movgt num_blocks, setup_sprite_block_count_##type(); \
4537 blgt setup_sprite_flush_blocks \
4540 #define setup_sprite_tile_full_4bpp(edge) \
4541 setup_sprite_tile_add_blocks(double); \
4544 and texture_block_ptr, texture_offset, texture_mask; \
4545 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4548 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4549 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4551 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4552 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4554 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4555 add texture_block_ptr, texture_offset, #8; \
4557 and texture_block_ptr, texture_block_ptr, texture_mask; \
4558 add block, block, #40; \
4560 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4561 add fb_ptr, fb_ptr, #16; \
4563 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4564 add block, block, #24; \
4566 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4567 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4570 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4571 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4573 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4574 add block, block, #40; \
4576 add texture_offset, texture_offset, #0x10; \
4577 add fb_ptr, fb_ptr, #(2048 - 16); \
4579 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4580 add block, block, #24; \
4582 subs sub_tile_height, sub_tile_height, #1; \
4585 add texture_offset, texture_offset, #0xF00; \
4586 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4589 #define setup_sprite_tile_half_4bpp(edge) \
4590 setup_sprite_tile_add_blocks(single); \
4593 and texture_block_ptr, texture_offset, texture_mask; \
4594 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4597 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4598 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4600 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4601 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4603 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4604 add block, block, #40; \
4606 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4607 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4609 add block, block, #24; \
4610 add texture_offset, texture_offset, #0x10; \
4612 add fb_ptr, fb_ptr, #2048; \
4613 subs sub_tile_height, sub_tile_height, #1; \
4617 add texture_offset, texture_offset, #0xF00; \
4618 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4621 #define setup_sprite_tile_full_8bpp(edge) \
4622 setup_sprite_tile_add_blocks(double); \
4623 add block, block, #16; \
4626 and texture_block_ptr, texture_offset, texture_mask; \
4627 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4630 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4631 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4633 add texture_block_ptr, texture_offset, #8; \
4634 vst1.u32 { texels }, [ block, :64 ]; \
4636 and texture_block_ptr, texture_block_ptr, texture_mask; \
4637 add block, block, #24; \
4639 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4641 add fb_ptr, fb_ptr, #16; \
4642 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4644 add block, block, #40; \
4645 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4648 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4649 vst1.u32 { texels }, [ block, :64 ]; \
4650 add block, block, #24; \
4652 add texture_offset, texture_offset, #0x10; \
4653 add fb_ptr, fb_ptr, #(2048 - 16); \
4655 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4656 add block, block, #40; \
4658 subs sub_tile_height, sub_tile_height, #1; \
4661 sub block, block, #16; \
4662 add texture_offset, texture_offset, #0xF00; \
4663 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4666 #define setup_sprite_tile_half_8bpp(edge) \
4667 setup_sprite_tile_add_blocks(single); \
4668 add block, block, #16; \
4671 and texture_block_ptr, texture_offset, texture_mask; \
4672 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4675 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4676 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4678 vst1.u32 { texels }, [ block, :64 ]; \
4679 add block, block, #24; \
4681 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4682 add block, block, #40; \
4684 add texture_offset, texture_offset, #0x10; \
4685 add fb_ptr, fb_ptr, #2048; \
4687 subs sub_tile_height, sub_tile_height, #1; \
4690 sub block, block, #16; \
4691 add texture_offset, texture_offset, #0xF00; \
4692 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4695 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4696 add texture_offset, texture_offset_base, #8; \
4697 add fb_ptr, fb_ptr, #16 \
4699 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4700 mov texture_offset, texture_offset_base \
4702 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4703 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4705 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4706 mov texture_offset, texture_offset_base \
4708 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4709 sub fb_ptr, fb_ptr, #16 \
4711 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4713 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4714 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4716 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4719 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \
4721 mov sub_tile_height, column_data; \
4722 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4723 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4724 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4726 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \
4728 and sub_tile_height, column_data, #0xFF; \
4729 mov tiles_remaining, column_data, lsr #16; \
4730 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4731 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4733 subs tiles_remaining, tiles_remaining, #1; \
4737 mov sub_tile_height, #16; \
4738 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4739 subs tiles_remaining, tiles_remaining, #1; \
4743 uxtb sub_tile_height, column_data, ror #8; \
4744 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4745 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4748 #define setup_sprite_column_data_single() \
4749 mov column_data, height; \
4750 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] \
4752 #define setup_sprite_column_data_multi() \
4753 and height_rounded, height_rounded, #0xF; \
4754 rsb column_data, offset_v, #16; \
4756 add height_rounded, height_rounded, #1; \
4757 sub tile_height, tile_height, #1; \
4759 orr column_data, column_data, tile_height, lsl #16; \
4760 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]; \
4762 orr column_data, column_data, height_rounded, lsl #8 \
4764 #define setup_sprite_setup_left_draw_mask_fb_ptr() \
4765 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4766 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4768 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column() \
4769 mov fb_ptr_advance_column, #32; \
4770 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4772 sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11; \
4773 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4775 #define setup_sprite_setup_right_draw_mask_fb_ptr() \
4776 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4777 vdup.u8 draw_mask_fb_ptr_right, block_masks[5] \
4779 #define setup_sprite_tile_column_width_single(tm, multi_height, edge_mode, \
4781 setup_sprite_##tm##_single_##multi_height##_##edge_mode##_##edge##x4mode: \
4782 setup_sprite_column_data_##multi_height(); \
4783 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4784 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4785 setup_sprite_setup_left_draw_mask_fb_ptr##x4mode(); \
4787 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, tm, x4mode); \
4788 ldmia sp!, { r4 - r11, pc } \
4790 #define setup_sprite_tiled_advance_column() \
4791 add texture_offset_base, texture_offset_base, #0x100; \
4792 tst texture_offset_base, #0xF00; \
4793 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4795 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4796 right_mode, x4mode) \
4797 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode##x4mode:\
4798 setup_sprite_column_data_##multi_height(); \
4800 setup_sprite_setup_left_draw_mask_fb_ptr_advance_column##x4mode(); \
4802 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm, x4mode);\
4804 subs tile_width, tile_width, #2; \
4805 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4809 vmov.u8 draw_masks_fb_ptrs, #0; \
4810 vmov.u8 draw_masks_fb_ptrs2, #0; \
4813 setup_sprite_tiled_advance_column(); \
4814 setup_sprite_tile_column_height_##multi_height(full, none, tm, x4mode); \
4815 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4816 subs tile_width, tile_width, #1; \
4820 setup_sprite_setup_right_draw_mask_fb_ptr##x4mode(); \
4822 setup_sprite_tiled_advance_column(); \
4823 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm, x4mode);\
4824 ldmia sp!, { r4 - r11, pc } \
4827 #define setup_sprite_offset_u_adjust() \
4829 #define setup_sprite_get_left_block_mask() \
4830 and left_block_mask, left_block_mask, #0xFF \
4832 #define setup_sprite_compare_left_block_mask() \
4833 cmp left_block_mask, #0xFF \
4835 #define setup_sprite_get_right_block_mask() \
4836 uxtb right_block_mask, right_block_mask, ror #8 \
4838 #define setup_sprite_compare_right_block_mask() \
4839 cmp right_block_mask, #0xFF \
4844 #define fb_ptr2 column_data
4846 #define setup_sprite_offset_u_adjust_4x() \
4847 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4848 lsl offset_u_right, #1; \
4850 add offset_u_right, #1 \
4852 #define setup_sprite_get_left_block_mask_4x() \
4853 sxth left_block_mask, left_block_mask \
4855 #define setup_sprite_compare_left_block_mask_4x() \
4856 cmp left_block_mask, #0xFFFFFFFF \
4858 #define setup_sprite_get_right_block_mask_4x() \
4859 sxth right_block_mask, right_block_mask, ror #16 \
4861 #define setup_sprite_compare_right_block_mask_4x() \
4862 cmp right_block_mask, #0xFFFFFFFF \
4865 #define widen_texels_16bpp(texels_) \
4866 vmov texels_wide_low, texels_; \
4867 vmov texels_wide_high, texels_; \
4868 vzip.16 texels_wide_low, texels_wide_high \
4870 #define widen_texels_8bpp(texels_) \
4871 vmov texels_wide_low, texels_; \
4872 vmov texels_wide_high, texels_; \
4873 vzip.8 texels_wide_low, texels_wide_high \
4875 #define write_block_16bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4876 vst1.u32 { texels_ }, [ block_, :128 ]; \
4877 add block_, block_, #40; \
4879 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4880 vst1.u32 { draw_mask_fb_ptr_ }, [ block_, :64 ]; \
4881 add block_, block_, #24 \
4883 /* assumes 16-byte offset already added to block_ */
4884 #define write_block_8bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4885 vst1.u32 { texels_ }, [ block_, :64 ]; \
4886 add block_, block_, #24; \
4888 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4889 vst1.u32 { draw_mask_fb_ptr_ }, [ block_, :64 ]; \
4890 add block_, block_, #40 \
4892 #define do_texture_block_16bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4893 draw_mask_fb_ptr_b_) \
4894 widen_texels_16bpp(texels_low); \
4895 add fb_ptr_tmp, fb_ptr, #1024*2; \
4897 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr); \
4899 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4900 widen_texels_16bpp(texels_high); \
4902 add fb_ptr_tmp, fb_ptr, #8*2; \
4903 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4905 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4906 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4908 #define do_texture_block_8bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4909 draw_mask_fb_ptr_b_) \
4910 widen_texels_8bpp(texels); \
4911 add fb_ptr_tmp, fb_ptr, #1024*2; \
4913 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr); \
4914 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4916 add fb_ptr_tmp, fb_ptr, #8*2; \
4917 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4919 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4920 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4923 #define setup_sprite_tiled_initialize_4bpp_4x() \
4924 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4925 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4927 vuzp.u8 clut_a, clut_b \
4929 #define setup_sprite_tiled_initialize_8bpp_4x() \
4932 #define setup_sprite_block_count_single_4x() \
4933 sub_tile_height, lsl #2 \
4935 #define setup_sprite_block_count_double_4x() \
4936 sub_tile_height, lsl #(1+2) \
4938 #define setup_sprite_tile_full_4bpp_4x(edge) \
4939 setup_sprite_tile_add_blocks(double_4x); \
4940 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4943 and texture_block_ptr, texture_offset, texture_mask; \
4946 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4947 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4949 add texture_block_ptr, texture_offset, #8; \
4950 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4952 and texture_block_ptr, texture_block_ptr, texture_mask; \
4953 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4955 vzip.8 texels_low, texels_high; \
4956 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
4957 draw_mask_fb_ptr_left_b); \
4959 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4960 pld [ fb_ptr, #2048 ]; \
4962 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4963 add fb_ptr, fb_ptr, #16*2; \
4965 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4966 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4968 vzip.8 texels_low, texels_high; \
4969 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
4970 draw_mask_fb_ptr_right_b); \
4972 add texture_offset, texture_offset, #0x10; \
4973 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
4975 subs sub_tile_height, sub_tile_height, #1; \
4978 ldr column_data, [sp], #8; /* fb_ptr2 */ \
4979 add texture_offset, texture_offset, #0xF00; \
4980 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4983 #define setup_sprite_tile_half_4bpp_4x(edge) \
4984 setup_sprite_tile_add_blocks(single_4x); \
4985 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4988 and texture_block_ptr, texture_offset, texture_mask; \
4991 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4992 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4994 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4995 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4997 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4998 add texture_offset, texture_offset, #0x10; \
5000 vzip.8 texels_low, texels_high; \
5001 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5002 draw_mask_fb_ptr_##edge##_b); \
5004 pld [ fb_ptr, #2048 ]; \
5005 add fb_ptr, fb_ptr, #2048 * 2; \
5007 subs sub_tile_height, sub_tile_height, #1; \
5010 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5011 add texture_offset, texture_offset, #0xF00; \
5012 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
5015 #define setup_sprite_tile_full_8bpp_4x(edge) \
5016 setup_sprite_tile_add_blocks(double_4x); \
5017 add block, block, #16; \
5018 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5021 and texture_block_ptr, texture_offset, texture_mask; \
5024 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5025 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
5027 add texture_block_ptr, texture_offset, #8; \
5028 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
5029 draw_mask_fb_ptr_left_b); \
5031 pld [ fb_ptr, #2048 ]; \
5032 and texture_block_ptr, texture_block_ptr, texture_mask; \
5034 add fb_ptr, fb_ptr, #16*2; \
5035 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5037 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
5039 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
5040 draw_mask_fb_ptr_right_b); \
5042 add texture_offset, texture_offset, #0x10; \
5043 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
5045 subs sub_tile_height, sub_tile_height, #1; \
5048 sub block, block, #16; \
5049 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5050 add texture_offset, texture_offset, #0xF00; \
5051 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
5054 #define setup_sprite_tile_half_8bpp_4x(edge) \
5055 setup_sprite_tile_add_blocks(single_4x); \
5056 add block, block, #16; \
5057 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5060 and texture_block_ptr, texture_offset, texture_mask; \
5063 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5064 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
5066 pld [ fb_ptr, #2048 ]; \
5067 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5068 draw_mask_fb_ptr_##edge##_b); \
5070 add texture_offset, texture_offset, #0x10; \
5071 add fb_ptr, fb_ptr, #2048 * 2; \
5073 subs sub_tile_height, sub_tile_height, #1; \
5076 sub block, block, #16; \
5077 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5078 add texture_offset, texture_offset, #0xF00; \
5079 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
5082 #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \
5083 add texture_offset, texture_offset_base, #8; \
5084 add fb_ptr, fb_ptr, #16 * 2 \
5086 #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \
5087 mov texture_offset, texture_offset_base \
5089 #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \
5090 setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \
5092 #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \
5093 mov texture_offset, texture_offset_base \
5095 #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \
5096 sub fb_ptr, fb_ptr, #16 * 2 \
5098 #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \
5100 #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \
5101 setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \
5103 #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \
5106 #define setup_sprite_setup_left_draw_mask_fb_ptr_4x() \
5107 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5108 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5109 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5110 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5112 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column_4x() \
5113 mov fb_ptr_advance_column, #32 * 2; \
5114 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5115 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5116 sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11 + 1; \
5117 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5118 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5120 #define setup_sprite_setup_right_draw_mask_fb_ptr_4x() \
5121 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[4]; \
5122 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[5]; \
5123 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[6]; \
5124 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[7] \
5132 // [ sp + 4 ]: width
5133 // [ sp + 8 ]: height
5134 // [ sp + 12 ]: color (unused)
5136 #define setup_sprite_tiled_builder(texture_mode, x4mode) \
5138 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \
5140 setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \
5142 setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \
5144 setup_sprite_tile_column_width_single(texture_mode, single, full, none, \
5146 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \
5148 setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \
5150 setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \
5152 setup_sprite_tile_column_width_single(texture_mode, single, half, right, \
5154 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \
5156 setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \
5158 setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \
5160 setup_sprite_tile_column_width_single(texture_mode, single, half, left, \
5162 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \
5164 setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \
5169 function(setup_sprite_##texture_mode##x4mode) \
5170 stmdb sp!, { r4 - r11, r14 }; \
5171 setup_sprite_tiled_initialize_##texture_mode##x4mode(); \
5173 ldr v, [ sp, #36 ]; \
5174 and offset_u, u, #0xF; \
5176 ldr width, [ sp, #40 ]; \
5177 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
5179 ldr height, [ sp, #44 ]; \
5180 add fb_ptr, fb_ptr, y, lsl #11; \
5182 add fb_ptr, fb_ptr, x, lsl #1; \
5183 and offset_v, v, #0xF; \
5185 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
5186 add width_rounded, offset_u, width; \
5188 add height_rounded, offset_v, height; \
5189 add width_rounded, width_rounded, #15; \
5191 add height_rounded, height_rounded, #15; \
5192 mov tile_width, width_rounded, lsr #4; \
5194 /* texture_offset_base = VH-VL-00-00 */\
5195 mov texture_offset_base, v, lsl #8; \
5196 and offset_u_right, width_rounded, #0xF; \
5198 /* texture_offset_base = VH-UH-UL-00 */\
5199 bfi texture_offset_base, u, #4, #8; \
5200 mov right_block_mask, #0xFFFFFFFE; \
5202 setup_sprite_offset_u_adjust##x4mode(); \
5204 /* texture_offset_base = VH-UH-VL-00 */\
5205 bfi texture_offset_base, v, #4, #4; \
5206 mov left_block_mask, #0xFFFFFFFF; \
5208 mov tile_height, height_rounded, lsr #4; \
5209 mvn left_block_mask, left_block_mask, lsl offset_u; \
5211 /* texture_mask = HH-HL-WH-WL */\
5212 ldrh texture_mask, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]; \
5213 mov right_block_mask, right_block_mask, lsl offset_u_right; \
5215 /* texture_mask_rev = WH-WL-HH-HL */\
5216 rev16 texture_mask_rev, texture_mask; \
5217 vmov block_masks, left_block_mask, right_block_mask; \
5219 /* texture_mask = HH-HL-HL-WL */\
5220 bfi texture_mask, texture_mask_rev, #4, #4; \
5221 /* texture_mask_rev = 00-00-00-WH */\
5222 mov texture_mask_rev, texture_mask_rev, lsr #12; \
5224 /* texture_mask = HH-WH-HL-WL */\
5225 bfi texture_mask, texture_mask_rev, #8, #4; \
5226 setup_sprite_get_left_block_mask##x4mode(); \
5228 mov control_mask, #0; \
5229 setup_sprite_compare_left_block_mask##x4mode(); \
5231 setup_sprite_get_right_block_mask##x4mode(); \
5232 orreq control_mask, control_mask, #0x4; \
5234 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
5235 setup_sprite_compare_right_block_mask##x4mode(); \
5237 orreq control_mask, control_mask, #0x8; \
5238 cmp tile_width, #1; \
5240 add block, psx_gpu, #psx_gpu_blocks_offset; \
5241 orreq control_mask, control_mask, #0x1; \
5243 cmp tile_height, #1; \
5244 add block, block, num_blocks, lsl #6; \
5246 orreq control_mask, control_mask, #0x2; \
5247 JT_OP_REL(9f, control_mask, temp); \
5248 JT_OP(ldr pc, [ pc, control_mask, lsl #2 ]); \
5252 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_full_full##x4mode); \
5253 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_full_none##x4mode); \
5254 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_full_full##x4mode); \
5255 .word JTE(9b, setup_sprite_##texture_mode##_single_single_full_none##x4mode); \
5256 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_half_full##x4mode); \
5257 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_half_right##x4mode); \
5258 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_half_full##x4mode); \
5259 .word JTE(9b, setup_sprite_##texture_mode##_single_single_half_right##x4mode);\
5260 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_full_half##x4mode); \
5261 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_half_left##x4mode); \
5262 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_full_half##x4mode); \
5263 .word JTE(9b, setup_sprite_##texture_mode##_single_single_half_left##x4mode); \
5264 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_half_half##x4mode); \
5266 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_half_half##x4mode); \
5269 setup_sprite_tiled_builder(4bpp,);
5270 setup_sprite_tiled_builder(8bpp,);
5272 #undef draw_mask_fb_ptr_left
5273 #undef draw_mask_fb_ptr_right
5275 setup_sprite_tiled_builder(4bpp, _4x);
5276 setup_sprite_tiled_builder(8bpp, _4x);
5284 #define block_ptr r0
5285 #define num_blocks r1
5287 #define texel_shift_mask r3
5288 #define block_pixels_a r4
5289 #define block_pixels_b r5
5298 #define texels_01 r6
5299 #define texels_23 r7
5300 #define texels_45 r8
5301 #define texels_67 r9
5303 function(texture_sprite_blocks_8bpp)
5304 stmdb sp!, { r4 - r11, r14 }
5305 movw texel_shift_mask, #(0xFF << 1)
5307 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5308 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
5310 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
5311 ldr block_pixels_a, [ block_ptr, #16 ]
5314 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
5315 ldr block_pixels_b, [ block_ptr, #20 ]
5317 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
5318 ldrh texel_0, [ clut_ptr, texel_0 ]
5320 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
5321 ldrh texel_1, [ clut_ptr, texel_1 ]
5323 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
5324 ldr block_pixels_a, [ block_ptr, #(64 + 16) ]
5326 ldrh texel_2, [ clut_ptr, texel_2 ]
5327 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
5329 ldrh texel_3, [ clut_ptr, texel_3 ]
5330 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
5332 ldrh texel_4, [ clut_ptr, texel_4 ]
5333 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
5335 ldrh texel_5, [ clut_ptr, texel_5 ]
5336 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
5338 ldrh texel_6, [ clut_ptr, texel_6 ]
5339 orr texels_01, texel_0, texel_1, lsl #16
5341 ldrh texel_7, [ clut_ptr, texel_7 ]
5342 orr texels_23, texel_2, texel_3, lsl #16
5344 orr texels_45, texel_4, texel_5, lsl #16
5345 str texels_01, [ block_ptr, #0 ]
5347 orr texels_67, texel_6, texel_7, lsl #16
5348 str texels_23, [ block_ptr, #4 ]
5350 subs num_blocks, num_blocks, #1
5351 str texels_45, [ block_ptr, #8 ]
5353 str texels_67, [ block_ptr, #12 ]
5354 add block_ptr, block_ptr, #64
5358 ldmia sp!, { r4 - r11, pc }
5361 #undef width_rounded
5364 #undef texture_offset
5367 #undef texels_wide_low
5368 #undef texels_wide_high
5380 #define left_offset r8
5381 #define width_rounded r9
5382 #define right_width r10
5384 #define block_width r11
5386 #define texture_offset_base r1
5387 #define texture_mask r2
5388 #define texture_page_ptr r3
5389 #define num_blocks r4
5392 #define texture_offset r8
5393 #define blocks_remaining r9
5395 #define fb_ptr_pitch r12
5396 #define texture_block_ptr r14
5398 #define texture_mask_width r2
5399 #define texture_mask_height r3
5400 #define left_mask_bits r4
5401 #define right_mask_bits r5
5405 #undef block_masks_shifted
5408 #define block_masks d0
5409 #define block_masks_shifted d1
5410 #define draw_mask_fb_ptr d2
5413 #define draw_mask_fb_ptr_a d2
5414 #define draw_mask_fb_ptr_b d3
5415 #define texels_low d4
5416 #define texels_high d5
5417 #define texels_wide_low d6
5418 #define texels_wide_high d7
5419 #define texels_wide q3
5422 setup_sprites_16bpp_flush:
5425 stmdb sp!, { r0 - r3, r12, r14 }
5426 bl flush_render_block_buffer
5427 ldmia sp!, { r0 - r3, r12, r14 }
5431 add block, psx_gpu, #psx_gpu_blocks_offset
5432 mov num_blocks, block_width
5436 function(setup_sprite_16bpp)
5437 stmdb sp!, { r4 - r11, r14 }
5438 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5441 add fb_ptr, fb_ptr, y, lsl #11
5443 ldr width, [ sp, #40 ]
5444 add fb_ptr, fb_ptr, x, lsl #1
5446 ldr height, [ sp, #44 ]
5447 and left_offset, u, #0x7
5449 add texture_offset_base, u, u
5450 add width_rounded, width, #7
5452 add texture_offset_base, texture_offset_base, v, lsl #11
5453 mov left_mask_bits, #0xFF
5455 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5456 add width_rounded, width_rounded, left_offset
5458 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5459 sub fb_ptr, fb_ptr, left_offset, lsl #1
5461 add texture_mask, texture_mask_width, texture_mask_width
5462 mov right_mask_bits, #0xFE
5464 and right_width, width_rounded, #0x7
5465 mvn left_mask_bits, left_mask_bits, lsl left_offset
5467 add texture_mask, texture_mask, texture_mask_height, lsl #11
5468 mov block_width, width_rounded, lsr #3
5470 mov right_mask_bits, right_mask_bits, lsl right_width
5471 movw fb_ptr_pitch, #(2048 + 16)
5473 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5474 vmov block_masks, left_mask_bits, right_mask_bits
5476 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5477 add block, psx_gpu, #psx_gpu_blocks_offset
5479 bic texture_offset_base, texture_offset_base, #0xF
5482 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5483 add block, block, num_blocks, lsl #6
5487 vext.32 block_masks_shifted, block_masks, block_masks, #1
5488 vorr.u32 block_masks, block_masks, block_masks_shifted
5489 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5492 add num_blocks, num_blocks, #1
5493 cmp num_blocks, #MAX_BLOCKS
5494 blgt setup_sprites_16bpp_flush
5496 and texture_block_ptr, texture_offset_base, texture_mask
5497 subs height, height, #1
5499 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5500 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5502 vst1.u32 { texels }, [ block, :128 ]
5503 add block, block, #40
5505 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5508 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5510 add block, block, #24
5511 add texture_offset_base, texture_offset_base, #2048
5512 add fb_ptr, fb_ptr, #2048
5513 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5516 ldmia sp!, { r4 - r11, pc }
5519 add num_blocks, num_blocks, block_width
5520 mov texture_offset, texture_offset_base
5522 cmp num_blocks, #MAX_BLOCKS
5523 blgt setup_sprites_16bpp_flush
5525 add texture_offset_base, texture_offset_base, #2048
5526 and texture_block_ptr, texture_offset, texture_mask
5528 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5529 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5531 vst1.u32 { texels }, [ block, :128 ]
5532 add block, block, #40
5534 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5535 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5538 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5539 subs blocks_remaining, block_width, #2
5541 add texture_offset, texture_offset, #16
5542 add fb_ptr, fb_ptr, #16
5544 vmov.u8 draw_mask_fb_ptr, #0
5546 add block, block, #24
5550 and texture_block_ptr, texture_offset, texture_mask
5551 subs blocks_remaining, blocks_remaining, #1
5553 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5554 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5556 vst1.u32 { texels }, [ block, :128 ]
5557 add block, block, #40
5559 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5562 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5564 add texture_offset, texture_offset, #16
5565 add fb_ptr, fb_ptr, #16
5567 add block, block, #24
5571 and texture_block_ptr, texture_offset, texture_mask
5572 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5574 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5575 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5577 vst1.u32 { texels }, [ block, :128 ]
5578 add block, block, #40
5580 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5581 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5583 add block, block, #24
5584 subs height, height, #1
5586 add fb_ptr, fb_ptr, fb_ptr_pitch
5587 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5591 ldmia sp!, { r4 - r11, pc }
5595 // FIXME: duplicate code with normal version :(
5596 #undef draw_mask_fb_ptr
5598 function(setup_sprite_16bpp_4x)
5599 stmdb sp!, { r4 - r11, r14 }
5600 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5603 add fb_ptr, fb_ptr, y, lsl #11
5605 ldr width, [ sp, #40 ]
5606 add fb_ptr, fb_ptr, x, lsl #1
5608 ldr height, [ sp, #44 ]
5609 and left_offset, u, #0x7
5611 add texture_offset_base, u, u
5612 add width_rounded, width, #7
5614 add texture_offset_base, texture_offset_base, v, lsl #11
5615 movw left_mask_bits, #0xFFFF
5617 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5618 add width_rounded, width_rounded, left_offset
5622 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5623 sub fb_ptr, fb_ptr, left_offset, lsl #1
5625 add texture_mask, texture_mask_width, texture_mask_width
5626 movw right_mask_bits, #0xFFFC
5628 and right_width, width_rounded, #0x7
5629 mvn left_mask_bits, left_mask_bits, lsl left_offset
5633 add texture_mask, texture_mask, texture_mask_height, lsl #11
5634 mov block_width, width_rounded, lsr #3
5636 mov right_mask_bits, right_mask_bits, lsl right_width
5637 movw fb_ptr_pitch, #(2048 + 16) * 2
5639 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4+1
5640 vmov block_masks, left_mask_bits, right_mask_bits
5642 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5643 add block, psx_gpu, #psx_gpu_blocks_offset
5645 bic texture_offset_base, texture_offset_base, #0xF
5648 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5649 add block, block, num_blocks, lsl #6
5654 vext.32 block_masks_shifted, block_masks, block_masks, #1
5655 vorr.u32 block_masks, block_masks, block_masks_shifted
5656 vdup.u8 draw_mask_fb_ptr_a, block_masks[0]
5657 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5660 add num_blocks, num_blocks, block_width
5661 cmp num_blocks, #MAX_BLOCKS
5662 blgt setup_sprites_16bpp_flush
5664 and texture_block_ptr, texture_offset_base, texture_mask
5665 subs height, height, #1
5667 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5668 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5670 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5672 add texture_offset_base, texture_offset_base, #2048
5673 add fb_ptr, fb_ptr, #2048*2
5674 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5677 ldmia sp!, { r4 - r11, pc }
5680 add num_blocks, num_blocks, block_width
5681 mov texture_offset, texture_offset_base
5683 vdup.u8 draw_mask_fb_ptr_a, block_masks[0] // left_mask_bits
5684 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5686 cmp num_blocks, #MAX_BLOCKS
5687 blgt setup_sprites_16bpp_flush
5689 add texture_offset_base, texture_offset_base, #2048
5690 and texture_block_ptr, texture_offset, texture_mask
5692 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5693 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5695 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5697 subs blocks_remaining, block_width, #2*4
5698 add texture_offset, texture_offset, #16
5700 vmov.u8 draw_mask_fb_ptr_a, #0
5701 vmov.u8 draw_mask_fb_ptr_b, #0
5703 add fb_ptr, fb_ptr, #16*2
5707 and texture_block_ptr, texture_offset, texture_mask
5708 subs blocks_remaining, blocks_remaining, #4
5710 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5711 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5713 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5714 add texture_offset, texture_offset, #16
5716 add fb_ptr, fb_ptr, #16*2
5720 vdup.u8 draw_mask_fb_ptr_a, block_masks[4] // right_mask_bits
5721 vdup.u8 draw_mask_fb_ptr_b, block_masks[5]
5723 and texture_block_ptr, texture_offset, texture_mask
5724 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5726 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5728 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5729 subs height, height, #1
5731 add fb_ptr, fb_ptr, fb_ptr_pitch
5732 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5736 ldmia sp!, { r4 - r11, pc }
5741 #undef right_mask_bits
5744 #undef blocks_remaining
5754 #define right_width r5
5755 #define right_mask_bits r6
5759 #define fb_ptr_pitch r12
5761 // referenced by setup_sprites_16bpp_flush
5762 #define num_blocks r4
5764 #define block_width r11
5769 #define blocks_remaining r6
5772 #define right_mask q1
5773 #define test_mask q2
5774 #define draw_mask q2
5775 #define draw_mask_bits_fb_ptr d6
5780 function(setup_sprite_untextured)
5781 ldrh r12, [ psx_gpu, #psx_gpu_render_state_offset ]
5782 tst r12, #(RENDER_STATE_MASK_EVALUATE | RENDER_FLAGS_MODULATE_TEXELS \
5783 | RENDER_FLAGS_BLEND)
5784 ldrbeq r12, [ psx_gpu, #psx_gpu_render_mode_offset ]
5785 tsteq r12, #RENDER_INTERLACE_ENABLED
5786 beq setup_sprite_untextured_simple
5788 stmdb sp!, { r4 - r11, r14 }
5790 ldr width, [ sp, #40 ]
5791 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5793 ldr height, [ sp, #44 ]
5794 add fb_ptr, fb_ptr, y, lsl #11
5796 add fb_ptr, fb_ptr, x, lsl #1
5797 sub right_width, width, #1
5799 ldr color, [ sp, #48 ]
5802 add block_width, width, #7
5806 mov right_mask_bits, #0xff
5808 sub fb_ptr_pitch, block_width, #1
5809 lsl right_mask_bits, right_width
5811 lsl fb_ptr_pitch, #3+1
5812 ubfx color_r, color, #3, #5
5814 rsb fb_ptr_pitch, #1024*2
5815 ubfx color_g, color, #11, #5
5817 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
5818 ubfx color_b, color, #19, #5
5820 vdup.u16 right_mask, right_mask_bits
5821 orr color, color_r, color_b, lsl #10
5823 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5824 orr color, color, color_g, lsl #5
5826 vtst.u16 right_mask, right_mask, test_mask
5827 add block, psx_gpu, #psx_gpu_blocks_offset
5829 vdup.u16 colors, color
5830 add block, block, num_blocks, lsl #6
5833 setup_sprite_untextured_height_loop:
5834 add num_blocks, block_width
5835 sub blocks_remaining, block_width, #1
5837 cmp num_blocks, #MAX_BLOCKS
5838 blgt setup_sprites_16bpp_flush
5840 cmp blocks_remaining, #0
5843 vmov.u8 draw_mask, #0 /* zero_mask */
5844 vmov.u8 draw_mask_bits_fb_ptr, #0
5847 vst1.u32 { draw_mask }, [ block, :128 ]!
5848 subs blocks_remaining, #1
5850 vst1.u32 { colors }, [ block, :128 ]
5851 add block, block, #24
5853 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5854 vst1.u32 { draw_mask_bits_fb_ptr }, [ block, :64 ]
5856 add block, block, #24
5861 vst1.u32 { right_mask }, [ block, :128 ]!
5864 vst1.u32 { colors }, [ block, :128 ]
5865 add block, block, #24
5867 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5868 vst1.u32 { draw_mask_bits_fb_ptr }, [ block, :64 ]
5870 add block, block, #24
5871 add fb_ptr, fb_ptr_pitch
5873 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5874 bgt setup_sprite_untextured_height_loop
5876 ldmia sp!, { r4 - r11, pc }
5880 #undef texture_page_ptr
5882 #undef dirty_textures_mask
5883 #undef current_texture_mask
5886 #define current_texture_page r1
5887 #define texture_page_ptr r2
5888 #define vram_ptr_a r3
5889 #define current_texture_page_x r12
5890 #define current_texture_page_y r4
5891 #define dirty_textures_mask r5
5895 #define current_texture_mask r9
5897 #define vram_ptr_b r11
5899 #define texel_block_a d0
5900 #define texel_block_b d1
5901 #define texel_block_expanded_a q1
5902 #define texel_block_expanded_b q2
5903 #define texel_block_expanded_ab q2
5904 #define texel_block_expanded_c q3
5905 #define texel_block_expanded_d q4
5906 #define texel_block_expanded_cd q3
5908 function(update_texture_4bpp_cache)
5909 stmdb sp!, { r4 - r11, r14 }
5912 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5914 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5915 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5917 and current_texture_page_x, current_texture_page, #0xF
5918 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
5920 mov current_texture_page_y, current_texture_page, lsr #4
5921 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5923 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5926 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5927 bic dirty_textures_mask, current_texture_mask
5930 str dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5935 add vram_ptr_b, vram_ptr_a, #2048
5938 vld1.u32 { texel_block_a }, [ vram_ptr_a, :64 ], c_4096
5939 vld1.u32 { texel_block_b }, [ vram_ptr_b, :64 ], c_4096
5941 vmovl.u8 texel_block_expanded_a, texel_block_a
5942 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5943 vmovl.u8 texel_block_expanded_c, texel_block_b
5944 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5946 vbic.u16 texel_block_expanded_a, #0x00F0
5947 vbic.u16 texel_block_expanded_b, #0x00F0
5948 vbic.u16 texel_block_expanded_c, #0x00F0
5949 vbic.u16 texel_block_expanded_d, #0x00F0
5951 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
5952 texel_block_expanded_b
5953 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
5954 texel_block_expanded_d
5956 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
5957 [ texture_page_ptr, :256 ]!
5959 subs sub_y, sub_y, #1
5963 add vram_ptr_a, vram_ptr_a, #8
5964 add vram_ptr_b, vram_ptr_b, #8
5966 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5967 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5969 subs tile_x, tile_x, #1
5973 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5974 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5976 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5977 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5979 subs tile_y, tile_y, #1
5983 ldmia sp!, { r4 - r11, pc }
5986 #undef current_texture_page
5989 #define texture_page r1
5990 #define texture_page_ptr r2
5991 #define vram_ptr_a r3
5992 #define texture_page_x r12
5993 #define texture_page_y r4
5994 #define current_texture_page r5
5999 #define vram_ptr_b r11
6011 function(update_texture_8bpp_cache_slice)
6012 stmdb sp!, { r4 - r11, r14 }
6015 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
6016 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
6018 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
6021 and texture_page_x, texture_page, #0xF
6022 mov texture_page_y, texture_page, lsr #4
6024 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
6027 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
6028 eor current_texture_page, current_texture_page, texture_page
6030 ands current_texture_page, current_texture_page, #0x1
6033 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6036 add vram_ptr_b, vram_ptr_a, #2048
6039 vld1.u32 { texels_a }, [ vram_ptr_a, :128 ], c_4096
6040 vld1.u32 { texels_b }, [ vram_ptr_b, :128 ], c_4096
6041 vld1.u32 { texels_c }, [ vram_ptr_a, :128 ], c_4096
6042 vld1.u32 { texels_d }, [ vram_ptr_b, :128 ], c_4096
6044 vst1.u32 { texels_a, texels_b }, [ texture_page_ptr, :256 ]!
6045 vst1.u32 { texels_c, texels_d }, [ texture_page_ptr, :256 ]!
6047 subs sub_y, sub_y, #1
6052 add vram_ptr_a, vram_ptr_a, #16
6053 add vram_ptr_b, vram_ptr_b, #16
6055 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
6056 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
6058 subs tile_x, tile_x, #1
6063 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
6064 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
6066 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
6067 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
6069 subs tile_y, tile_y, #1
6070 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6075 ldmia sp!, { r4 - r11, pc }
6078 /* void scale2x_tiles8(void *dst, const void *src, int w8, int h) */
6079 function(scale2x_tiles8)
6083 add r12, r0, #1024*2
6087 vld1.u16 { q0 }, [ r1, :128 ]!
6088 vld1.u16 { q2 }, [ r1, :128 ]!
6094 vst1.u16 { q0, q1 }, [ r0, :128 ]!
6095 vst1.u16 { q0, q1 }, [ r12, :128 ]!
6097 vst1.u16 { q2, q3 }, [ r0, :128 ]!
6098 vst1.u16 { q2, q3 }, [ r12, :128 ]!
6105 sub r0, r0, r2, lsl #4+1
6107 add r12, r0, #1024*2
6113 // vim:filetype=armasm