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 #include "psx_gpu_offsets.h"
22 #define psx_gpu_b_dx_offset (psx_gpu_b_block_span_offset + 4)
24 #define edge_data_left_x_offset 0
25 #define edge_data_num_blocks_offset 2
26 #define edge_data_right_mask_offset 4
27 #define edge_data_y_offset 6
71 #define gw_by_l gw_bx_l
101 #define uvrg_xxxx2 q5
106 #define yyyy_uvrg0 q6
111 #define yyyy_uvrg1 q7
116 #define yyyy_uvrg2 q8
143 #define uvrg_base q11
186 #define uvrgb_phase q13
190 /* FIXME: users of this should be in psx_gpu instead */
192 #define load_pointer(register, pointer) \
193 movw register, :lower16:pointer; \
194 movt register, :upper16:pointer; \
197 #define load_pointer(register, pointer) \
198 ldr register, =pointer \
202 #define function(name) \
211 function(compute_all_gradients)
212 // First compute the triangle area reciprocal and shift. The division will
213 // happen concurrently with much of the work which follows.
214 @ r12 = psx_gpu->triangle_area
215 ldr r12, [ psx_gpu, #psx_gpu_triangle_area_offset ]
216 stmdb sp!, { r4 - r11, lr }
218 @ load exponent of 62 into upper half of double
220 clz r14, r12 @ r14 = shift
222 movt r4, #((62 + 1023) << 4)
223 mov r12, r12, lsl r14 @ r12 = triangle_area_normalized
225 @ load area normalized into lower half of double
227 vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n
229 movt r4, #((1022 + 31) << 4)
232 add r4, r4, r12, lsr #11
235 vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n
237 // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) =
238 // ( d0 * d1 ) - ( d2 * d3 ) =
239 // ( m0 ) - ( m1 ) = gradient
241 // This is split to do 12 elements at a time over three sets: a, b, and c.
242 // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so
243 // two of the slots are unused.
245 // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as
248 // First type is: uvrg bxxx xxxx
249 // Second type is: yyyy ybyy uvrg
250 // Since x_a and y_c are the same the same variable is used for both.
252 vld1.u32 { v0 }, [ v_a, : 128 ] @ v0 = { uvrg0, b0, x0, y0 }
253 ldrsh x0, [ v_a, #8 ] @ load x0
255 vld1.u32 { v1 }, [ v_b, : 128 ] @ v1 = { uvrg1, b1, x1, y1}
256 ldrh x1, [ v_b, #8 ] @ load x1
258 vld1.u32 { v2 }, [ v_c, : 128 ] @ v2 = { uvrg2, b2, x2, y2 }
259 ldrh x2, [ v_c, #8 ] @ load x2
261 vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- }
262 ldrh y0, [ v_a, #10 ] @ load y0
264 vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- }
265 ldrh y1, [ v_b, #10 ] @ load y1
267 vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- }
268 ldrh y2, [ v_c, #10 ] @ load y2
270 vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 }
271 vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 }
273 orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 }
274 pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 }
276 vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 }
277 vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 }
279 vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 }
280 vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 }
282 ldrb b2, [ v_c, #4 ] @ load b2
283 orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 }
285 ldrb b1, [ v_b, #4 ] @ load b1
286 orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 }
288 vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 }
289 vsub.s16 d0_ab, x1_ab, x0_ab
291 ldrb b0, [ v_a, #4 ] @ load b0
292 orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 }
294 vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 }
295 vsub.s16 d2_ab, x2_ab, x1_ab
297 vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 }
298 vsub.s16 d1_ab, y2_ab, y1_ab
300 orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 }
301 ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 }
303 ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 }
304 ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 }
306 vsub.s16 d3_ab, y1_ab, y0_ab
307 smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) -
308 @ ((x2 - X1) * (b1 - b0))
309 vmull.s16 ga_uvrg_x, d0_a, d1_a
310 smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) -
311 @ ((b2 - b1) * (y1 - y0))
312 vmlsl.s16 ga_uvrg_x, d2_a, d3_a
313 movs gs_bx, ga_bx, asr #31
315 vmull.s16 ga_uvrg_y, d0_b, d1_b
316 rsbmi ga_bx, ga_bx, #0
318 @ r12 = psx_gpu->uvrgb_phase
319 ldr r12, [ psx_gpu, #psx_gpu_uvrgb_phase_offset ]
321 vmlsl.s16 ga_uvrg_y, d2_b, d3_b
322 movs gs_by, ga_by, asr #31
324 vshr.u64 d0, d30, #22
325 add b_base, r12, b0, lsl #16
327 vdup.u32 uvrgb_phase, r12
329 rsbmi ga_by, ga_by, #0
330 vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0
332 @ r12 = psx_gpu->triangle_winding_offset
333 ldrb r12, [ psx_gpu, #psx_gpu_triangle_winding_offset ]
334 vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0
336 rsb r12, r12, #0 @ r12 = -(triangle->winding)
338 vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w }
339 sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS)
341 vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16
342 vdup.u32 r_shift, r14 @ r_shift = { shift, shift, shift, shift }
344 vadd.u32 uvrg_base, uvrgb_phase
345 vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x)
347 vmov area_r_s, s0 @ area_r_s = triangle_reciprocal
348 vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y)
350 vmull.u32 gw_rg_x, ga_rg_x, d0[0]
351 vmull.u32 gw_uv_x, ga_uv_x, d0[0]
352 vmull.u32 gw_rg_y, ga_rg_y, d0[0]
353 vmull.u32 gw_uv_y, ga_uv_y, d0[0]
355 vshl.u64 gw_rg_x, gw_rg_x, r_shift
356 vshl.u64 gw_uv_x, gw_uv_x, r_shift
357 vshl.u64 gw_rg_y, gw_rg_y, r_shift
358 vshl.u64 gw_uv_y, gw_uv_y, r_shift
360 veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask
361 vmovn.u64 g_uv_x, gw_uv_x
363 veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask
364 vmovn.u64 g_rg_x, gw_rg_x
366 veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
367 vmovn.u64 g_uv_y, gw_uv_y
369 vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
370 vmovn.u64 g_rg_y, gw_rg_y
372 veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
373 mov ga_bx, ga_bx, lsl #13
375 vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
376 mov ga_by, ga_by, lsl #13
379 umull gw_bx_l, gw_bx_h, ga_bx, area_r_s
381 vshl.u32 g_uvrg_x, g_uvrg_x, #4
382 vshl.u32 g_uvrg_y, g_uvrg_y, #4
384 umull gw_by_l, gw_by_h, ga_by, area_r_s
385 vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0]
387 eor gs_bx, gs_bx, r12
388 vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1
390 veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0
391 eor gs_by, gs_by, r12
393 rsb r11, r14, #0 @ r11 = negative shift for scalar lsr
394 add store_a, psx_gpu, #psx_gpu_uvrg_offset
396 sub r11, r11, #(32 - 13)
398 add store_b, store_a, #16
401 vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1
402 vst1.u32 { uvrg_base }, [ store_a, : 128 ], store_inc
404 vst1.u32 { uvrg_dx1 }, [ store_b, : 128 ], store_inc
405 mov g_bx, gw_bx_h, lsr r11
407 vst1.u32 { g_uvrg_y }, [ store_a, : 128 ], store_inc
408 mov g_by, gw_by_h, lsr r11
410 vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \
411 [ store_b, : 128 ], store_inc
412 eor g_bx, g_bx, gs_bx
414 vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \
415 [ store_b, : 128 ], store_inc
416 sub g_bx, g_bx, gs_bx
419 eor g_by, g_by, gs_by
421 mls b_base, g_bx, x0, b_base
422 sub g_by, g_by, gs_by
427 add g_bx2, g_bx, g_bx
428 add g_bx3, g_bx, g_bx2
430 stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by }
432 ldmia sp!, { r4 - r11, pc }
449 #define height_minor_a r7
450 #define height_minor_b r8
451 #define height_major r9
454 #define reciprocal_table_ptr r10
456 #define edge_alt_low r4
457 #define edge_alt_high r5
458 #define edge_dx_dy_alt r6
459 #define edge_shift_alt r10
461 #define edge_dx_dy_alt_low r4
462 #define edge_dx_dy_alt_high r5
464 #define span_edge_data r4
465 #define span_uvrg_offset r5
466 #define span_b_offset r6
474 #define alternate_x q0
475 #define alternate_dx_dy q1
476 #define alternate_x_32 q2
478 #define alternate_x_low d0
479 #define alternate_x_high d1
480 #define alternate_dx_dy_low d2
481 #define alternate_dx_dy_high d3
482 #define alternate_x_32_low d4
483 #define alternate_x_32_high d5
487 #define left_dx_dy q5
488 #define right_dx_dy q6
490 #define right_edge q8
492 #define left_x_low d6
493 #define left_x_high d7
494 #define right_x_low d8
495 #define right_x_high d9
496 #define left_dx_dy_low d10
497 #define left_dx_dy_high d11
498 #define right_dx_dy_low d12
499 #define right_dx_dy_high d13
500 #define left_edge_low d14
501 #define left_edge_high d15
502 #define right_edge_low d16
503 #define right_edge_high d17
505 #define y_mid_point d18
508 #define left_right_x_16 q11
509 #define span_shifts_y q12
512 #define span_shifts d24
517 #define left_right_x_16_low d22
518 #define left_right_x_16_high d23
523 #define alternate_x_16 d4
526 #define v_clip_low d6
528 #define right_x_32 q10
529 #define left_x_32 q11
530 #define alternate_select d24
532 #define right_x_32_low d20
533 #define right_x_32_high d21
534 #define left_x_32_low d22
535 #define left_x_32_high d23
538 #define edges_dx_dy d2
539 #define edge_shifts d3
540 #define edge_shifts_64 q2
542 #define edges_xy_left d0
543 #define edges_xy_right d1
545 #define height_reciprocals d6
553 #define heights_b d12
554 #define edges_dx_dy_64 q10
556 #define edges_dx_dy_64_left d20
557 #define edges_dx_dy_64_right d21
560 #define setup_spans_prologue() \
561 stmdb sp!, { r4 - r11, lr }; \
563 ldrsh x_a, [ v_a, #8 ]; \
564 ldrsh x_b, [ v_b, #8 ]; \
565 ldrsh x_c, [ v_c, #8 ]; \
566 ldrsh y_a, [ v_a, #10 ]; \
567 ldrsh y_b, [ v_b, #10 ]; \
568 ldrsh y_c, [ v_c, #10 ]; \
570 add temp, psx_gpu, #psx_gpu_uvrg_offset; \
571 vld1.32 { uvrg }, [ temp ]; \
572 add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \
573 vld1.32 { uvrg_dy }, [ temp ]; \
574 load_pointer(reciprocal_table_ptr, reciprocal_table); \
576 vmov.u32 c_0x01, #0x01 \
578 #define setup_spans_load_b() \
579 ldr b, [ psx_gpu, #psx_gpu_b_offset ]; \
580 ldr b_dy, [ psx_gpu, #psx_gpu_b_dy_offset ] \
582 #define setup_spans_prologue_b() \
583 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
584 add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \
586 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
587 vmov.u16 c_0x0004, #0x0004; \
589 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
590 vmov.u16 c_0x0001, #0x0001; \
592 vld1.u16 { left_edge_low[], left_edge_high[] }, [ temp ]; \
593 add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \
595 vld1.u16 { right_edge_low[], right_edge_high[] }, [ temp ]; \
596 vadd.u16 right_edge, right_edge, c_0x0001; \
598 vmov.u16 c_0x0007, #0x0007; \
599 vmvn.u16 c_0xFFFE, #0x0001 \
602 #define compute_edge_delta_x2() \
603 ldr temp, [ reciprocal_table_ptr, height, lsl #2 ]; \
605 vdup.u32 heights, height; \
606 vsub.u32 widths, x_ends, x_starts; \
608 vdup.u32 edge_shifts, temp; \
609 vsub.u32 heights_b, heights, c_0x01; \
610 vshr.u32 height_reciprocals, edge_shifts, #10; \
612 vmla.s32 heights_b, x_starts, heights; \
613 vbic.u16 edge_shifts, #0xE0; \
614 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
615 vmull.s32 edges_xy, heights_b, height_reciprocals \
618 #define height_reciprocal_alt r11
619 #define height_b_alt r12
621 #define compute_edge_delta_x3(start_c, height_a, height_b) \
622 vmov.u32 heights, height_a, height_b; \
623 ldr temp, [ reciprocal_table_ptr, height_a, lsl #2 ]; \
624 vmov.u32 edge_shifts[0], temp; \
625 ldr temp, [ reciprocal_table_ptr, height_b, lsl #2 ]; \
626 vmov.u32 edge_shifts[1], temp; \
627 ldr edge_shift_alt, [ reciprocal_table_ptr, height_minor_b, lsl #2 ]; \
629 vsub.u32 widths, x_ends, x_starts; \
630 sub width_alt, x_c, start_c; \
632 vsub.u32 heights_b, heights, c_0x01; \
633 sub height_b_alt, height_minor_b, #1; \
635 vshr.u32 height_reciprocals, edge_shifts, #10; \
636 lsr height_reciprocal_alt, edge_shift_alt, #10; \
638 vmla.s32 heights_b, x_starts, heights; \
639 mla height_b_alt, height_minor_b, start_c, height_b_alt; \
641 vbic.u16 edge_shifts, #0xE0; \
642 and edge_shift_alt, edge_shift_alt, #0x1F; \
644 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
645 mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \
647 vmull.s32 edges_xy, heights_b, height_reciprocals; \
648 smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \
651 #define setup_spans_adjust_y_up() \
652 vsub.u32 y_x4, y_x4, c_0x0004 \
654 #define setup_spans_adjust_y_down() \
655 vadd.u32 y_x4, y_x4, c_0x0004 \
657 #define setup_spans_adjust_interpolants_up() \
658 vsub.u32 uvrg, uvrg, uvrg_dy; \
661 #define setup_spans_adjust_interpolants_down() \
662 vadd.u32 uvrg, uvrg, uvrg_dy; \
666 #define setup_spans_clip_interpolants_increment() \
667 mla b, b_dy, clip, b; \
668 vmla.s32 uvrg, uvrg_dy, v_clip \
670 #define setup_spans_clip_interpolants_decrement() \
671 mls b, b_dy, clip, b; \
672 vmls.s32 uvrg, uvrg_dy, v_clip \
674 #define setup_spans_clip_alternate_yes() \
675 smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \
677 #define setup_spans_clip_alternate_no() \
679 #define setup_spans_clip(direction, alternate_active) \
680 vdup.u32 v_clip, clip; \
681 setup_spans_clip_alternate_##alternate_active(); \
682 setup_spans_clip_interpolants_##direction(); \
683 vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \
686 #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \
687 vmovl.s32 edge_shifts_64, edge_shifts; \
688 vmovl.s32 edges_dx_dy_64, edges_dx_dy; \
690 vshl.s64 edges_xy, edges_xy, edge_shifts_64; \
691 vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \
693 vmov left_x_low, edges_xy_##left_index; \
694 vmov right_x_low, edges_xy_##right_index; \
696 vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \
697 vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \
698 vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \
699 vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \
701 vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \
702 vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \
704 vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \
705 vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \
708 #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \
709 setup_spans_adjust_edges_alternate_no(left_index, right_index); \
711 vdup.u16 y_mid_point, y_b; \
712 rsb temp, edge_shift_alt, #32; \
714 lsl edge_alt_high, edge_alt_high, edge_shift_alt; \
715 orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \
716 lsl edge_alt_low, edge_alt_low, edge_shift_alt; \
717 vmov alternate_x_low, edge_alt_low, edge_alt_high; \
719 asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \
720 lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \
721 vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \
722 vmov alternate_dx_dy_high, alternate_dx_dy_low; \
724 vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \
725 vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \
728 #define setup_spans_y_select_up() \
729 vclt.s16 alternate_select, y_x4, y_mid_point \
731 #define setup_spans_y_select_down() \
732 vcgt.s16 alternate_select, y_x4, y_mid_point \
735 #define setup_spans_alternate_select_left() \
736 vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \
738 #define setup_spans_alternate_select_right() \
739 vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \
742 #define setup_spans_set_x4_alternate_yes(alternate, direction) \
743 vshrn.s64 alternate_x_32_low, alternate_x, #32; \
744 vshrn.s64 left_x_32_low, left_x, #32; \
745 vshrn.s64 right_x_32_low, right_x, #32; \
747 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
748 vadd.u64 left_x, left_x, left_dx_dy; \
749 vadd.u64 right_x, right_x, right_dx_dy; \
751 vshrn.s64 alternate_x_32_high, alternate_x, #32; \
752 vshrn.s64 left_x_32_high, left_x, #32; \
753 vshrn.s64 right_x_32_high, right_x, #32; \
755 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
756 vadd.u64 left_x, left_x, left_dx_dy; \
757 vadd.u64 right_x, right_x, right_dx_dy; \
759 vmovn.u32 alternate_x_16, alternate_x_32; \
760 setup_spans_y_select_##direction(); \
761 vmovn.u32 left_right_x_16_low, left_x_32; \
763 vmovn.u32 left_right_x_16_high, right_x_32; \
764 setup_spans_alternate_select_##alternate(); \
766 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
767 str b, [ span_b_offset ], #4; \
768 setup_spans_adjust_interpolants_##direction(); \
770 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
772 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
773 str b, [ span_b_offset ], #4; \
774 setup_spans_adjust_interpolants_##direction(); \
776 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
778 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
779 str b, [ span_b_offset ], #4; \
780 setup_spans_adjust_interpolants_##direction(); \
782 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
783 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
784 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
786 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
787 str b, [ span_b_offset ], #4; \
788 setup_spans_adjust_interpolants_##direction(); \
790 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
791 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
793 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
795 setup_spans_adjust_y_##direction() \
798 #define setup_spans_set_x4_alternate_no(alternate, direction) \
799 vshrn.s64 left_x_32_low, left_x, #32; \
800 vshrn.s64 right_x_32_low, right_x, #32; \
802 vadd.u64 left_x, left_x, left_dx_dy; \
803 vadd.u64 right_x, right_x, right_dx_dy; \
805 vshrn.s64 left_x_32_high, left_x, #32; \
806 vshrn.s64 right_x_32_high, right_x, #32; \
808 vadd.u64 left_x, left_x, left_dx_dy; \
809 vadd.u64 right_x, right_x, right_dx_dy; \
811 vmovn.u32 left_right_x_16_low, left_x_32; \
812 vmovn.u32 left_right_x_16_high, right_x_32; \
814 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
815 str b, [ span_b_offset ], #4; \
816 setup_spans_adjust_interpolants_##direction(); \
818 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
820 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
821 str b, [ span_b_offset ], #4; \
822 setup_spans_adjust_interpolants_##direction(); \
824 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
826 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
827 str b, [ span_b_offset ], #4; \
828 setup_spans_adjust_interpolants_##direction(); \
830 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
831 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
832 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
834 vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \
835 str b, [ span_b_offset ], #4; \
836 setup_spans_adjust_interpolants_##direction(); \
838 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
839 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
841 vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \
843 setup_spans_adjust_y_##direction() \
846 #define edge_adjust_low r11
847 #define edge_adjust_high r12
849 #define setup_spans_alternate_adjust_yes() \
850 smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \
851 subs edge_alt_low, edge_alt_low, edge_adjust_low; \
852 sbc edge_alt_high, edge_alt_high, edge_adjust_high \
854 #define setup_spans_alternate_adjust_no() \
857 #define setup_spans_down(left_index, right_index, alternate, alternate_active) \
858 setup_spans_alternate_adjust_##alternate_active(); \
859 setup_spans_load_b(); \
861 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
862 subs y_c, y_c, temp; \
863 subgt height, height, y_c; \
864 addgt height, height, #1; \
866 ldrsh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
867 subs clip, temp, y_a; \
870 sub height, height, clip; \
871 add y_a, y_a, clip; \
872 setup_spans_clip(increment, alternate_active); \
878 orr temp, y_a, y_a, lsl #16; \
879 add temp, temp, #(1 << 16); \
881 add y_a, y_a, #(2 << 16); \
882 vmov.u32 y_x4, temp, y_a; \
884 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
886 setup_spans_prologue_b(); \
888 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
891 setup_spans_set_x4_alternate_##alternate_active(alternate, down); \
892 subs height, height, #4; \
898 #define setup_spans_alternate_pre_increment_yes() \
899 adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \
900 adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \
902 #define setup_spans_alternate_pre_increment_no() \
905 #define setup_spans_up_decrement_yes() \
906 suble height, height, #1 \
908 #define setup_spans_up_decrement_no() \
911 #define setup_spans_up(left_index, right_index, alternate, alternate_active) \
912 setup_spans_alternate_adjust_##alternate_active(); \
913 setup_spans_load_b(); \
916 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \
917 subs temp, temp, y_c; \
918 subgt height, height, temp; \
919 setup_spans_up_decrement_##alternate_active(); \
921 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \
922 subs clip, y_a, temp; \
925 sub height, height, clip; \
926 sub y_a, y_a, clip; \
927 setup_spans_clip(decrement, alternate_active); \
933 orr temp, y_a, y_a, lsl #16; \
934 sub temp, temp, #(1 << 16); \
936 sub y_a, y_a, #(2 << 16); \
937 vmov.u32 y_x4, temp, y_a; \
939 vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \
941 setup_spans_alternate_pre_increment_##alternate_active(); \
942 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
944 setup_spans_adjust_interpolants_up(); \
945 setup_spans_prologue_b(); \
947 strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
950 setup_spans_set_x4_alternate_##alternate_active(alternate, up); \
951 subs height, height, #4; \
957 #define setup_spans_epilogue() \
958 ldmia sp!, { r4 - r11, pc } \
961 #define setup_spans_up_up(minor, major) \
962 setup_spans_prologue(); \
963 sub height_minor_a, y_a, y_b; \
964 sub height_minor_b, y_b, y_c; \
965 sub height, y_a, y_c; \
967 vdup.u32 x_starts, x_a; \
968 vmov.u32 x_ends, x_c, x_b; \
970 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
971 setup_spans_up(major, minor, minor, yes); \
972 setup_spans_epilogue() \
974 function(setup_spans_up_left)
975 setup_spans_up_up(left, right)
977 function(setup_spans_up_right)
978 setup_spans_up_up(right, left)
982 #define setup_spans_down_down(minor, major) \
983 setup_spans_prologue(); \
984 sub height_minor_a, y_b, y_a; \
985 sub height_minor_b, y_c, y_b; \
986 sub height, y_c, y_a; \
988 vdup.u32 x_starts, x_a; \
989 vmov.u32 x_ends, x_c, x_b; \
991 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
992 setup_spans_down(major, minor, minor, yes); \
993 setup_spans_epilogue() \
995 function(setup_spans_down_left)
996 setup_spans_down_down(left, right)
998 function(setup_spans_down_right)
999 setup_spans_down_down(right, left)
1002 #define setup_spans_up_flat() \
1003 sub height, y_a, y_c; \
1005 compute_edge_delta_x2(); \
1006 setup_spans_up(left, right, none, no); \
1007 setup_spans_epilogue() \
1009 function(setup_spans_up_a)
1010 setup_spans_prologue()
1012 vmov.u32 x_starts, x_a, x_b
1013 vdup.u32 x_ends, x_c
1015 setup_spans_up_flat()
1017 function(setup_spans_up_b)
1018 setup_spans_prologue()
1020 vdup.u32 x_starts, x_a
1021 vmov.u32 x_ends, x_b, x_c
1023 setup_spans_up_flat()
1025 #define setup_spans_down_flat() \
1026 sub height, y_c, y_a; \
1028 compute_edge_delta_x2(); \
1029 setup_spans_down(left, right, none, no); \
1030 setup_spans_epilogue() \
1032 function(setup_spans_down_a)
1033 setup_spans_prologue()
1035 vmov.u32 x_starts, x_a, x_b
1036 vdup.u32 x_ends, x_c
1038 setup_spans_down_flat()
1040 function(setup_spans_down_b)
1041 setup_spans_prologue()
1043 vdup.u32 x_starts, x_a
1044 vmov.u32 x_ends, x_b, x_c
1046 setup_spans_down_flat()
1051 #define edges_xy_b q11
1052 #define edges_dx_dy_b d26
1053 #define edge_shifts_b d27
1054 #define edges_dx_dy_and_shifts_b q13
1055 #define height_increment d20
1057 #define edges_dx_dy_and_shifts q1
1059 #define edges_xy_b_left d22
1060 #define edges_xy_b_right d23
1062 #define setup_spans_up_down_load_edge_set_b() \
1063 vmov edges_xy, edges_xy_b; \
1064 vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \
1067 function(setup_spans_up_down)
1068 setup_spans_prologue()
1070 // s32 middle_y = y_a;
1071 sub height_minor_a, y_a, y_b
1072 sub height_minor_b, y_c, y_a
1073 sub height_major, y_c, y_b
1075 vmov.u32 x_starts, x_a, x_c
1076 vdup.u32 x_ends, x_b
1078 compute_edge_delta_x3(x_a, height_minor_a, height_major)
1081 vmov.u32 height_increment, temp, height_minor_b
1082 vmlal.s32 edges_xy, edges_dx_dy, height_increment
1084 vmov edges_xy_b_left, edge_alt_low, edge_alt_high
1085 vmov edges_xy_b_right, edges_xy_right
1087 vmov edge_shifts_b, edge_shifts
1088 vmov.u32 edge_shifts_b[0], edge_shift_alt
1090 vneg.s32 edges_dx_dy_b, edges_dx_dy
1091 vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt
1095 setup_spans_load_b()
1098 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1099 subs temp, temp, y_b
1100 subgt height_minor_a, height_minor_a, temp
1102 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1103 subs clip, y_a, temp
1106 sub height_minor_a, height_minor_a, clip
1108 setup_spans_clip(decrement, no)
1111 cmp height_minor_a, #0
1114 orr temp, y_a, y_a, lsl #16
1115 sub temp, temp, #(1 << 16)
1117 sub y_a, y_a, #(2 << 16)
1118 vmov.u32 y_x4, temp, y_a
1120 vaddw.s32 edges_xy, edges_xy, edges_dx_dy
1122 strh height_minor_a, [ psx_gpu, #psx_gpu_num_spans_offset ]
1124 setup_spans_adjust_edges_alternate_no(left, right);
1125 setup_spans_adjust_interpolants_up()
1126 setup_spans_up_down_load_edge_set_b()
1128 setup_spans_prologue_b()
1132 setup_spans_set_x4_alternate_no(none, up)
1133 subs height_minor_a, height_minor_a, #4
1136 add span_edge_data, span_edge_data, height_minor_a, lsl #3
1137 add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4
1138 add span_b_offset, span_b_offset, height_minor_a, lsl #2
1141 add temp, psx_gpu, #psx_gpu_uvrg_offset
1142 vld1.32 { uvrg }, [ temp ]
1145 setup_spans_load_b()
1147 ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]
1149 subgt height_minor_b, height_minor_b, y_c
1150 addgt height_minor_b, height_minor_b, #1
1152 ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]
1153 subs clip, temp, y_a
1156 sub height_minor_b, height_minor_b, clip
1158 setup_spans_clip(increment, no)
1161 cmp height_minor_b, #0
1164 orr temp, y_a, y_a, lsl #16
1165 add temp, temp, #(1 << 16)
1167 add y_a, y_a, #(2 << 16)
1168 vmov.u32 y_x4, temp, y_a
1170 setup_spans_adjust_edges_alternate_no(left, right)
1172 ldrh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1173 add temp, temp, height_minor_b
1174 strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ]
1177 setup_spans_set_x4_alternate_no(none, down)
1178 subs height_minor_b, height_minor_b, #4
1182 setup_spans_epilogue()
1185 setup_spans_up_down_load_edge_set_b()
1186 setup_spans_prologue_b()
1191 #undef span_uvrg_offset
1192 #undef span_edge_data
1193 #undef span_b_offset
1198 #define num_spans r1
1199 #define span_uvrg_offset r2
1200 #define span_edge_data r3
1201 #define span_b_offset r4
1203 #define span_num_blocks r6
1207 #define dither_offset_ptr r10
1208 #define block_ptr_a r11
1210 #define num_blocks r14
1212 #define uvrg_dx_ptr r2
1213 #define texture_mask_ptr r3
1214 #define dither_shift r8
1215 #define dither_row r10
1220 #define block_ptr_b r10
1222 #define block_span_ptr r10
1223 #define right_mask r8
1242 #define b_whole_8 d14
1243 #define fb_mask_ptrs d15
1256 #define u_whole_low d16
1257 #define u_whole_high d17
1258 #define v_whole_low d18
1259 #define v_whole_high d19
1260 #define r_whole_low d20
1261 #define r_whole_high d21
1262 #define g_whole_low d22
1263 #define g_whole_high d23
1264 #define b_whole_low d24
1265 #define b_whole_high d25
1270 #define u_whole_8 d26
1271 #define v_whole_8 d27
1272 #define u_whole_8b d24
1273 #define r_whole_8 d24
1274 #define g_whole_8 d25
1276 #define uv_whole_8 q13
1277 #define uv_whole_8b q14
1279 #define dither_offsets q14
1280 #define texture_mask q15
1281 #define texture_mask_u d30
1282 #define texture_mask_v d31
1284 #define dither_offsets_short d28
1288 #define block_span q10
1293 #define draw_mask q1
1294 #define draw_mask_edge q13
1295 #define test_mask q0
1301 #define setup_blocks_texture_swizzled() \
1302 vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \
1303 vsli.u8 u_whole_8, v_whole_8, #4; \
1304 vsri.u8 v_whole_8, u_whole_8b, #4 \
1306 #define setup_blocks_texture_unswizzled() \
1309 #define setup_blocks_shaded_textured_builder(swizzling) \
1312 function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \
1313 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1314 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1316 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1317 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1319 cmp num_spans, #0; \
1322 stmdb sp!, { r4 - r11, r14 }; \
1323 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1325 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
1326 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1328 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1329 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1331 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1332 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1334 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
1335 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1337 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1340 vmov.u8 fb_mask_ptrs, #0; \
1342 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1343 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1345 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1346 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
1348 cmp span_num_blocks, #0; \
1351 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1352 add num_blocks, span_num_blocks, num_blocks; \
1354 cmp num_blocks, #MAX_BLOCKS; \
1358 ldr b, [ span_b_offset ]; \
1359 add fb_ptr, fb_ptr, y, lsl #11; \
1361 vdup.u32 v_left_x, left_x; \
1364 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1365 add fb_ptr, fb_ptr, left_x, lsl #1; \
1367 mla b, b_dx, left_x, b; \
1368 and dither_shift, left_x, #0x03; \
1370 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1371 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1373 mov dither_shift, dither_shift, lsl #3; \
1374 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1377 subs span_num_blocks, span_num_blocks, #1; \
1379 mov dither_row, dither_row, ror dither_shift; \
1380 mov b_dx4, b_dx, lsl #2; \
1382 vdup.u32 dither_offsets_short, dither_row; \
1383 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1385 vdup.u32 b_block, b; \
1386 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1388 vdup.u32 u_block, uv[0]; \
1389 mov b_dx8, b_dx, lsl #3; \
1391 vdup.u32 v_block, uv[1]; \
1392 vdup.u32 r_block, rg[0]; \
1393 vdup.u32 g_block, rg[1]; \
1395 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1397 vadd.u32 u_block, u_block, block_span; \
1398 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1400 vadd.u32 v_block, v_block, block_span; \
1401 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1403 vadd.u32 r_block, r_block, block_span; \
1404 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1406 vadd.u32 g_block, g_block, block_span; \
1407 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
1409 vadd.u32 b_block, b_block, block_span; \
1410 add block_ptr_b, block_ptr_a, #16; \
1412 vshrn.u32 u_whole_low, u_block, #16; \
1413 vshrn.u32 v_whole_low, v_block, #16; \
1414 vshrn.u32 r_whole_low, r_block, #16; \
1415 vshrn.u32 g_whole_low, g_block, #16; \
1417 vdup.u32 dx4, uv_dx4[0]; \
1418 vshrn.u32 b_whole_low, b_block, #16; \
1420 vaddhn.u32 u_whole_high, u_block, dx4; \
1421 vdup.u32 dx4, uv_dx4[1]; \
1423 vaddhn.u32 v_whole_high, v_block, dx4; \
1424 vdup.u32 dx4, rg_dx4[0]; \
1426 vaddhn.u32 r_whole_high, r_block, dx4; \
1427 vdup.u32 dx4, rg_dx4[1]; \
1429 vaddhn.u32 g_whole_high, g_block, dx4; \
1430 vdup.u32 dx4, b_dx4; \
1432 vaddhn.u32 b_whole_high, b_block, dx4; \
1433 vdup.u32 dx8, uv_dx8[0]; \
1435 vadd.u32 u_block, u_block, dx8; \
1436 vdup.u32 dx8, uv_dx8[1]; \
1438 vadd.u32 v_block, v_block, dx8; \
1439 vdup.u32 dx8, rg_dx8[0]; \
1441 vadd.u32 r_block, r_block, dx8; \
1442 vdup.u32 dx8, rg_dx8[1]; \
1444 vadd.u32 g_block, g_block, dx8; \
1445 vdup.u32 dx8, b_dx8; \
1447 vadd.u32 b_block, b_block, dx8; \
1448 vmovn.u16 u_whole_8, u_whole; \
1450 vmovn.u16 v_whole_8, v_whole; \
1452 vmovn.u16 b_whole_8, b_whole; \
1454 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1456 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1457 setup_blocks_texture_##swizzling(); \
1459 vmovn.u16 r_whole_8, r_whole; \
1463 vmovn.u16 g_whole_8, g_whole; \
1464 vshrn.u32 u_whole_low, u_block, #16; \
1466 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1467 vshrn.u32 v_whole_low, v_block, #16; \
1469 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1470 vshrn.u32 r_whole_low, r_block, #16; \
1472 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1473 vshrn.u32 g_whole_low, g_block, #16; \
1475 vdup.u32 dx4, uv_dx4[0]; \
1476 vshrn.u32 b_whole_low, b_block, #16; \
1478 vaddhn.u32 u_whole_high, u_block, dx4; \
1479 vdup.u32 dx4, uv_dx4[1]; \
1481 vaddhn.u32 v_whole_high, v_block, dx4; \
1482 vdup.u32 dx4, rg_dx4[0]; \
1484 vaddhn.u32 r_whole_high, r_block, dx4; \
1485 vdup.u32 dx4, rg_dx4[1]; \
1487 vaddhn.u32 g_whole_high, g_block, dx4; \
1488 vdup.u32 dx4, b_dx4; \
1490 vaddhn.u32 b_whole_high, b_block, dx4; \
1491 vdup.u32 dx8, uv_dx8[0]; \
1493 vadd.u32 u_block, u_block, dx8; \
1494 vdup.u32 dx8, uv_dx8[1]; \
1496 vadd.u32 v_block, v_block, dx8; \
1497 vdup.u32 dx8, rg_dx8[0]; \
1499 vadd.u32 r_block, r_block, dx8; \
1500 vdup.u32 dx8, rg_dx8[1]; \
1502 vadd.u32 g_block, g_block, dx8; \
1503 vdup.u32 dx8, b_dx8; \
1505 vadd.u32 b_block, b_block, dx8; \
1506 vmovn.u16 u_whole_8, u_whole; \
1508 add fb_ptr, fb_ptr, #16; \
1509 vmovn.u16 v_whole_8, v_whole; \
1511 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1512 vmovn.u16 b_whole_8, b_whole; \
1516 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1517 subs span_num_blocks, span_num_blocks, #1; \
1519 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1520 setup_blocks_texture_##swizzling(); \
1522 vmovn.u16 r_whole_8, r_whole; \
1526 vmovn.u16 g_whole_8, g_whole; \
1527 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1529 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1530 vdup.u8 draw_mask, right_mask; \
1532 vmov.u32 fb_mask_ptrs[0], right_mask; \
1533 vtst.u16 draw_mask, draw_mask, test_mask; \
1534 vzip.u8 u_whole_8, v_whole_8; \
1536 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1537 vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \
1538 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1539 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1540 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1543 add span_uvrg_offset, span_uvrg_offset, #16; \
1544 add span_b_offset, span_b_offset, #4; \
1546 add span_edge_data, span_edge_data, #8; \
1547 subs num_spans, num_spans, #1; \
1549 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1552 ldmia sp!, { r4 - r11, pc }; \
1555 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1556 vpush { texture_mask }; \
1557 vpush { uvrg_dx4 }; \
1559 stmdb sp!, { r0 - r3, r12, r14 }; \
1560 bl flush_render_block_buffer; \
1561 ldmia sp!, { r0 - r3, r12, r14 }; \
1563 vpop { uvrg_dx4 }; \
1564 vpop { texture_mask }; \
1566 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1567 vmov.u8 fb_mask_ptrs, #0; \
1569 mov num_blocks, span_num_blocks; \
1570 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1574 setup_blocks_shaded_textured_builder(swizzled)
1575 setup_blocks_shaded_textured_builder(unswizzled)
1578 #define setup_blocks_unshaded_textured_builder(swizzling) \
1581 function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \
1582 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
1583 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1585 vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \
1586 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1588 cmp num_spans, #0; \
1591 stmdb sp!, { r4 - r11, r14 }; \
1592 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1594 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1596 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \
1597 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1599 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1600 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1602 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1604 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1607 vmov.u8 fb_mask_ptrs, #0; \
1609 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
1610 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1612 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
1613 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
1615 cmp span_num_blocks, #0; \
1618 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
1619 add num_blocks, span_num_blocks, num_blocks; \
1621 cmp num_blocks, #MAX_BLOCKS; \
1625 add fb_ptr, fb_ptr, y, lsl #11; \
1627 vdup.u32 v_left_x, left_x; \
1630 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
1631 add fb_ptr, fb_ptr, left_x, lsl #1; \
1633 and dither_shift, left_x, #0x03; \
1635 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
1636 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1638 mov dither_shift, dither_shift, lsl #3; \
1639 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1642 subs span_num_blocks, span_num_blocks, #1; \
1644 mov dither_row, dither_row, ror dither_shift; \
1646 vdup.u32 dither_offsets_short, dither_row; \
1647 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1649 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1651 vdup.u32 u_block, uv[0]; \
1653 vdup.u32 v_block, uv[1]; \
1654 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1656 vadd.u32 u_block, u_block, block_span; \
1657 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
1659 vadd.u32 v_block, v_block, block_span; \
1660 add block_ptr_b, block_ptr_a, #16; \
1662 vshrn.u32 u_whole_low, u_block, #16; \
1663 vshrn.u32 v_whole_low, v_block, #16; \
1665 vdup.u32 dx4, uv_dx4[0]; \
1667 vaddhn.u32 u_whole_high, u_block, dx4; \
1668 vdup.u32 dx4, uv_dx4[1]; \
1670 vaddhn.u32 v_whole_high, v_block, dx4; \
1671 vdup.u32 dx8, uv_dx8[0]; \
1673 vadd.u32 u_block, u_block, dx8; \
1674 vdup.u32 dx8, uv_dx8[1]; \
1676 vadd.u32 v_block, v_block, dx8; \
1677 vmovn.u16 u_whole_8, u_whole; \
1679 vmovn.u16 v_whole_8, v_whole; \
1682 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1684 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1685 setup_blocks_texture_##swizzling(); \
1690 vshrn.u32 u_whole_low, u_block, #16; \
1692 vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1693 vshrn.u32 v_whole_low, v_block, #16; \
1695 add block_ptr_b, block_ptr_b, #32; \
1696 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1698 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 add fb_ptr, fb_ptr, #16; \
1712 vmovn.u16 v_whole_8, v_whole; \
1714 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1717 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1718 subs span_num_blocks, span_num_blocks, #1; \
1720 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1721 setup_blocks_texture_##swizzling(); \
1726 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
1728 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
1729 vdup.u8 draw_mask, right_mask; \
1731 vmov.u32 fb_mask_ptrs[0], right_mask; \
1732 vtst.u16 draw_mask, draw_mask, test_mask; \
1733 vzip.u8 u_whole_8, v_whole_8; \
1735 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1736 add block_ptr_b, block_ptr_b, #32; \
1737 vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \
1738 vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \
1739 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \
1742 add span_uvrg_offset, span_uvrg_offset, #16; \
1743 add span_edge_data, span_edge_data, #8; \
1744 subs num_spans, num_spans, #1; \
1746 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
1749 ldmia sp!, { r4 - r11, pc }; \
1752 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1753 vpush { texture_mask }; \
1754 vpush { uvrg_dx4 }; \
1756 stmdb sp!, { r0 - r3, r12, r14 }; \
1757 bl flush_render_block_buffer; \
1758 ldmia sp!, { r0 - r3, r12, r14 }; \
1760 vpop { uvrg_dx4 }; \
1761 vpop { texture_mask }; \
1763 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1764 vmov.u8 fb_mask_ptrs, #0; \
1766 mov num_blocks, span_num_blocks; \
1767 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1771 setup_blocks_unshaded_textured_builder(swizzled)
1772 setup_blocks_unshaded_textured_builder(unswizzled)
1777 function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect)
1778 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1779 veor.u32 draw_mask, draw_mask, draw_mask
1784 stmdb sp!, { r4 - r11, r14 }
1785 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1787 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1789 ubfx color_r, color, #3, #5
1790 ubfx color_g, color, #11, #5
1791 ubfx color_b, color, #19, #5
1793 orr color, color_r, color_b, lsl #10
1794 orr color, color, color_g, lsl #5
1796 vdup.u16 colors, color
1798 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1799 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1801 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1802 add block_ptr_a, block_ptr_a, num_blocks, lsl #6
1805 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1806 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1808 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
1810 cmp span_num_blocks, #0
1813 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1814 add num_blocks, span_num_blocks, num_blocks
1816 cmp num_blocks, #MAX_BLOCKS
1820 add fb_ptr, fb_ptr, y, lsl #11
1823 add fb_ptr, fb_ptr, left_x, lsl #1
1826 subs span_num_blocks, span_num_blocks, #1
1828 add block_ptr_b, block_ptr_a, #16
1831 vmov.u32 fb_mask_ptrs[1], fb_ptr
1835 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_32
1836 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1837 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1839 add fb_ptr, fb_ptr, #16
1840 add block_ptr_b, block_ptr_b, #32
1844 vmov.u32 fb_mask_ptrs[1], fb_ptr
1845 subs span_num_blocks, span_num_blocks, #1
1850 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1852 vdup.u8 draw_mask_edge, right_mask
1853 vtst.u16 draw_mask_edge, draw_mask_edge, test_mask
1855 vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32
1856 vst1.u32 { draw_mask_edge }, [ block_ptr_a, :128 ], c_32
1857 add block_ptr_b, block_ptr_b, #32
1858 vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32
1861 add span_edge_data, span_edge_data, #8
1862 subs num_spans, num_spans, #1
1864 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
1867 ldmia sp!, { r4 - r11, pc }
1872 stmdb sp!, { r0 - r3, r12, r14 }
1873 bl flush_render_block_buffer
1874 ldmia sp!, { r0 - r3, r12, r14 }
1878 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
1879 veor.u32 draw_mask, draw_mask, draw_mask
1881 mov num_blocks, span_num_blocks
1882 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1886 #define mask_msb_scalar r14
1888 #define msb_mask q15
1890 #define pixels_low d16
1892 #define msb_mask_low d30
1893 #define msb_mask_high d31
1898 function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct)
1899 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]
1904 stmdb sp!, { r4 - r11, r14 }
1906 ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ]
1908 ubfx color_r, color, #3, #5
1909 ubfx color_g, color, #11, #5
1911 ldrh mask_msb_scalar, [ psx_gpu, #psx_gpu_mask_msb_offset ]
1912 ubfx color_b, color, #19, #5
1914 orr color, color_r, color_b, lsl #10
1915 orr color, color, color_g, lsl #5
1916 orr color, color, mask_msb_scalar
1918 vdup.u16 colors, color
1920 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1921 orr color, color, lsl #16
1925 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]
1926 ldrh y, [ span_edge_data, #edge_data_y_offset ]
1928 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
1930 cmp span_num_blocks, #0
1933 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]
1935 add fb_ptr, fb_ptr, y, lsl #11
1936 subs span_num_blocks, span_num_blocks, #1
1938 add fb_ptr, fb_ptr, left_x, lsl #1
1942 vst1.u32 { colors }, [ fb_ptr ]!
1943 subs span_num_blocks, span_num_blocks, #1
1948 ldrb right_mask, [ span_edge_data, #edge_data_right_mask_offset ]
1950 cmp right_mask, #0x0
1953 tst right_mask, #0xF
1954 streq color, [ fb_ptr ], #4
1955 moveq right_mask, right_mask, lsr #4
1956 streq color, [ fb_ptr ], #4
1958 tst right_mask, #0x3
1959 streq color, [ fb_ptr ], #4
1960 moveq right_mask, right_mask, lsr #2
1962 tst right_mask, #0x1
1963 streqh color, [ fb_ptr ]
1966 add span_edge_data, span_edge_data, #8
1967 subs num_spans, num_spans, #1
1970 ldmia sp!, { r4 - r11, pc }
1973 vst1.u32 { colors }, [ fb_ptr ]
1980 #define rg_dx_ptr r2
1998 #undef dither_offsets
2018 #define r_whole_low d6
2019 #define r_whole_high d7
2020 #define g_whole_low d8
2021 #define g_whole_high d9
2022 #define b_whole_low d10
2023 #define b_whole_high d11
2025 #define gb_whole_8 q6
2027 #define g_whole_8 d12
2028 #define b_whole_8 d13
2030 #define r_whole_8 d14
2042 #define block_span q5
2050 #define d128_0x7 q13
2054 #define dither_offsets q14
2055 #define draw_mask q15
2057 #define dither_offsets_low d28
2060 #define test_mask q10
2063 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2064 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2065 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2067 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2068 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2069 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2071 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2073 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2076 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2079 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2080 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2081 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2083 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2085 cmp num_spans, #0; \
2088 stmdb sp!, { r4 - r11, r14 }; \
2089 vshl.u32 rg_dx4, rg_dx, #2; \
2091 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2092 vshl.u32 rg_dx8, rg_dx, #3; \
2094 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2096 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2097 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2099 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2100 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2102 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2103 vmov.u8 d64_1, #1; \
2105 vmov.u8 d128_4, #4; \
2106 vmov.u8 d64_128, #128; \
2108 vmov.u8 d128_0x7, #0x7; \
2111 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2112 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2114 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2115 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
2117 cmp span_num_blocks, #0; \
2120 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2121 add num_blocks, span_num_blocks, num_blocks; \
2123 cmp num_blocks, #MAX_BLOCKS; \
2127 ldr b, [ span_b_offset ]; \
2128 add fb_ptr, fb_ptr, y, lsl #11; \
2130 vdup.u32 v_left_x, left_x; \
2133 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2134 add fb_ptr, fb_ptr, left_x, lsl #1; \
2136 mla b, b_dx, left_x, b; \
2137 and dither_shift, left_x, #0x03; \
2139 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2140 vshr.u32 rg_dx, rg_dx4, #2; \
2142 mov dither_shift, dither_shift, lsl #3; \
2143 vmla.u32 rg, rg_dx, v_left_x; \
2146 subs span_num_blocks, span_num_blocks, #1; \
2148 mov dither_row, dither_row, ror dither_shift; \
2149 mov b_dx4, b_dx, lsl #2; \
2151 vdup.u32 dither_offsets, dither_row; \
2152 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2154 vdup.u32 b_block, b; \
2155 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2157 mov b_dx8, b_dx, lsl #3; \
2158 vdup.u32 r_block, rg[0]; \
2159 vdup.u32 g_block, rg[1]; \
2161 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2163 vadd.u32 r_block, r_block, block_span; \
2164 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2166 vadd.u32 g_block, g_block, block_span; \
2167 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2169 vadd.u32 b_block, b_block, block_span; \
2170 add block_ptr_b, block_ptr_a, #16; \
2172 vshrn.u32 r_whole_low, r_block, #16; \
2173 vshrn.u32 g_whole_low, g_block, #16; \
2174 vshrn.u32 b_whole_low, b_block, #16; \
2175 vdup.u32 dx4, rg_dx4[0]; \
2177 vaddhn.u32 r_whole_high, r_block, dx4; \
2178 vdup.u32 dx4, rg_dx4[1]; \
2180 vaddhn.u32 g_whole_high, g_block, dx4; \
2181 vdup.u32 dx4, b_dx4; \
2183 vaddhn.u32 b_whole_high, b_block, dx4; \
2184 vdup.u32 dx8, rg_dx8[0]; \
2186 vadd.u32 r_block, r_block, dx8; \
2187 vdup.u32 dx8, rg_dx8[1]; \
2189 vadd.u32 g_block, g_block, dx8; \
2190 vdup.u32 dx8, b_dx8; \
2192 vadd.u32 b_block, b_block, dx8; \
2194 vmovn.u16 r_whole_8, r_whole; \
2195 vmovn.u16 g_whole_8, g_whole; \
2196 vmovn.u16 b_whole_8, b_whole; \
2199 veor.u32 draw_mask, draw_mask, draw_mask; \
2202 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2203 vshrn.u32 r_whole_low, r_block, #16; \
2205 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2206 vshrn.u32 g_whole_low, g_block, #16; \
2208 vshrn.u32 b_whole_low, b_block, #16; \
2209 str fb_ptr, [ block_ptr_a, #44 ]; \
2211 vdup.u32 dx4, rg_dx4[0]; \
2212 vshr.u8 r_whole_8, r_whole_8, #3; \
2213 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2215 vaddhn.u32 r_whole_high, r_block, dx4; \
2216 vdup.u32 dx4, rg_dx4[1]; \
2218 vaddhn.u32 g_whole_high, g_block, dx4; \
2219 vdup.u32 dx4, b_dx4; \
2221 vaddhn.u32 b_whole_high, b_block, dx4; \
2222 vdup.u32 dx8, rg_dx8[0]; \
2224 vmull.u8 pixels, r_whole_8, d64_1; \
2225 vmlal.u8 pixels, g_whole_8, d64_4; \
2226 vmlal.u8 pixels, b_whole_8, d64_128; \
2228 vadd.u32 r_block, r_block, dx8; \
2229 vdup.u32 dx8, rg_dx8[1]; \
2231 vadd.u32 g_block, g_block, dx8; \
2232 vdup.u32 dx8, b_dx8; \
2234 vadd.u32 b_block, b_block, dx8; \
2235 add fb_ptr, fb_ptr, #16; \
2237 vmovn.u16 r_whole_8, r_whole; \
2238 vmovn.u16 g_whole_8, g_whole; \
2239 vmovn.u16 b_whole_8, b_whole; \
2241 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2242 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2246 subs span_num_blocks, span_num_blocks, #1; \
2250 str fb_ptr, [ block_ptr_a, #44 ]; \
2251 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2253 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2254 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2256 vshr.u8 r_whole_8, r_whole_8, #3; \
2257 vdup.u8 draw_mask, right_mask; \
2259 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2260 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
2262 vtst.u16 draw_mask, draw_mask, test_mask; \
2264 vmull.u8 pixels, r_whole_8, d64_1; \
2265 vmlal.u8 pixels, g_whole_8, d64_4; \
2266 vmlal.u8 pixels, b_whole_8, d64_128; \
2268 vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \
2269 vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \
2272 add span_uvrg_offset, span_uvrg_offset, #16; \
2273 add span_b_offset, span_b_offset, #4; \
2275 add span_edge_data, span_edge_data, #8; \
2276 subs num_spans, num_spans, #1; \
2278 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
2281 ldmia sp!, { r4 - r11, pc }; \
2284 /* TODO: Load from psx_gpu instead of saving/restoring these */\
2287 stmdb sp!, { r0 - r3, r12, r14 }; \
2288 bl flush_render_block_buffer; \
2289 ldmia sp!, { r0 - r3, r12, r14 }; \
2293 vmov.u8 d64_1, #1; \
2294 vmov.u8 d128_4, #4; \
2295 vmov.u8 d64_128, #128; \
2296 vmov.u8 d128_0x7, #0x7; \
2298 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2300 mov num_blocks, span_num_blocks; \
2301 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2305 setup_blocks_shaded_untextured_indirect_builder(undithered)
2306 setup_blocks_shaded_untextured_indirect_builder(dithered)
2311 #define mask_msb_ptr r14
2313 #define draw_mask q0
2314 #define pixels_low d16
2315 #define pixels_high d17
2319 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2322 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2323 ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \
2324 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2326 vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \
2328 cmp num_spans, #0; \
2331 stmdb sp!, { r4 - r11, r14 }; \
2332 vshl.u32 rg_dx4, rg_dx, #2; \
2334 ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \
2335 vshl.u32 rg_dx8, rg_dx, #3; \
2337 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2338 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2340 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2341 vmov.u8 d64_1, #1; \
2343 vmov.u8 d128_4, #4; \
2344 vmov.u8 d64_128, #128; \
2346 vmov.u8 d128_0x7, #0x7; \
2347 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2348 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
2351 ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \
2352 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2354 ldrh y, [ span_edge_data, #edge_data_y_offset ]; \
2355 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
2357 cmp span_num_blocks, #0; \
2360 ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \
2361 add fb_ptr, fb_ptr, y, lsl #11; \
2363 ldr b, [ span_b_offset ]; \
2364 vdup.u32 v_left_x, left_x; \
2367 ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \
2368 add fb_ptr, fb_ptr, left_x, lsl #1; \
2370 mla b, b_dx, left_x, b; \
2371 and dither_shift, left_x, #0x03; \
2373 vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \
2374 vshr.u32 rg_dx, rg_dx4, #2; \
2376 mov dither_shift, dither_shift, lsl #3; \
2377 vmla.u32 rg, rg_dx, v_left_x; \
2379 subs span_num_blocks, span_num_blocks, #1; \
2381 mov dither_row, dither_row, ror dither_shift; \
2382 mov b_dx4, b_dx, lsl #2; \
2384 vdup.u32 dither_offsets, dither_row; \
2385 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2387 vdup.u32 b_block, b; \
2388 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2390 mov b_dx8, b_dx, lsl #3; \
2391 vdup.u32 r_block, rg[0]; \
2392 vdup.u32 g_block, rg[1]; \
2394 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2396 vadd.u32 r_block, r_block, block_span; \
2397 vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \
2399 vadd.u32 g_block, g_block, block_span; \
2400 vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \
2402 vadd.u32 b_block, b_block, block_span; \
2403 add block_ptr_b, block_ptr_a, #16; \
2405 vshrn.u32 r_whole_low, r_block, #16; \
2406 vshrn.u32 g_whole_low, g_block, #16; \
2407 vshrn.u32 b_whole_low, b_block, #16; \
2408 vdup.u32 dx4, rg_dx4[0]; \
2410 vaddhn.u32 r_whole_high, r_block, dx4; \
2411 vdup.u32 dx4, rg_dx4[1]; \
2413 vaddhn.u32 g_whole_high, g_block, dx4; \
2414 vdup.u32 dx4, b_dx4; \
2416 vaddhn.u32 b_whole_high, b_block, dx4; \
2417 vdup.u32 dx8, rg_dx8[0]; \
2419 vadd.u32 r_block, r_block, dx8; \
2420 vdup.u32 dx8, rg_dx8[1]; \
2422 vadd.u32 g_block, g_block, dx8; \
2423 vdup.u32 dx8, b_dx8; \
2425 vadd.u32 b_block, b_block, dx8; \
2427 vmovn.u16 r_whole_8, r_whole; \
2428 vmovn.u16 g_whole_8, g_whole; \
2429 vmovn.u16 b_whole_8, b_whole; \
2434 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2435 vshrn.u32 r_whole_low, r_block, #16; \
2437 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2438 vshrn.u32 g_whole_low, g_block, #16; \
2440 vshrn.u32 b_whole_low, b_block, #16; \
2442 vdup.u32 dx4, rg_dx4[0]; \
2443 vshr.u8 r_whole_8, r_whole_8, #3; \
2444 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2446 vaddhn.u32 r_whole_high, r_block, dx4; \
2447 vdup.u32 dx4, rg_dx4[1]; \
2449 vmov pixels, msb_mask; \
2450 vaddhn.u32 g_whole_high, g_block, dx4; \
2451 vdup.u32 dx4, b_dx4; \
2453 vaddhn.u32 b_whole_high, b_block, dx4; \
2454 vdup.u32 dx8, rg_dx8[0]; \
2456 vmlal.u8 pixels, r_whole_8, d64_1; \
2457 vmlal.u8 pixels, g_whole_8, d64_4; \
2458 vmlal.u8 pixels, b_whole_8, d64_128; \
2460 vadd.u32 r_block, r_block, dx8; \
2461 vdup.u32 dx8, rg_dx8[1]; \
2463 vadd.u32 g_block, g_block, dx8; \
2464 vdup.u32 dx8, b_dx8; \
2466 vadd.u32 b_block, b_block, dx8; \
2468 vmovn.u16 r_whole_8, r_whole; \
2469 vmovn.u16 g_whole_8, g_whole; \
2470 vmovn.u16 b_whole_8, b_whole; \
2472 vst1.u32 { pixels }, [ fb_ptr ]!; \
2473 subs span_num_blocks, span_num_blocks, #1; \
2477 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2479 ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \
2480 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2482 vshr.u8 r_whole_8, r_whole_8, #3; \
2483 rbit right_mask, right_mask; \
2484 vmov pixels, msb_mask; \
2485 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2486 clz right_mask, right_mask; \
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 ldr pc, [ pc, right_mask, lsl #2 ]; \
2505 vst1.u16 { pixels_low[0] }, [ fb_ptr ]; \
2509 vst1.u32 { pixels_low[0] }, [ fb_ptr ]; \
2513 vst1.u32 { pixels_low[0] }, [ fb_ptr ]!; \
2514 vst1.u16 { pixels_low[2] }, [ fb_ptr ]; \
2518 vst1.u32 { pixels_low }, [ fb_ptr ]; \
2522 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2523 vst1.u16 { pixels_high[0] }, [ fb_ptr ]; \
2527 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2528 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2532 vst1.u32 { pixels_low }, [ fb_ptr ]!; \
2533 vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \
2534 vst1.u16 { pixels_high[2] }, [ fb_ptr ]; \
2538 vst1.u32 { pixels }, [ fb_ptr ]; \
2542 add span_uvrg_offset, span_uvrg_offset, #16; \
2543 add span_b_offset, span_b_offset, #4; \
2545 add span_edge_data, span_edge_data, #8; \
2546 subs num_spans, num_spans, #1; \
2550 ldmia sp!, { r4 - r11, pc } \
2552 setup_blocks_shaded_untextured_direct_builder(undithered)
2553 setup_blocks_shaded_untextured_direct_builder(dithered)
2562 #define block_ptr r1
2563 #define num_blocks r2
2576 #define texture_ptr r11
2590 #define pixels_d r10
2594 #define clut_ptr r12
2595 #define current_texture_mask r5
2596 #define dirty_textures_mask r6
2600 #define clut_low_a d2
2601 #define clut_low_b d3
2602 #define clut_high_a d4
2603 #define clut_high_b d5
2608 #define texels_low d6
2609 #define texels_high d7
2613 function(texture_blocks_untextured)
2619 function(texture_blocks_4bpp)
2620 stmdb sp!, { r3 - r11, r14 }
2621 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2623 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2624 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2626 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2627 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]
2629 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2630 vuzp.u8 clut_a, clut_b
2632 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
2633 tst dirty_textures_mask, current_texture_mask
2639 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2641 uxtah uv_0, texture_ptr, uv_01
2642 uxtah uv_1, texture_ptr, uv_01, ror #16
2644 uxtah uv_2, texture_ptr, uv_23
2645 uxtah uv_3, texture_ptr, uv_23, ror #16
2647 uxtah uv_4, texture_ptr, uv_45
2648 ldrb pixel_0, [ uv_0 ]
2650 uxtah uv_5, texture_ptr, uv_45, ror #16
2651 ldrb pixel_1, [ uv_1 ]
2653 uxtah uv_6, texture_ptr, uv_67
2654 ldrb pixel_2, [ uv_2 ]
2656 uxtah uv_7, texture_ptr, uv_67, ror #16
2657 ldrb pixel_3, [ uv_3 ]
2659 ldrb pixel_4, [ uv_4 ]
2660 subs num_blocks, num_blocks, #1
2662 ldrb pixel_5, [ uv_5 ]
2663 orr pixels_a, pixel_0, pixel_1, lsl #8
2665 ldrb pixel_6, [ uv_6 ]
2666 orr pixels_b, pixel_4, pixel_5, lsl #8
2668 ldrb pixel_7, [ uv_7 ]
2669 orr pixels_a, pixels_a, pixel_2, lsl #16
2671 orr pixels_b, pixels_b, pixel_6, lsl #16
2672 orr pixels_a, pixels_a, pixel_3, lsl #24
2674 orr pixels_b, pixels_b, pixel_7, lsl #24
2675 vmov.u32 texels, pixels_a, pixels_b
2677 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2678 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2680 vst2.u8 { texels_low, texels_high }, [ block_ptr, :128 ], c_64
2683 ldmia sp!, { r3 - r11, pc }
2686 stmdb sp!, { r1 - r2 }
2687 bl update_texture_4bpp_cache
2690 ldmia sp!, { r1 - r2 }
2696 function(texture_blocks_8bpp)
2697 stmdb sp!, { r3 - r11, r14 }
2698 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2700 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2701 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2703 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
2704 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
2706 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]
2707 tst dirty_textures_mask, current_texture_mask
2713 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2715 uxtah uv_0, texture_ptr, uv_01
2716 uxtah uv_1, texture_ptr, uv_01, ror #16
2718 uxtah uv_2, texture_ptr, uv_23
2719 uxtah uv_3, texture_ptr, uv_23, ror #16
2721 uxtah uv_4, texture_ptr, uv_45
2722 ldrb pixel_0, [ uv_0 ]
2724 uxtah uv_5, texture_ptr, uv_45, ror #16
2725 ldrb pixel_1, [ uv_1 ]
2727 uxtah uv_6, texture_ptr, uv_67
2728 ldrb pixel_2, [ uv_2 ]
2730 uxtah uv_7, texture_ptr, uv_67, ror #16
2731 ldrb pixel_3, [ uv_3 ]
2733 ldrb pixel_4, [ uv_4 ]
2734 add pixel_0, pixel_0, pixel_0
2736 ldrb pixel_5, [ uv_5 ]
2737 add pixel_1, pixel_1, pixel_1
2739 ldrb pixel_6, [ uv_6 ]
2740 add pixel_2, pixel_2, pixel_2
2742 ldrb pixel_7, [ uv_7 ]
2743 add pixel_3, pixel_3, pixel_3
2745 ldrh pixel_0, [ clut_ptr, pixel_0 ]
2746 add pixel_4, pixel_4, pixel_4
2748 ldrh pixel_1, [ clut_ptr, pixel_1 ]
2749 add pixel_5, pixel_5, pixel_5
2751 ldrh pixel_2, [ clut_ptr, pixel_2 ]
2752 add pixel_6, pixel_6, pixel_6
2754 ldrh pixel_3, [ clut_ptr, pixel_3 ]
2755 add pixel_7, pixel_7, pixel_7
2757 ldrh pixel_4, [ clut_ptr, pixel_4 ]
2758 orr pixels_a, pixel_0, pixel_1, lsl #16
2760 ldrh pixel_5, [ clut_ptr, pixel_5 ]
2761 orr pixels_c, pixel_2, pixel_3, lsl #16
2763 ldrh pixel_6, [ clut_ptr, pixel_6 ]
2764 subs num_blocks, num_blocks, #1
2766 ldrh pixel_7, [ clut_ptr, pixel_7 ]
2767 orr pixels_b, pixel_4, pixel_5, lsl #16
2769 orr pixels_d, pixel_6, pixel_7, lsl #16
2770 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2772 add block_ptr, block_ptr, #64
2775 ldmia sp!, { r3 - r11, pc }
2778 stmdb sp!, { r1 - r2, r12 }
2780 bl update_texture_8bpp_cache
2782 ldmia sp!, { r1 - r2, r12 }
2812 #define block_ptr r1
2813 #define num_blocks r2
2857 #define texture_ptr r12
2862 function(texture_blocks_16bpp)
2863 stmdb sp!, { r3 - r11, r14 }
2864 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2866 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
2867 ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
2870 ldrh uv_0, [ block_ptr ]
2871 subs num_blocks, num_blocks, #1
2873 ldrh uv_1, [ block_ptr, #2 ]
2875 and v_0, uv_0, #0xFF00
2876 and v_1, uv_1, #0xFF00
2878 and u_0, uv_0, #0xFF
2879 and u_1, uv_1, #0xFF
2881 add uv_0, u_0, v_0, lsl #2
2882 ldrh uv_2, [ block_ptr, #4 ]
2884 add uv_1, u_1, v_1, lsl #2
2885 ldrh uv_3, [ block_ptr, #6 ]
2887 add uv_0, uv_0, uv_0
2888 add uv_1, uv_1, uv_1
2890 and v_2, uv_2, #0xFF00
2891 and v_3, uv_3, #0xFF00
2893 and u_2, uv_2, #0xFF
2894 and u_3, uv_3, #0xFF
2896 add uv_2, u_2, v_2, lsl #2
2897 ldrh uv_4, [ block_ptr, #8 ]
2899 add uv_3, u_3, v_3, lsl #2
2900 ldrh uv_5, [ block_ptr, #10 ]
2902 add uv_2, uv_2, uv_2
2903 add uv_3, uv_3, uv_3
2905 and v_4, uv_4, #0xFF00
2906 and v_5, uv_5, #0xFF00
2908 and u_4, uv_4, #0xFF
2909 and u_5, uv_5, #0xFF
2911 add uv_4, u_4, v_4, lsl #2
2912 ldrh uv_6, [ block_ptr, #12 ]
2914 add uv_5, u_5, v_5, lsl #2
2915 ldrh uv_7, [ block_ptr, #14 ]
2917 add uv_4, uv_4, uv_4
2918 ldrh pixel_0, [ texture_ptr, uv_0 ]
2920 add uv_5, uv_5, uv_5
2921 ldrh pixel_1, [ texture_ptr, uv_1 ]
2923 and v_6, uv_6, #0xFF00
2924 ldrh pixel_2, [ texture_ptr, uv_2 ]
2926 and v_7, uv_7, #0xFF00
2927 ldrh pixel_3, [ texture_ptr, uv_3 ]
2929 and u_6, uv_6, #0xFF
2930 ldrh pixel_4, [ texture_ptr, uv_4 ]
2932 and u_7, uv_7, #0xFF
2933 ldrh pixel_5, [ texture_ptr, uv_5 ]
2935 add uv_6, u_6, v_6, lsl #2
2936 add uv_7, u_7, v_7, lsl #2
2938 add uv_6, uv_6, uv_6
2939 add uv_7, uv_7, uv_7
2941 orr pixels_a, pixel_0, pixel_1, lsl #16
2942 orr pixels_b, pixel_2, pixel_3, lsl #16
2944 ldrh pixel_6, [ texture_ptr, uv_6 ]
2945 orr pixels_c, pixel_4, pixel_5, lsl #16
2947 ldrh pixel_7, [ texture_ptr, uv_7 ]
2948 orr pixels_d, pixel_6, pixel_7, lsl #16
2950 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
2951 add block_ptr, block_ptr, #64
2955 ldmia sp!, { r3 - r11, pc }
2970 #undef msb_mask_high
2978 #define num_blocks r1
2979 #define color_ptr r2
2980 #define colors_scalar r2
2981 #define colors_scalar_compare r3
2982 #define mask_msb_ptr r2
2984 #define block_ptr_load_a r0
2985 #define block_ptr_store r3
2986 #define block_ptr_load_b r12
2991 #define draw_mask_bits_scalar r5
2993 #define d128_0x07 q0
2994 #define d128_0x1F q1
2995 #define d128_0x8000 q2
2996 #define test_mask q3
2998 #define colors_rg q5
2999 #define colors_b_dm_bits q6
3000 #define texels_rg q7
3003 #define pixels_b q10
3005 #define zero_mask q4
3006 #define draw_mask q12
3007 #define msb_mask q13
3009 #define fb_pixels q8
3011 #define pixels_gb_low q9
3013 #define colors_r d10
3014 #define colors_g d11
3015 #define colors_b d12
3016 #define draw_mask_bits d13
3017 #define texels_r d14
3018 #define texels_g d15
3019 #define pixels_r_low d16
3020 #define pixels_g_low d18
3021 #define pixels_b_low d19
3022 #define msb_mask_low d26
3023 #define msb_mask_high d27
3028 #define texels_b d31
3030 #define shade_blocks_textured_modulated_prologue_indirect() \
3032 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3034 #define shade_blocks_textured_modulated_prologue_direct() \
3035 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3036 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ] \
3039 #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \
3041 #define shade_blocks_textured_false_modulation_check_undithered(target) \
3042 ldr colors_scalar, [ psx_gpu, #psx_gpu_triangle_color_offset ]; \
3043 movw colors_scalar_compare, #0x8080; \
3045 movt colors_scalar_compare, #0x80; \
3046 cmp colors_scalar, colors_scalar_compare; \
3047 beq shade_blocks_textured_unmodulated_##target \
3049 #define shade_blocks_textured_false_modulation_check_dithered(target) \
3051 #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \
3052 shade_blocks_textured_false_modulation_check_##dithering(target); \
3053 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3054 vld1.u32 { colors_r[] }, [ color_ptr, :32 ]; \
3055 vdup.u8 colors_g, colors_r[1]; \
3056 vdup.u8 colors_b, colors_r[2]; \
3057 vdup.u8 colors_r, colors_r[0] \
3060 #define shade_blocks_textured_modulated_load_dithered(target) \
3061 vld1.u32 { target }, [ block_ptr_load_b, :128 ] \
3063 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3064 vld1.u32 { target }, [ block_ptr_load_b, :128 ], c_32 \
3066 #define shade_blocks_textured_modulated_load_undithered(target) \
3068 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3069 add block_ptr_load_b, block_ptr_load_b, #32 \
3071 #define shade_blocks_textured_modulate_dithered(channel) \
3072 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3074 #define shade_blocks_textured_modulate_undithered(channel) \
3075 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3078 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3079 vst1.u32 { draw_mask }, [ block_ptr_store, :128 ]! \
3081 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3082 ldr fb_ptr, [ block_ptr_load_b, #(offset - 64) ]; \
3083 vld1.u32 { fb_pixels }, [ fb_ptr ]; \
3084 vbit.u16 pixels, fb_pixels, draw_mask \
3086 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3087 vst1.u32 { pixels }, [ block_ptr_store, :128 ], c_48 \
3089 #define shade_blocks_textured_modulated_store_pixels_direct() \
3090 vst1.u32 { pixels }, [ fb_ptr ] \
3093 #define shade_blocks_textured_modulated_load_rg_shaded() \
3094 vld1.u32 { colors_r, colors_g }, [ block_ptr_load_b, :128 ], c_32 \
3096 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3097 add block_ptr_load_b, block_ptr_load_b, #32 \
3099 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3100 vld1.u32 { colors_b, draw_mask_bits }, [ block_ptr_load_a, :128 ], c_32 \
3102 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3103 ldr draw_mask_bits_scalar, [ block_ptr_load_a, #8 ]; \
3104 add block_ptr_load_a, block_ptr_load_a, #32 \
3106 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3107 vdup.u16 draw_mask, draw_mask_bits[0] \
3109 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3110 vdup.u16 draw_mask, draw_mask_bits_scalar \
3113 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3115 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3116 vorr.u16 pixels, pixels, msb_mask \
3119 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3122 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3123 shade_blocks_textured_modulated_prologue_##shading(dithering, target); \
3124 stmdb sp!, { r4 - r5, lr }; \
3125 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3127 vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \
3129 shade_blocks_textured_modulated_prologue_##target(); \
3131 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3134 add block_ptr_load_b, block_ptr_load_a, #16; \
3135 vmov.u8 d64_1, #1; \
3136 vmov.u8 d64_4, #4; \
3137 vmov.u8 d64_128, #128; \
3139 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3140 vmov.u8 d128_0x07, #0x07; \
3142 shade_blocks_textured_modulated_load_rg_##shading(); \
3143 vmov.u8 d128_0x1F, #0x1F; \
3145 shade_blocks_textured_modulated_load_bdm_##shading(); \
3146 vmov.u16 d128_0x8000, #0x8000; \
3148 vmovn.u16 texels_r, texels; \
3149 vshrn.u16 texels_g, texels, #5; \
3151 vshrn.u16 texels_b, texels, #7; \
3152 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3154 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3155 vtst.u16 draw_mask, draw_mask, test_mask; \
3157 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3158 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3160 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3161 vshr.u8 texels_b, texels_b, #3; \
3163 shade_blocks_textured_modulate_##dithering(r); \
3164 shade_blocks_textured_modulate_##dithering(g); \
3165 shade_blocks_textured_modulate_##dithering(b); \
3167 vand.u16 pixels, texels, d128_0x8000; \
3168 vceq.u16 zero_mask, texels, #0; \
3170 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3171 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3172 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3174 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3175 vorr.u16 draw_mask, draw_mask, zero_mask; \
3176 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3177 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3179 subs num_blocks, num_blocks, #1; \
3185 vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \
3186 shade_blocks_textured_modulated_load_rg_##shading(); \
3187 vshrn.u16 texels_g, texels, #5; \
3189 shade_blocks_textured_modulated_load_bdm_##shading(); \
3190 vshrn.u16 texels_b, texels, #7; \
3192 pld [ block_ptr_load_a ]; \
3193 vmovn.u16 texels_r, texels; \
3194 vmlal.u8 pixels, pixels_r_low, d64_1; \
3196 vmlal.u8 pixels, pixels_g_low, d64_4; \
3197 vmlal.u8 pixels, pixels_b_low, d64_128; \
3198 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3200 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3201 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3203 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3204 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3206 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3207 vtst.u16 draw_mask, draw_mask, test_mask; \
3209 shade_blocks_textured_modulated_store_pixels_##target(); \
3210 vshr.u8 texels_b, texels_b, #3; \
3212 shade_blocks_textured_modulate_##dithering(r); \
3213 shade_blocks_textured_modulate_##dithering(g); \
3214 shade_blocks_textured_modulate_##dithering(b); \
3216 vand.u16 pixels, texels, d128_0x8000; \
3217 vceq.u16 zero_mask, texels, #0; \
3219 subs num_blocks, num_blocks, #1; \
3221 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3222 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3223 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3225 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3226 vorr.u16 draw_mask, draw_mask, zero_mask; \
3227 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3228 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3233 vmlal.u8 pixels, pixels_r_low, d64_1; \
3234 vmlal.u8 pixels, pixels_g_low, d64_4; \
3235 vmlal.u8 pixels, pixels_b_low, d64_128; \
3237 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3238 shade_blocks_textured_modulated_store_pixels_##target(); \
3240 ldmia sp!, { r4 - r5, pc } \
3243 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3244 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3245 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3246 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3248 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3249 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3250 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3251 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3269 #undef msb_mask_high
3272 #define num_blocks r1
3273 #define mask_msb_ptr r2
3274 #define color_ptr r3
3276 #define block_ptr_load r0
3277 #define draw_mask_store_ptr r3
3278 #define draw_mask_bits_ptr r12
3279 #define draw_mask_ptr r12
3280 #define pixel_store_ptr r14
3282 #define fb_ptr_cmp r4
3285 #define fb_ptr_next r14
3289 #define test_mask q0
3291 #define draw_mask q2
3292 #define zero_mask q3
3293 #define draw_mask_combined q4
3294 #define fb_pixels q5
3295 #define fb_pixels_next q6
3298 #define draw_mask_low d4
3299 #define draw_mask_high d5
3300 #define msb_mask_low d14
3301 #define msb_mask_high d15
3304 function(shade_blocks_textured_unmodulated_indirect)
3305 str r14, [ sp, #-4 ]
3306 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3308 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3309 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3311 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3312 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3315 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3317 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3318 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3319 [ draw_mask_bits_ptr, :16 ], c_64
3320 vceq.u16 zero_mask, pixels, #0
3322 vtst.u16 draw_mask, draw_mask, test_mask
3323 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3325 subs num_blocks, num_blocks, #1
3329 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3330 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3332 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3333 [ draw_mask_bits_ptr, :16 ], c_64
3334 vceq.u16 zero_mask, pixels, #0
3336 vtst.u16 draw_mask, draw_mask, test_mask
3337 vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64
3339 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3340 subs num_blocks, num_blocks, #1
3345 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3346 vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64
3353 function(shade_blocks_textured_unmodulated_direct)
3354 stmdb sp!, { r4, r14 }
3355 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3357 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3358 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3360 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3363 vld1.u32 { test_mask }, [ psx_gpu, :128 ]
3364 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3366 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3367 [ draw_mask_bits_ptr, :16 ], c_64
3368 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3370 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3371 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3372 vceq.u16 zero_mask, pixels, #0
3373 vtst.u16 draw_mask, draw_mask, test_mask
3375 subs num_blocks, num_blocks, #1
3379 mov fb_ptr, fb_ptr_next
3380 ldr fb_ptr_next, [ block_ptr_load, #44 ]
3382 vorr.u16 pixels, pixels, msb_mask
3384 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3385 vmov fb_pixels, fb_pixels_next
3387 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3388 [ draw_mask_bits_ptr, :16 ], c_64
3389 vbif.u16 fb_pixels, pixels, draw_mask_combined
3391 vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64
3393 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3394 add fb_ptr_cmp, fb_ptr_cmp, #14
3398 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3399 vceq.u16 zero_mask, pixels, #0
3401 vst1.u16 { fb_pixels }, [ fb_ptr ]
3402 vtst.u16 draw_mask, draw_mask, test_mask
3405 subs num_blocks, num_blocks, #1
3409 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3410 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3412 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3414 ldmia sp!, { r4, pc }
3417 vst1.u16 { fb_pixels }, [ fb_ptr ]
3418 vceq.u16 zero_mask, pixels, #0
3420 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3421 vtst.u16 draw_mask, draw_mask, test_mask
3426 function(shade_blocks_unshaded_untextured_indirect)
3431 function(shade_blocks_unshaded_untextured_direct)
3432 stmdb sp!, { r4, r14 }
3433 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3435 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
3436 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3438 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
3439 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3441 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3442 vld1.u16 { pixels }, [ color_ptr, :128 ]
3445 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3447 vorr.u16 pixels, pixels, msb_mask
3448 subs num_blocks, num_blocks, #1
3450 ldr fb_ptr_next, [ block_ptr_load ], #64
3452 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3456 vmov fb_pixels, fb_pixels_next
3457 mov fb_ptr, fb_ptr_next
3458 ldr fb_ptr_next, [ block_ptr_load ], #64
3460 vbif.u16 fb_pixels, pixels, draw_mask
3461 vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
3463 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3464 add fb_ptr_cmp, fb_ptr_cmp, #14
3468 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3469 vst1.u16 { fb_pixels }, [ fb_ptr ]
3472 subs num_blocks, num_blocks, #1
3476 vbif.u16 fb_pixels_next, pixels, draw_mask
3477 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3479 ldmia sp!, { r4, pc }
3482 vst1.u16 { fb_pixels }, [ fb_ptr ]
3483 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]
3487 #undef draw_mask_ptr
3494 #define num_blocks r1
3495 #define msb_mask_ptr r2
3496 #define pixel_ptr r3
3497 #define draw_mask_ptr r0
3500 #define fb_ptr_next r14
3501 #define fb_ptr_cmp r4
3509 #undef msb_mask_high
3510 #undef draw_mask_next
3513 #undef fb_pixels_next
3516 #define draw_mask q1
3518 #define fb_pixels q3
3519 #define blend_pixels q4
3520 #define pixels_no_msb q5
3521 #define blend_mask q6
3522 #define fb_pixels_no_msb q7
3523 #define d128_0x8000 q8
3524 #define d128_0x0421 q9
3525 #define fb_pixels_next q10
3526 #define blend_pixels_next q11
3527 #define pixels_next q12
3528 #define draw_mask_next q13
3529 #define write_mask q14
3531 #define pixels_rb q5
3532 #define pixels_mg q7
3534 #define d128_0x7C1F q8
3535 #define d128_0x03E0 q9
3536 #define fb_pixels_rb q10
3537 #define fb_pixels_g q11
3538 #define fb_pixels_masked q11
3539 #define d128_0x83E0 q15
3540 #define pixels_fourth q7
3541 #define d128_0x1C07 q12
3542 #define d128_0x00E0 q13
3543 #define d128_0x80E0 q13
3545 #define msb_mask_low d0
3546 #define msb_mask_high d1
3548 #define blend_blocks_average_set_blend_mask_textured(source) \
3549 vclt.s16 blend_mask, source, #0 \
3551 #define blend_blocks_average_set_stp_bit_textured() \
3552 vorr.u16 blend_pixels, #0x8000 \
3554 #define blend_blocks_average_combine_textured(source) \
3555 vbif.u16 blend_pixels, source, blend_mask \
3557 #define blend_blocks_average_set_blend_mask_untextured(source) \
3559 #define blend_blocks_average_set_stp_bit_untextured() \
3561 #define blend_blocks_average_combine_untextured(source) \
3563 #define blend_blocks_average_mask_set_on() \
3564 vclt.s16 write_mask, fb_pixels_next, #0 \
3566 #define blend_blocks_average_mask_copy_on() \
3567 vorr.u16 draw_mask, draw_mask_next, write_mask \
3569 #define blend_blocks_average_mask_copy_b_on() \
3570 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3572 #define blend_blocks_average_mask_set_off() \
3574 #define blend_blocks_average_mask_copy_off() \
3575 vmov draw_mask, draw_mask_next \
3577 #define blend_blocks_average_mask_copy_b_off() \
3579 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3582 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3583 stmdb sp!, { r4, r14 }; \
3584 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3585 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3587 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3588 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3590 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3593 vmov.u16 d128_0x8000, #0x8000; \
3594 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3595 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3597 vmov.u16 d128_0x0421, #0x0400; \
3598 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3600 vorr.u16 d128_0x0421, #0x0021; \
3601 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3603 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3604 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3605 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3606 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3607 blend_blocks_average_mask_set_##mask_evaluate(); \
3608 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3610 subs num_blocks, num_blocks, #1; \
3614 mov fb_ptr, fb_ptr_next; \
3615 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3617 vmov pixels, pixels_next; \
3618 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3620 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3622 blend_blocks_average_mask_copy_##mask_evaluate(); \
3623 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3625 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3626 blend_blocks_average_set_stp_bit_##texturing(); \
3627 vmov fb_pixels, fb_pixels_next; \
3628 blend_blocks_average_combine_##texturing(pixels); \
3630 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3631 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3632 cmp fb_ptr_cmp, #28; \
3635 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3636 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3638 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3639 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3641 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3642 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3644 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3645 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3646 blend_blocks_average_mask_set_##mask_evaluate(); \
3647 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3650 subs num_blocks, num_blocks, #1; \
3654 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3655 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3657 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3658 blend_blocks_average_set_stp_bit_##texturing(); \
3659 blend_blocks_average_combine_##texturing(pixels_next); \
3661 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3662 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3663 vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3665 ldmia sp!, { r4, pc }; \
3668 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3669 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3670 vst1.u16 { fb_pixels }, [ fb_ptr ]; \
3672 vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \
3673 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3674 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3675 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3676 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3677 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3681 blend_blocks_average_builder(textured, off)
3682 blend_blocks_average_builder(untextured, off)
3683 blend_blocks_average_builder(textured, on)
3684 blend_blocks_average_builder(untextured, on)
3687 #define blend_blocks_add_mask_set_on() \
3688 vclt.s16 write_mask, fb_pixels, #0 \
3690 #define blend_blocks_add_mask_copy_on() \
3691 vorr.u16 draw_mask, draw_mask, write_mask \
3693 #define blend_blocks_add_mask_set_off() \
3695 #define blend_blocks_add_mask_copy_off() \
3698 #define blend_blocks_add_textured_builder(mask_evaluate) \
3701 function(blend_blocks_textured_add_##mask_evaluate) \
3702 stmdb sp!, { r4, r14 }; \
3703 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3704 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3706 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3707 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3709 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3712 vmov.u16 d128_0x7C1F, #0x7C00; \
3713 vmov.u16 d128_0x03E0, #0x0300; \
3714 vmov.u16 d128_0x83E0, #0x8000; \
3715 vorr.u16 d128_0x03E0, #0x00E0; \
3716 vorr.u16 d128_0x7C1F, #0x001F; \
3717 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3719 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3720 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3721 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3722 vclt.s16 blend_mask, pixels, #0; \
3723 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3724 blend_blocks_add_mask_set_##mask_evaluate(); \
3725 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3727 blend_blocks_add_mask_copy_##mask_evaluate(); \
3728 vorr.u16 pixels, pixels, msb_mask; \
3729 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3730 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3731 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3732 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3733 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3734 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3735 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3736 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3738 subs num_blocks, num_blocks, #1; \
3742 mov fb_ptr, fb_ptr_next; \
3744 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3746 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3747 vclt.s16 blend_mask, pixels, #0; \
3749 vorr.u16 pixels, pixels, msb_mask; \
3750 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3751 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3753 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3754 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3756 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3757 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3758 cmp fb_ptr_cmp, #28; \
3761 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3762 blend_blocks_add_mask_set_##mask_evaluate(); \
3763 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3764 blend_blocks_add_mask_copy_##mask_evaluate(); \
3765 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3766 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3767 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3770 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3771 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3772 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3773 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3774 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3776 subs num_blocks, num_blocks, #1; \
3780 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3781 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3782 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3784 ldmia sp!, { r4, pc }; \
3787 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3788 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3790 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3791 blend_blocks_add_mask_set_##mask_evaluate(); \
3792 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3793 blend_blocks_add_mask_copy_##mask_evaluate(); \
3794 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3798 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3801 function(blend_blocks_untextured_add_##mask_evaluate) \
3802 stmdb sp!, { r4, r14 }; \
3803 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3804 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3806 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3807 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3809 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3812 vmov.u16 d128_0x7C1F, #0x7C00; \
3813 vmov.u16 d128_0x03E0, #0x0300; \
3814 vorr.u16 d128_0x7C1F, #0x001F; \
3815 vorr.u16 d128_0x03E0, #0x00E0; \
3817 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3818 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3819 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3820 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3821 blend_blocks_add_mask_set_##mask_evaluate(); \
3822 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3824 blend_blocks_add_mask_copy_##mask_evaluate(); \
3825 vand.u16 pixels_g, pixels, d128_0x03E0; \
3826 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3827 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3828 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3829 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3830 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3831 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3833 subs num_blocks, num_blocks, #1; \
3837 mov fb_ptr, fb_ptr_next; \
3839 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3841 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
3843 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3844 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3845 vand.u16 pixels_g, pixels, d128_0x03E0; \
3847 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3848 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
3850 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3851 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3852 cmp fb_ptr_cmp, #28; \
3855 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3856 blend_blocks_add_mask_set_##mask_evaluate(); \
3857 blend_blocks_add_mask_copy_##mask_evaluate(); \
3858 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3859 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3860 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3863 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3864 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3865 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3866 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3867 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3869 subs num_blocks, num_blocks, #1; \
3873 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3874 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3875 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3876 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
3878 ldmia sp!, { r4, pc }; \
3881 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3882 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3884 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3885 blend_blocks_add_mask_set_##mask_evaluate(); \
3886 blend_blocks_add_mask_copy_##mask_evaluate(); \
3887 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3891 blend_blocks_add_textured_builder(off)
3892 blend_blocks_add_textured_builder(on)
3893 blend_blocks_add_untextured_builder(off)
3894 blend_blocks_add_untextured_builder(on)
3896 #define blend_blocks_subtract_set_blend_mask_textured() \
3897 vclt.s16 blend_mask, pixels_next, #0 \
3899 #define blend_blocks_subtract_combine_textured() \
3900 vbif.u16 blend_pixels, pixels, blend_mask \
3902 #define blend_blocks_subtract_set_stb_textured() \
3903 vorr.u16 blend_pixels, #0x8000 \
3905 #define blend_blocks_subtract_msb_mask_textured() \
3906 vorr.u16 pixels, pixels_next, msb_mask \
3908 #define blend_blocks_subtract_set_blend_mask_untextured() \
3910 #define blend_blocks_subtract_combine_untextured() \
3912 #define blend_blocks_subtract_set_stb_untextured() \
3913 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3915 #define blend_blocks_subtract_msb_mask_untextured() \
3918 #define blend_blocks_subtract_mask_set_on() \
3919 vclt.s16 write_mask, fb_pixels, #0 \
3921 #define blend_blocks_subtract_mask_copy_on() \
3922 vorr.u16 draw_mask, draw_mask_next, write_mask \
3924 #define blend_blocks_subtract_mask_set_off() \
3926 #define blend_blocks_subtract_mask_copy_off() \
3927 vmov draw_mask, draw_mask_next \
3930 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
3933 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
3934 stmdb sp!, { r4, r14 }; \
3935 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3936 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
3938 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3939 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
3941 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3944 vmov.u16 d128_0x7C1F, #0x7C00; \
3945 vmov.u16 d128_0x03E0, #0x0300; \
3946 vorr.u16 d128_0x7C1F, #0x001F; \
3947 vorr.u16 d128_0x03E0, #0x00E0; \
3949 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3950 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3951 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3952 blend_blocks_subtract_set_blend_mask_##texturing(); \
3953 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3954 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3955 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3957 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3958 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3959 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3960 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3961 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3963 subs num_blocks, num_blocks, #1; \
3967 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
3968 mov fb_ptr, fb_ptr_next; \
3969 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
3971 vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \
3972 blend_blocks_subtract_msb_mask_##texturing(); \
3974 vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \
3975 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3976 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3977 blend_blocks_subtract_set_stb_##texturing(); \
3978 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3979 blend_blocks_subtract_combine_##texturing(); \
3980 blend_blocks_subtract_set_blend_mask_##texturing(); \
3981 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3983 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3984 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3985 cmp fb_ptr_cmp, #28; \
3988 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
3989 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3990 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3991 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3992 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3993 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
3994 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3997 subs num_blocks, num_blocks, #1; \
4001 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4003 blend_blocks_subtract_msb_mask_##texturing(); \
4004 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4005 blend_blocks_subtract_set_stb_##texturing(); \
4006 blend_blocks_subtract_combine_##texturing(); \
4007 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4008 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4010 ldmia sp!, { r4, pc }; \
4013 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4014 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4015 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4016 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4017 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4018 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4019 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4023 blend_blocks_subtract_builder(textured, off)
4024 blend_blocks_subtract_builder(textured, on)
4025 blend_blocks_subtract_builder(untextured, off)
4026 blend_blocks_subtract_builder(untextured, on)
4029 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4032 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4033 stmdb sp!, { r4, r14 }; \
4034 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4035 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4037 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4038 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4040 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4043 vmov.u16 d128_0x7C1F, #0x7C00; \
4044 vmov.u16 d128_0x03E0, #0x0300; \
4045 vmov.u16 d128_0x1C07, #0x1C00; \
4046 vmov.u16 d128_0x00E0, #0x00E0; \
4047 vorr.u16 d128_0x7C1F, #0x001F; \
4048 vorr.u16 d128_0x03E0, #0x00E0; \
4049 vorr.u16 d128_0x1C07, #0x0007; \
4051 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4052 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4053 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4054 vclt.s16 blend_mask, pixels, #0; \
4055 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4056 blend_blocks_add_mask_set_##mask_evaluate(); \
4057 vshr.s16 pixels_fourth, pixels, #2; \
4058 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4060 blend_blocks_add_mask_copy_##mask_evaluate(); \
4061 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4062 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4063 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4064 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4065 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4066 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4067 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4069 subs num_blocks, num_blocks, #1; \
4073 mov fb_ptr, fb_ptr_next; \
4074 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4076 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4077 vbif.u16 blend_pixels, pixels, blend_mask; \
4079 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4080 vclt.s16 blend_mask, pixels, #0; \
4081 vshr.s16 pixels_fourth, pixels, #2; \
4082 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4083 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4085 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4086 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4088 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4089 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4090 cmp fb_ptr_cmp, #28; \
4093 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4094 blend_blocks_add_mask_set_##mask_evaluate(); \
4095 blend_blocks_add_mask_copy_##mask_evaluate(); \
4096 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4097 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4098 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4101 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4102 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4103 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4104 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4105 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4107 subs num_blocks, num_blocks, #1; \
4111 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4112 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4113 vbif.u16 blend_pixels, pixels, blend_mask; \
4114 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4115 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4117 ldmia sp!, { r4, pc }; \
4120 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4121 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4123 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4124 blend_blocks_add_mask_set_##mask_evaluate(); \
4125 blend_blocks_add_mask_copy_##mask_evaluate(); \
4126 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4131 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4134 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4135 stmdb sp!, { r4, r14 }; \
4136 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4137 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
4139 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4140 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \
4142 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4145 vmov.u16 d128_0x7C1F, #0x7C00; \
4146 vmov.u16 d128_0x03E0, #0x0300; \
4147 vmov.u16 d128_0x1C07, #0x1C00; \
4148 vmov.u16 d128_0x00E0, #0x00E0; \
4149 vorr.u16 d128_0x7C1F, #0x001F; \
4150 vorr.u16 d128_0x03E0, #0x00E0; \
4151 vorr.u16 d128_0x1C07, #0x0007; \
4153 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4154 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4155 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4156 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4157 blend_blocks_add_mask_set_##mask_evaluate(); \
4158 vshr.s16 pixels_fourth, pixels, #2; \
4159 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4161 blend_blocks_add_mask_copy_##mask_evaluate(); \
4162 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4163 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4164 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4165 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4166 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4167 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4168 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4170 subs num_blocks, num_blocks, #1; \
4174 mov fb_ptr, fb_ptr_next; \
4175 ldr fb_ptr_next, [ pixel_ptr, #28 ]; \
4177 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \
4179 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4180 vshr.s16 pixels_fourth, pixels, #2; \
4181 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4182 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4184 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4185 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \
4187 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4188 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4189 cmp fb_ptr_cmp, #28; \
4192 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4193 blend_blocks_add_mask_set_##mask_evaluate(); \
4194 blend_blocks_add_mask_copy_##mask_evaluate(); \
4195 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4196 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4197 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4200 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4201 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4202 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4203 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4204 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4206 subs num_blocks, num_blocks, #1; \
4210 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4211 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4212 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4213 vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \
4215 ldmia sp!, { r4, pc }; \
4218 vst1.u16 { blend_pixels }, [ fb_ptr ]; \
4219 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4221 vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \
4222 blend_blocks_add_mask_set_##mask_evaluate(); \
4223 blend_blocks_add_mask_copy_##mask_evaluate(); \
4224 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4228 blend_blocks_add_fourth_textured_builder(off)
4229 blend_blocks_add_fourth_textured_builder(on)
4230 blend_blocks_add_fourth_untextured_builder(off)
4231 blend_blocks_add_fourth_untextured_builder(on)
4233 // TODO: Optimize this more. Need a scene that actually uses it for
4238 function(blend_blocks_textured_unblended_on)
4239 stmdb sp!, { r4, r14 }
4240 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4241 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
4243 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4244 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]
4246 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4249 ldr fb_ptr, [ pixel_ptr, #28 ]
4250 vld1.u16 { fb_pixels }, [ fb_ptr ]
4251 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4252 vclt.s16 write_mask, fb_pixels, #0
4253 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4255 subs num_blocks, num_blocks, #1
4259 vorr.u16 draw_mask, draw_mask, write_mask
4260 vbif.u16 fb_pixels, pixels, draw_mask
4261 vst1.u16 { fb_pixels }, [ fb_ptr ]
4263 ldr fb_ptr, [ pixel_ptr, #28 ]
4264 vld1.u16 { fb_pixels }, [ fb_ptr ]
4265 vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64
4266 vclt.s16 write_mask, fb_pixels, #0
4267 vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64
4269 subs num_blocks, num_blocks, #1
4273 vorr.u16 draw_mask, draw_mask, write_mask
4274 vbif.u16 fb_pixels, pixels, draw_mask
4275 vst1.u16 { fb_pixels }, [ fb_ptr ]
4277 ldmia sp!, { r4, pc }
4280 function(blend_blocks_textured_unblended_off)
4290 vld1.u32 { u_whole_8, v_whole_8 }, [ r1, :128 ], r3
4310 #define num_width r12
4320 function(render_block_fill_body)
4321 vdup.u16 colors_a, color
4324 vmov colors_b, colors_a
4325 sub pitch, pitch, width, lsl #1
4327 mov num_width, width
4330 vst1.u32 { colors_a, colors_b }, [ vram_ptr, :256 ]!
4332 subs num_width, num_width, #16
4335 add vram_ptr, vram_ptr, pitch
4336 mov num_width, width
4338 subs height, height, #1
4352 #undef dirty_textures_mask
4354 #undef current_texture_mask
4365 #define offset_u_right r10
4366 #define width_rounded r11
4367 #define height_rounded r12
4369 #define texture_offset_base r1
4370 #define tile_width r2
4371 #define tile_height r3
4372 #define num_blocks r4
4374 #define sub_tile_height r6
4376 #define texture_mask r8
4377 #define column_data r9
4378 #define texture_offset r10
4379 #define tiles_remaining r11
4380 #define fb_ptr_advance_column r12
4381 #define texture_block_ptr r14
4383 #define texture_page_ptr r3
4384 #define left_block_mask r4
4385 #define right_block_mask r5
4386 #define texture_mask_rev r10
4387 #define control_mask r11
4389 #define dirty_textures_mask r4
4391 #define current_texture_mask r6
4405 #define draw_masks_fb_ptrs q1
4407 #define draw_mask_fb_ptr_left d2
4408 #define draw_mask_fb_ptr_right d3
4410 #define draw_mask_fb_ptr_left_a d2
4411 #define draw_mask_fb_ptr_left_b d3
4412 #define draw_mask_fb_ptr_right_a d10
4413 #define draw_mask_fb_ptr_right_b d11
4414 #define draw_masks_fb_ptrs2 q5
4416 #define clut_low_a d4
4417 #define clut_low_b d5
4418 #define clut_high_a d6
4419 #define clut_high_b d7
4421 #define block_masks d8
4422 #define block_masks_shifted d9
4427 #define texels_low d12
4428 #define texels_high d13
4430 #define texels_wide_low d14
4431 #define texels_wide_high d15
4432 #define texels_wide q7
4435 setup_sprite_flush_blocks:
4438 stmdb sp!, { r0 - r3, r12, r14 }
4439 bl flush_render_block_buffer
4440 ldmia sp!, { r0 - r3, r12, r14 }
4444 add block, psx_gpu, #psx_gpu_blocks_offset
4448 setup_sprite_update_texture_4bpp_cache:
4449 stmdb sp!, { r0 - r3, r14 }
4450 bl update_texture_4bpp_cache
4451 ldmia sp!, { r0 - r3, pc }
4454 setup_sprite_update_texture_8bpp_cache:
4455 stmdb sp!, { r0 - r3, r14 }
4456 bl update_texture_8bpp_cache
4457 ldmia sp!, { r0 - r3, pc }
4460 #define setup_sprite_tiled_initialize_4bpp() \
4461 ldr dirty_textures_mask, \
4462 [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]; \
4463 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4465 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4466 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4468 tst current_texture_mask, dirty_textures_mask; \
4469 vuzp.u8 clut_a, clut_b; \
4471 blne setup_sprite_update_texture_4bpp_cache \
4473 #define setup_sprite_tiled_initialize_8bpp() \
4474 ldr dirty_textures_mask, \
4475 [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]; \
4476 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \
4478 tst current_texture_mask, dirty_textures_mask; \
4479 blne setup_sprite_update_texture_8bpp_cache \
4482 #define setup_sprite_block_count_single() \
4485 #define setup_sprite_block_count_double() \
4486 sub_tile_height, lsl #1 \
4488 #define setup_sprite_tile_add_blocks(type) \
4489 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4490 cmp num_blocks, #MAX_BLOCKS; \
4492 movgt num_blocks, setup_sprite_block_count_##type(); \
4493 blgt setup_sprite_flush_blocks \
4496 #define setup_sprite_tile_full_4bpp(edge) \
4497 setup_sprite_tile_add_blocks(double); \
4500 and texture_block_ptr, texture_offset, texture_mask; \
4501 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4504 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4505 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4507 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4508 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4510 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4511 add texture_block_ptr, texture_offset, #8; \
4513 and texture_block_ptr, texture_block_ptr, texture_mask; \
4514 add block, block, #40; \
4516 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4517 add fb_ptr, fb_ptr, #16; \
4519 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4520 add block, block, #24; \
4522 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4523 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4526 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4527 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4529 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4530 add block, block, #40; \
4532 add texture_offset, texture_offset, #0x10; \
4533 add fb_ptr, fb_ptr, #(2048 - 16); \
4535 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4536 add block, block, #24; \
4538 subs sub_tile_height, sub_tile_height, #1; \
4541 add texture_offset, texture_offset, #0xF00; \
4542 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4545 #define setup_sprite_tile_half_4bpp(edge) \
4546 setup_sprite_tile_add_blocks(single); \
4549 and texture_block_ptr, texture_offset, texture_mask; \
4550 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4553 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4554 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4556 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4557 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4559 vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \
4560 add block, block, #40; \
4562 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4563 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4565 add block, block, #24; \
4566 add texture_offset, texture_offset, #0x10; \
4568 add fb_ptr, fb_ptr, #2048; \
4569 subs sub_tile_height, sub_tile_height, #1; \
4573 add texture_offset, texture_offset, #0xF00; \
4574 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4577 #define setup_sprite_tile_full_8bpp(edge) \
4578 setup_sprite_tile_add_blocks(double); \
4579 add block, block, #16; \
4582 and texture_block_ptr, texture_offset, texture_mask; \
4583 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4586 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4587 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4589 add texture_block_ptr, texture_offset, #8; \
4590 vst1.u32 { texels }, [ block, :64 ]; \
4592 and texture_block_ptr, texture_block_ptr, texture_mask; \
4593 add block, block, #24; \
4595 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4597 add fb_ptr, fb_ptr, #16; \
4598 vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \
4600 add block, block, #40; \
4601 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4604 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4605 vst1.u32 { texels }, [ block, :64 ]; \
4606 add block, block, #24; \
4608 add texture_offset, texture_offset, #0x10; \
4609 add fb_ptr, fb_ptr, #(2048 - 16); \
4611 vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \
4612 add block, block, #40; \
4614 subs sub_tile_height, sub_tile_height, #1; \
4617 sub block, block, #16; \
4618 add texture_offset, texture_offset, #0xF00; \
4619 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4622 #define setup_sprite_tile_half_8bpp(edge) \
4623 setup_sprite_tile_add_blocks(single); \
4624 add block, block, #16; \
4627 and texture_block_ptr, texture_offset, texture_mask; \
4628 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4631 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4632 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4634 vst1.u32 { texels }, [ block, :64 ]; \
4635 add block, block, #24; \
4637 vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \
4638 add block, block, #40; \
4640 add texture_offset, texture_offset, #0x10; \
4641 add fb_ptr, fb_ptr, #2048; \
4643 subs sub_tile_height, sub_tile_height, #1; \
4646 sub block, block, #16; \
4647 add texture_offset, texture_offset, #0xF00; \
4648 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4651 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4652 add texture_offset, texture_offset_base, #8; \
4653 add fb_ptr, fb_ptr, #16 \
4655 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4656 mov texture_offset, texture_offset_base \
4658 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4659 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4661 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4662 mov texture_offset, texture_offset_base \
4664 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4665 sub fb_ptr, fb_ptr, #16 \
4667 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4669 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4670 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4672 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4675 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \
4677 mov sub_tile_height, column_data; \
4678 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4679 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4680 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4682 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \
4684 and sub_tile_height, column_data, #0xFF; \
4685 mov tiles_remaining, column_data, lsr #16; \
4686 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4687 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4689 subs tiles_remaining, tiles_remaining, #1; \
4693 mov sub_tile_height, #16; \
4694 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4695 subs tiles_remaining, tiles_remaining, #1; \
4699 uxtb sub_tile_height, column_data, ror #8; \
4700 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4701 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4704 #define setup_sprite_column_data_single() \
4705 mov column_data, height; \
4706 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] \
4708 #define setup_sprite_column_data_multi() \
4709 and height_rounded, height_rounded, #0xF; \
4710 rsb column_data, offset_v, #16; \
4712 add height_rounded, height_rounded, #1; \
4713 sub tile_height, tile_height, #1; \
4715 orr column_data, column_data, tile_height, lsl #16; \
4716 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]; \
4718 orr column_data, column_data, height_rounded, lsl #8 \
4720 #define setup_sprite_setup_left_draw_mask_fb_ptr() \
4721 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4722 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4724 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column() \
4725 mov fb_ptr_advance_column, #32; \
4726 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4728 sub fb_ptr_advance_column, height, lsl #11; \
4729 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4731 #define setup_sprite_setup_right_draw_mask_fb_ptr() \
4732 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4733 vdup.u8 draw_mask_fb_ptr_right, block_masks[5] \
4735 #define setup_sprite_tile_column_width_single(tm, multi_height, edge_mode, \
4737 setup_sprite_##tm##_single_##multi_height##_##edge_mode##_##edge##x4mode: \
4738 setup_sprite_column_data_##multi_height(); \
4739 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4740 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4741 setup_sprite_setup_left_draw_mask_fb_ptr##x4mode(); \
4743 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, tm, x4mode); \
4744 ldmia sp!, { r4 - r11, pc } \
4746 #define setup_sprite_tiled_advance_column() \
4747 add texture_offset_base, texture_offset_base, #0x100; \
4748 tst texture_offset_base, #0xF00; \
4749 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4751 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4752 right_mode, x4mode) \
4753 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode##x4mode:\
4754 setup_sprite_column_data_##multi_height(); \
4756 setup_sprite_setup_left_draw_mask_fb_ptr_advance_column##x4mode(); \
4758 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm, x4mode);\
4760 subs tile_width, tile_width, #2; \
4761 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4765 vmov.u8 draw_masks_fb_ptrs, #0; \
4766 vmov.u8 draw_masks_fb_ptrs2, #0; \
4769 setup_sprite_tiled_advance_column(); \
4770 setup_sprite_tile_column_height_##multi_height(full, none, tm, x4mode); \
4771 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4772 subs tile_width, tile_width, #1; \
4776 setup_sprite_setup_right_draw_mask_fb_ptr##x4mode(); \
4778 setup_sprite_tiled_advance_column(); \
4779 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm, x4mode);\
4780 ldmia sp!, { r4 - r11, pc } \
4783 #define setup_sprite_offset_u_adjust() \
4785 #define setup_sprite_get_left_block_mask() \
4786 and left_block_mask, left_block_mask, #0xFF \
4788 #define setup_sprite_compare_left_block_mask() \
4789 cmp left_block_mask, #0xFF \
4791 #define setup_sprite_get_right_block_mask() \
4792 uxtb right_block_mask, right_block_mask, ror #8 \
4794 #define setup_sprite_compare_right_block_mask() \
4795 cmp right_block_mask, #0xFF \
4800 #define fb_ptr2 column_data
4802 #define setup_sprite_offset_u_adjust_4x() \
4803 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4804 lsl offset_u_right, #1; \
4806 add offset_u_right, #1 \
4808 #define setup_sprite_get_left_block_mask_4x() \
4809 sxth left_block_mask, left_block_mask \
4811 #define setup_sprite_compare_left_block_mask_4x() \
4812 cmp left_block_mask, #0xFFFFFFFF \
4814 #define setup_sprite_get_right_block_mask_4x() \
4815 sxth right_block_mask, right_block_mask, ror #16 \
4817 #define setup_sprite_compare_right_block_mask_4x() \
4818 cmp right_block_mask, #0xFFFFFFFF \
4821 #define widen_texels_16bpp(texels_) \
4822 vmov texels_wide_low, texels_; \
4823 vmov texels_wide_high, texels_; \
4824 vzip.16 texels_wide_low, texels_wide_high \
4826 #define widen_texels_8bpp(texels_) \
4827 vmov texels_wide_low, texels_; \
4828 vmov texels_wide_high, texels_; \
4829 vzip.8 texels_wide_low, texels_wide_high \
4831 #define write_block_16bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4832 vst1.u32 { texels_ }, [ block_, :128 ]; \
4833 add block_, block_, #40; \
4835 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4836 vst1.u32 { draw_mask_fb_ptr_ }, [ block_, :64 ]; \
4837 add block_, block_, #24 \
4839 /* assumes 16-byte offset already added to block_ */
4840 #define write_block_8bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4841 vst1.u32 { texels_ }, [ block_, :64 ]; \
4842 add block_, block_, #24; \
4844 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4845 vst1.u32 { draw_mask_fb_ptr_ }, [ block_, :64 ]; \
4846 add block_, block_, #40 \
4848 #define do_texture_block_16bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4849 draw_mask_fb_ptr_b_) \
4850 widen_texels_16bpp(texels_low); \
4851 add fb_ptr_tmp, fb_ptr, #1024*2; \
4853 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr); \
4855 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4856 widen_texels_16bpp(texels_high); \
4858 add fb_ptr_tmp, fb_ptr, #8*2; \
4859 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4861 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4862 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4864 #define do_texture_block_8bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4865 draw_mask_fb_ptr_b_) \
4866 widen_texels_8bpp(texels); \
4867 add fb_ptr_tmp, fb_ptr, #1024*2; \
4869 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr); \
4870 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4872 add fb_ptr_tmp, fb_ptr, #8*2; \
4873 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4875 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4876 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4879 #define setup_sprite_tiled_initialize_4bpp_4x() \
4880 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \
4881 vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \
4883 vuzp.u8 clut_a, clut_b \
4885 #define setup_sprite_tiled_initialize_8bpp_4x() \
4888 #define setup_sprite_block_count_single_4x() \
4889 sub_tile_height, lsl #2 \
4891 #define setup_sprite_block_count_double_4x() \
4892 sub_tile_height, lsl #(1+2) \
4894 #define setup_sprite_tile_full_4bpp_4x(edge) \
4895 setup_sprite_tile_add_blocks(double_4x); \
4896 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4899 and texture_block_ptr, texture_offset, texture_mask; \
4902 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4903 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4905 add texture_block_ptr, texture_offset, #8; \
4906 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4908 and texture_block_ptr, texture_block_ptr, texture_mask; \
4909 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4911 vzip.8 texels_low, texels_high; \
4912 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
4913 draw_mask_fb_ptr_left_b); \
4915 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4916 add fb_ptr, fb_ptr, #16*2; \
4918 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4919 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4922 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4924 vzip.8 texels_low, texels_high; \
4925 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
4926 draw_mask_fb_ptr_right_b); \
4928 add texture_offset, texture_offset, #0x10; \
4929 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
4931 subs sub_tile_height, sub_tile_height, #1; \
4934 ldr column_data, [sp], #8; /* fb_ptr2 */ \
4935 add texture_offset, texture_offset, #0xF00; \
4936 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4939 #define setup_sprite_tile_half_4bpp_4x(edge) \
4940 setup_sprite_tile_add_blocks(single_4x); \
4941 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4944 and texture_block_ptr, texture_offset, texture_mask; \
4947 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4948 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4950 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4951 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4953 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4954 add texture_offset, texture_offset, #0x10; \
4956 vzip.8 texels_low, texels_high; \
4957 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
4958 draw_mask_fb_ptr_##edge##_b); \
4960 add fb_ptr, fb_ptr, #2048 * 2; \
4961 subs sub_tile_height, sub_tile_height, #1; \
4965 ldr column_data, [sp], #8; /* fb_ptr2 */ \
4966 add texture_offset, texture_offset, #0xF00; \
4967 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
4970 #define setup_sprite_tile_full_8bpp_4x(edge) \
4971 setup_sprite_tile_add_blocks(double_4x); \
4972 add block, block, #16; \
4973 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4976 and texture_block_ptr, texture_offset, texture_mask; \
4979 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4980 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4982 add texture_block_ptr, texture_offset, #8; \
4983 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
4984 draw_mask_fb_ptr_left_b); \
4986 and texture_block_ptr, texture_block_ptr, texture_mask; \
4988 add fb_ptr, fb_ptr, #16*2; \
4989 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4991 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
4994 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
4995 draw_mask_fb_ptr_right_b); \
4997 add texture_offset, texture_offset, #0x10; \
4998 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
5000 subs sub_tile_height, sub_tile_height, #1; \
5003 sub block, block, #16; \
5004 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5005 add texture_offset, texture_offset, #0xF00; \
5006 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
5009 #define setup_sprite_tile_half_8bpp_4x(edge) \
5010 setup_sprite_tile_add_blocks(single_4x); \
5011 add block, block, #16; \
5012 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5015 and texture_block_ptr, texture_offset, texture_mask; \
5018 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5019 vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \
5021 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5022 draw_mask_fb_ptr_##edge##_b); \
5024 add texture_offset, texture_offset, #0x10; \
5025 add fb_ptr, fb_ptr, #2048 * 2; \
5027 subs sub_tile_height, sub_tile_height, #1; \
5030 sub block, block, #16; \
5031 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5032 add texture_offset, texture_offset, #0xF00; \
5033 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \
5036 #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \
5037 add texture_offset, texture_offset_base, #8; \
5038 add fb_ptr, fb_ptr, #16 * 2 \
5040 #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \
5041 mov texture_offset, texture_offset_base \
5043 #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \
5044 setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \
5046 #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \
5047 mov texture_offset, texture_offset_base \
5049 #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \
5050 sub fb_ptr, fb_ptr, #16 * 2 \
5052 #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \
5054 #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \
5055 setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \
5057 #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \
5060 #define setup_sprite_setup_left_draw_mask_fb_ptr_4x() \
5061 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5062 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5063 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5064 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5066 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column_4x() \
5067 mov fb_ptr_advance_column, #32 * 2; \
5068 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5069 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5070 sub fb_ptr_advance_column, height, lsl #11 + 1; \
5071 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5072 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5074 #define setup_sprite_setup_right_draw_mask_fb_ptr_4x() \
5075 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[4]; \
5076 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[5]; \
5077 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[6]; \
5078 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[7] \
5086 // [ sp + 4 ]: width
5087 // [ sp + 8 ]: height
5088 // [ sp + 12 ]: color (unused)
5090 #define setup_sprite_tiled_builder(texture_mode, x4mode) \
5092 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \
5094 setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \
5096 setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \
5098 setup_sprite_tile_column_width_single(texture_mode, single, full, none, \
5100 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \
5102 setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \
5104 setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \
5106 setup_sprite_tile_column_width_single(texture_mode, single, half, right, \
5108 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \
5110 setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \
5112 setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \
5114 setup_sprite_tile_column_width_single(texture_mode, single, half, left, \
5116 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \
5118 setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \
5123 function(setup_sprite_##texture_mode##x4mode) \
5124 stmdb sp!, { r4 - r11, r14 }; \
5125 setup_sprite_tiled_initialize_##texture_mode##x4mode(); \
5127 ldr v, [ sp, #36 ]; \
5128 and offset_u, u, #0xF; \
5130 ldr width, [ sp, #40 ]; \
5131 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \
5133 ldr height, [ sp, #44 ]; \
5134 add fb_ptr, fb_ptr, y, lsl #11; \
5136 add fb_ptr, fb_ptr, x, lsl #1; \
5137 and offset_v, v, #0xF; \
5139 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
5140 add width_rounded, offset_u, width; \
5142 add height_rounded, offset_v, height; \
5143 add width_rounded, width_rounded, #15; \
5145 add height_rounded, height_rounded, #15; \
5146 mov tile_width, width_rounded, lsr #4; \
5148 /* texture_offset_base = VH-VL-00-00 */\
5149 mov texture_offset_base, v, lsl #8; \
5150 and offset_u_right, width_rounded, #0xF; \
5152 /* texture_offset_base = VH-UH-UL-00 */\
5153 bfi texture_offset_base, u, #4, #8; \
5154 mov right_block_mask, #0xFFFFFFFE; \
5156 setup_sprite_offset_u_adjust##x4mode(); \
5158 /* texture_offset_base = VH-UH-VL-00 */\
5159 bfi texture_offset_base, v, #4, #4; \
5160 mov left_block_mask, #0xFFFFFFFF; \
5162 mov tile_height, height_rounded, lsr #4; \
5163 mvn left_block_mask, left_block_mask, lsl offset_u; \
5165 /* texture_mask = HH-HL-WH-WL */\
5166 ldrh texture_mask, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]; \
5167 mov right_block_mask, right_block_mask, lsl offset_u_right; \
5169 /* texture_mask_rev = WH-WL-HH-HL */\
5170 rev16 texture_mask_rev, texture_mask; \
5171 vmov block_masks, left_block_mask, right_block_mask; \
5173 /* texture_mask = HH-HL-HL-WL */\
5174 bfi texture_mask, texture_mask_rev, #4, #4; \
5175 /* texture_mask_rev = 00-00-00-WH */\
5176 mov texture_mask_rev, texture_mask_rev, lsr #12; \
5178 /* texture_mask = HH-WH-HL-WL */\
5179 bfi texture_mask, texture_mask_rev, #8, #4; \
5180 setup_sprite_get_left_block_mask##x4mode(); \
5182 mov control_mask, #0; \
5183 setup_sprite_compare_left_block_mask##x4mode(); \
5185 setup_sprite_get_right_block_mask##x4mode(); \
5186 orreq control_mask, control_mask, #0x4; \
5188 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \
5189 setup_sprite_compare_right_block_mask##x4mode(); \
5191 orreq control_mask, control_mask, #0x8; \
5192 cmp tile_width, #1; \
5194 add block, psx_gpu, #psx_gpu_blocks_offset; \
5195 orreq control_mask, control_mask, #0x1; \
5197 cmp tile_height, #1; \
5198 add block, block, num_blocks, lsl #6; \
5200 orreq control_mask, control_mask, #0x2; \
5201 ldr pc, [ pc, control_mask, lsl #2 ]; \
5204 .word setup_sprite_##texture_mode##_multi_multi_full_full##x4mode; \
5205 .word setup_sprite_##texture_mode##_single_multi_full_none##x4mode; \
5206 .word setup_sprite_##texture_mode##_multi_single_full_full##x4mode; \
5207 .word setup_sprite_##texture_mode##_single_single_full_none##x4mode; \
5208 .word setup_sprite_##texture_mode##_multi_multi_half_full##x4mode; \
5209 .word setup_sprite_##texture_mode##_single_multi_half_right##x4mode; \
5210 .word setup_sprite_##texture_mode##_multi_single_half_full##x4mode; \
5211 .word setup_sprite_##texture_mode##_single_single_half_right##x4mode; \
5212 .word setup_sprite_##texture_mode##_multi_multi_full_half##x4mode; \
5213 .word setup_sprite_##texture_mode##_single_multi_half_left##x4mode; \
5214 .word setup_sprite_##texture_mode##_multi_single_full_half##x4mode; \
5215 .word setup_sprite_##texture_mode##_single_single_half_left##x4mode; \
5216 .word setup_sprite_##texture_mode##_multi_multi_half_half##x4mode; \
5218 .word setup_sprite_##texture_mode##_multi_single_half_half##x4mode; \
5221 setup_sprite_tiled_builder(4bpp,);
5222 setup_sprite_tiled_builder(8bpp,);
5224 #undef draw_mask_fb_ptr_left
5225 #undef draw_mask_fb_ptr_right
5227 setup_sprite_tiled_builder(4bpp, _4x);
5228 setup_sprite_tiled_builder(8bpp, _4x);
5236 #define block_ptr r0
5237 #define num_blocks r1
5239 #define texel_shift_mask r3
5240 #define block_pixels_a r4
5241 #define block_pixels_b r5
5250 #define texels_01 r6
5251 #define texels_23 r7
5252 #define texels_45 r8
5253 #define texels_67 r9
5255 function(texture_sprite_blocks_8bpp)
5256 stmdb sp!, { r4 - r11, r14 }
5257 movw texel_shift_mask, #(0xFF << 1)
5259 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5260 ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]
5262 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
5263 ldr block_pixels_a, [ block_ptr, #16 ]
5266 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
5267 ldr block_pixels_b, [ block_ptr, #20 ]
5269 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
5270 ldrh texel_0, [ clut_ptr, texel_0 ]
5272 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
5273 ldrh texel_1, [ clut_ptr, texel_1 ]
5275 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
5276 ldr block_pixels_a, [ block_ptr, #(64 + 16) ]
5278 ldrh texel_2, [ clut_ptr, texel_2 ]
5279 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
5281 ldrh texel_3, [ clut_ptr, texel_3 ]
5282 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
5284 ldrh texel_4, [ clut_ptr, texel_4 ]
5285 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
5287 ldrh texel_5, [ clut_ptr, texel_5 ]
5288 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
5290 ldrh texel_6, [ clut_ptr, texel_6 ]
5291 orr texels_01, texel_0, texel_1, lsl #16
5293 ldrh texel_7, [ clut_ptr, texel_7 ]
5294 orr texels_23, texel_2, texel_3, lsl #16
5296 orr texels_45, texel_4, texel_5, lsl #16
5297 str texels_01, [ block_ptr, #0 ]
5299 orr texels_67, texel_6, texel_7, lsl #16
5300 str texels_23, [ block_ptr, #4 ]
5302 subs num_blocks, num_blocks, #1
5303 str texels_45, [ block_ptr, #8 ]
5305 str texels_67, [ block_ptr, #12 ]
5306 add block_ptr, block_ptr, #64
5310 ldmia sp!, { r4 - r11, pc }
5313 #undef width_rounded
5316 #undef texture_offset
5319 #undef texels_wide_low
5320 #undef texels_wide_high
5331 #define left_offset r8
5332 #define width_rounded r9
5333 #define right_width r10
5335 #define block_width r11
5337 #define texture_offset_base r1
5338 #define texture_mask r2
5339 #define texture_page_ptr r3
5340 #define num_blocks r4
5343 #define texture_offset r8
5344 #define blocks_remaining r9
5346 #define fb_ptr_pitch r12
5347 #define texture_block_ptr r14
5349 #define texture_mask_width r2
5350 #define texture_mask_height r3
5351 #define left_mask_bits r4
5352 #define right_mask_bits r5
5356 #undef block_masks_shifted
5359 #define block_masks d0
5360 #define block_masks_shifted d1
5361 #define draw_mask_fb_ptr d2
5364 #define draw_mask_fb_ptr_a d2
5365 #define draw_mask_fb_ptr_b d3
5366 #define texels_low d4
5367 #define texels_high d5
5368 #define texels_wide_low d6
5369 #define texels_wide_high d7
5370 #define texels_wide q3
5373 setup_sprites_16bpp_flush:
5376 stmdb sp!, { r0 - r3, r12, r14 }
5377 bl flush_render_block_buffer
5378 ldmia sp!, { r0 - r3, r12, r14 }
5382 add block, psx_gpu, #psx_gpu_blocks_offset
5383 mov num_blocks, block_width
5387 function(setup_sprite_16bpp)
5388 stmdb sp!, { r4 - r11, r14 }
5389 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5392 add fb_ptr, fb_ptr, y, lsl #11
5394 ldr width, [ sp, #40 ]
5395 add fb_ptr, fb_ptr, x, lsl #1
5397 ldr height, [ sp, #44 ]
5398 and left_offset, u, #0x7
5400 add texture_offset_base, u, u
5401 add width_rounded, width, #7
5403 add texture_offset_base, v, lsl #11
5404 mov left_mask_bits, #0xFF
5406 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5407 add width_rounded, width_rounded, left_offset
5409 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5410 sub fb_ptr, fb_ptr, left_offset, lsl #1
5412 add texture_mask, texture_mask_width, texture_mask_width
5413 mov right_mask_bits, #0xFE
5415 and right_width, width_rounded, #0x7
5416 mvn left_mask_bits, left_mask_bits, lsl left_offset
5418 add texture_mask, texture_mask_height, lsl #11
5419 mov block_width, width_rounded, lsr #3
5421 mov right_mask_bits, right_mask_bits, lsl right_width
5422 movw fb_ptr_pitch, #(2048 + 16)
5424 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5425 vmov block_masks, left_mask_bits, right_mask_bits
5427 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5428 add block, psx_gpu, #psx_gpu_blocks_offset
5430 bic texture_offset_base, texture_offset_base, #0xF
5433 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5434 add block, block, num_blocks, lsl #6
5438 vext.32 block_masks_shifted, block_masks, block_masks, #1
5439 vorr.u32 block_masks, block_masks, block_masks_shifted
5440 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5443 add num_blocks, num_blocks, #1
5444 cmp num_blocks, #MAX_BLOCKS
5445 blgt setup_sprites_16bpp_flush
5447 and texture_block_ptr, texture_offset_base, texture_mask
5448 subs height, height, #1
5450 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5451 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5453 vst1.u32 { texels }, [ block, :128 ]
5454 add block, block, #40
5456 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5459 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5461 add block, block, #24
5462 add texture_offset_base, texture_offset_base, #2048
5463 add fb_ptr, fb_ptr, #2048
5464 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5467 ldmia sp!, { r4 - r11, pc }
5470 add num_blocks, num_blocks, block_width
5471 mov texture_offset, texture_offset_base
5473 cmp num_blocks, #MAX_BLOCKS
5474 blgt setup_sprites_16bpp_flush
5476 add texture_offset_base, texture_offset_base, #2048
5477 and texture_block_ptr, texture_offset, texture_mask
5479 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5480 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5482 vst1.u32 { texels }, [ block, :128 ]
5483 add block, block, #40
5485 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5486 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5489 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5490 subs blocks_remaining, block_width, #2
5492 add texture_offset, texture_offset, #16
5493 add fb_ptr, fb_ptr, #16
5495 vmov.u8 draw_mask_fb_ptr, #0
5497 add block, block, #24
5501 and texture_block_ptr, texture_offset, texture_mask
5502 subs blocks_remaining, blocks_remaining, #1
5504 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5505 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5507 vst1.u32 { texels }, [ block, :128 ]
5508 add block, block, #40
5510 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5513 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5515 add texture_offset, texture_offset, #16
5516 add fb_ptr, fb_ptr, #16
5518 add block, block, #24
5522 and texture_block_ptr, texture_offset, texture_mask
5523 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5525 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5526 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5528 vst1.u32 { texels }, [ block, :128 ]
5529 add block, block, #40
5531 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5532 vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ]
5534 add block, block, #24
5535 subs height, height, #1
5537 add fb_ptr, fb_ptr, fb_ptr_pitch
5538 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5542 ldmia sp!, { r4 - r11, pc }
5546 // FIXME: duplicate code with normal version :(
5547 #undef draw_mask_fb_ptr
5549 function(setup_sprite_16bpp_4x)
5550 stmdb sp!, { r4 - r11, r14 }
5551 ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]
5554 add fb_ptr, fb_ptr, y, lsl #11
5556 ldr width, [ sp, #40 ]
5557 add fb_ptr, fb_ptr, x, lsl #1
5559 ldr height, [ sp, #44 ]
5560 and left_offset, u, #0x7
5562 add texture_offset_base, u, u
5563 add width_rounded, width, #7
5565 add texture_offset_base, v, lsl #11
5566 movw left_mask_bits, #0xFFFF
5568 ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]
5569 add width_rounded, width_rounded, left_offset
5573 ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ]
5574 sub fb_ptr, fb_ptr, left_offset, lsl #1
5576 add texture_mask, texture_mask_width, texture_mask_width
5577 movw right_mask_bits, #0xFFFC
5579 and right_width, width_rounded, #0x7
5580 mvn left_mask_bits, left_mask_bits, lsl left_offset
5584 add texture_mask, texture_mask_height, lsl #11
5585 mov block_width, width_rounded, lsr #3
5587 mov right_mask_bits, right_mask_bits, lsl right_width
5588 movw fb_ptr_pitch, #(2048 + 16) * 2
5590 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4+1
5591 vmov block_masks, left_mask_bits, right_mask_bits
5593 ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5594 add block, psx_gpu, #psx_gpu_blocks_offset
5596 bic texture_offset_base, texture_offset_base, #0xF
5599 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]
5600 add block, block, num_blocks, lsl #6
5605 vext.32 block_masks_shifted, block_masks, block_masks, #1
5606 vorr.u32 block_masks, block_masks, block_masks_shifted
5607 vdup.u8 draw_mask_fb_ptr_a, block_masks[0]
5608 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5611 add num_blocks, num_blocks, block_width
5612 cmp num_blocks, #MAX_BLOCKS
5613 blgt setup_sprites_16bpp_flush
5615 and texture_block_ptr, texture_offset_base, texture_mask
5616 subs height, height, #1
5618 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5619 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5621 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5623 add texture_offset_base, texture_offset_base, #2048
5624 add fb_ptr, fb_ptr, #2048*2
5625 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5628 ldmia sp!, { r4 - r11, pc }
5631 add num_blocks, num_blocks, block_width
5632 mov texture_offset, texture_offset_base
5634 vdup.u8 draw_mask_fb_ptr_a, block_masks[0] // left_mask_bits
5635 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5637 cmp num_blocks, #MAX_BLOCKS
5638 blgt setup_sprites_16bpp_flush
5640 add texture_offset_base, texture_offset_base, #2048
5641 and texture_block_ptr, texture_offset, texture_mask
5643 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5644 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5646 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5648 subs blocks_remaining, block_width, #2*4
5649 add texture_offset, texture_offset, #16
5651 vmov.u8 draw_mask_fb_ptr_a, #0
5652 vmov.u8 draw_mask_fb_ptr_b, #0
5654 add fb_ptr, fb_ptr, #16*2
5658 and texture_block_ptr, texture_offset, texture_mask
5659 subs blocks_remaining, blocks_remaining, #4
5661 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5662 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5664 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5665 add texture_offset, texture_offset, #16
5667 add fb_ptr, fb_ptr, #16*2
5671 vdup.u8 draw_mask_fb_ptr_a, block_masks[4] // right_mask_bits
5672 vdup.u8 draw_mask_fb_ptr_b, block_masks[5]
5674 and texture_block_ptr, texture_offset, texture_mask
5675 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5677 vld1.u32 { texels }, [ texture_block_ptr, :128 ]
5679 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5680 subs height, height, #1
5682 add fb_ptr, fb_ptr, fb_ptr_pitch
5683 strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
5687 ldmia sp!, { r4 - r11, pc }
5690 #undef texture_page_ptr
5692 #undef dirty_textures_mask
5693 #undef current_texture_mask
5696 #define current_texture_page r1
5697 #define texture_page_ptr r2
5698 #define vram_ptr_a r3
5699 #define current_texture_page_x r12
5700 #define current_texture_page_y r4
5701 #define dirty_textures_mask r5
5705 #define current_texture_mask r9
5707 #define vram_ptr_b r11
5709 #define texel_block_a d0
5710 #define texel_block_b d1
5711 #define texel_block_expanded_a q1
5712 #define texel_block_expanded_b q2
5713 #define texel_block_expanded_ab q2
5714 #define texel_block_expanded_c q3
5715 #define texel_block_expanded_d q4
5716 #define texel_block_expanded_cd q3
5718 function(update_texture_4bpp_cache)
5719 stmdb sp!, { r4 - r11, r14 }
5722 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5724 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5725 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5727 and current_texture_page_x, current_texture_page, #0xF
5728 ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]
5730 mov current_texture_page_y, current_texture_page, lsr #4
5731 ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5733 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5736 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5737 bic dirty_textures_mask, current_texture_mask
5740 str dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]
5745 add vram_ptr_b, vram_ptr_a, #2048
5748 vld1.u32 { texel_block_a }, [ vram_ptr_a, :64 ], c_4096
5749 vld1.u32 { texel_block_b }, [ vram_ptr_b, :64 ], c_4096
5751 vmovl.u8 texel_block_expanded_a, texel_block_a
5752 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5753 vmovl.u8 texel_block_expanded_c, texel_block_b
5754 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5756 vbic.u16 texel_block_expanded_a, #0x00F0
5757 vbic.u16 texel_block_expanded_b, #0x00F0
5758 vbic.u16 texel_block_expanded_c, #0x00F0
5759 vbic.u16 texel_block_expanded_d, #0x00F0
5761 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
5762 texel_block_expanded_b
5763 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
5764 texel_block_expanded_d
5766 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
5767 [ texture_page_ptr, :256 ]!
5769 subs sub_y, sub_y, #1
5773 add vram_ptr_a, vram_ptr_a, #8
5774 add vram_ptr_b, vram_ptr_b, #8
5776 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5777 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5779 subs tile_x, tile_x, #1
5783 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5784 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5786 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5787 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5789 subs tile_y, tile_y, #1
5793 ldmia sp!, { r4 - r11, pc }
5796 #undef current_texture_page
5799 #define texture_page r1
5800 #define texture_page_ptr r2
5801 #define vram_ptr_a r3
5802 #define texture_page_x r12
5803 #define texture_page_y r4
5804 #define current_texture_page r5
5809 #define vram_ptr_b r11
5821 function(update_texture_8bpp_cache_slice)
5822 stmdb sp!, { r4 - r11, r14 }
5825 ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ]
5826 ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
5828 ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ]
5831 and texture_page_x, texture_page, #0xF
5832 mov texture_page_y, texture_page, lsr #4
5834 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
5837 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
5838 eor current_texture_page, current_texture_page, texture_page
5840 ands current_texture_page, current_texture_page, #0x1
5843 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
5846 add vram_ptr_b, vram_ptr_a, #2048
5849 vld1.u32 { texels_a }, [ vram_ptr_a, :128 ], c_4096
5850 vld1.u32 { texels_b }, [ vram_ptr_b, :128 ], c_4096
5851 vld1.u32 { texels_c }, [ vram_ptr_a, :128 ], c_4096
5852 vld1.u32 { texels_d }, [ vram_ptr_b, :128 ], c_4096
5854 vst1.u32 { texels_a, texels_b }, [ texture_page_ptr, :256 ]!
5855 vst1.u32 { texels_c, texels_d }, [ texture_page_ptr, :256 ]!
5857 subs sub_y, sub_y, #1
5862 add vram_ptr_a, vram_ptr_a, #16
5863 add vram_ptr_b, vram_ptr_b, #16
5865 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5866 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5868 subs tile_x, tile_x, #1
5873 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5874 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5876 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5877 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5879 subs tile_y, tile_y, #1
5880 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
5885 ldmia sp!, { r4 - r11, pc }
5888 /* void scale2x_tiles8(void *dst, const void *src, int w8, int h) */
5889 function(scale2x_tiles8)
5893 add r12, r0, #1024*2
5897 vld1.u16 { q0 }, [ r1, :128 ]!
5898 vld1.u16 { q2 }, [ r1, :128 ]!
5904 vst1.u16 { q0, q1 }, [ r0, :128 ]!
5905 vst1.u16 { q0, q1 }, [ r12, :128 ]!
5907 vst1.u16 { q2, q3 }, [ r0, :128 ]!
5908 vst1.u16 { q2, q3 }, [ r12, :128 ]!
5915 sub r0, r2, lsl #4+1
5917 add r12, r0, #1024*2
5923 // vim:filetype=armasm