2 * Copyright (C) 2011 Gilead Kutnick "Exophase" <exophase@gmail.com>
3 * Copyright (C) 2012 GraÅžvydas Ignotas "notaz" <notasas@gmail.com>
5 * This program is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU General Public License as
7 * published by the Free Software Foundation; either version 2 of
8 * the License, or (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License for more details.
18 #define MAX_BLOCKS_PER_ROW 128
20 #define RENDER_STATE_MASK_EVALUATE 0x20
21 #define RENDER_FLAGS_MODULATE_TEXELS 0x1
22 #define RENDER_FLAGS_BLEND 0x2
23 #define RENDER_INTERLACE_ENABLED 0x1
25 #include "psx_gpu_offsets.h"
27 #define psx_gpu_b_dx_offset (psx_gpu_b_block_span_offset + 4)
29 #define edge_data_left_x_offset 0
30 #define edge_data_num_blocks_offset 2
31 #define edge_data_right_mask_offset 4
32 #define edge_data_y_offset 6
78 #define gw_by_l gw_bx_l
103 #define uvrg_xxxx1 q4
108 #define uvrg_xxxx2 q5
113 #define yyyy_uvrg0 q6
118 #define yyyy_uvrg1 q7
123 #define yyyy_uvrg2 q8
150 #define uvrg_base q11
193 #define uvrgb_phase q13
197 #include "arm_features.h"
199 #define function(name) FUNCTION(name):
201 #ifndef TEXRELS_FORBIDDEN
203 #define JT_OP_REL(table_label, index_reg, temp)
204 #define JT_OP(x...) x
205 #define JTE(start, target) target
209 #define JT_OP_REL(table_label, index_reg, temp) \
210 adr temp, table_label; \
211 ldr temp, [temp, index_reg, lsl #2]; \
215 #define JTE(start, target) (target - start)
220 #define flush_render_block_buffer _flush_render_block_buffer
221 #define setup_sprite_untextured_simple _setup_sprite_untextured_simple
222 #define update_texture_8bpp_cache _update_texture_8bpp_cache
230 function(compute_all_gradients)
231 // First compute the triangle area reciprocal and shift. The division will
232 // happen concurrently with much of the work which follows.
233 @ r12 = psx_gpu->triangle_area
234 ldr r12, [psx_gpu, #psx_gpu_triangle_area_offset]
235 stmdb sp!, { r4 - r11, lr }
237 @ load exponent of 62 into upper half of double
239 clz r14, r12 @ r14 = shift
241 movt r4, #((62 + 1023) << 4)
242 mov r12, r12, lsl r14 @ r12 = triangle_area_normalized
244 @ load area normalized into lower half of double
246 vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n
248 movt r4, #((1022 + 31) << 4)
251 add r4, r4, r12, lsr #11
254 vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n
256 // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) =
257 // ( d0 * d1 ) - ( d2 * d3 ) =
258 // ( m0 ) - ( m1 ) = gradient
260 // This is split to do 12 elements at a time over three sets: a, b, and c.
261 // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so
262 // two of the slots are unused.
264 // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as
267 // First type is: uvrg bxxx xxxx
268 // Second type is: yyyy ybyy uvrg
269 // Since x_a and y_c are the same the same variable is used for both.
271 vld1.u32 { v0 }, [v_a, :128] @ v0 = { uvrg0, b0, x0, y0 }
272 ldrsh x0, [v_a, #8] @ load x0
274 vld1.u32 { v1 }, [v_b, :128] @ v1 = { uvrg1, b1, x1, y1}
275 ldrh x1, [v_b, #8] @ load x1
277 vld1.u32 { v2 }, [v_c, :128] @ v2 = { uvrg2, b2, x2, y2 }
278 ldrh x2, [v_c, #8] @ load x2
280 vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- }
281 ldrh y0, [v_a, #10] @ load y0
283 vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- }
284 ldrh y1, [v_b, #10] @ load y1
286 vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- }
287 ldrh y2, [v_c, #10] @ load y2
289 vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 }
290 vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 }
292 orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 }
293 pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 }
295 vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 }
296 vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 }
298 vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 }
299 vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 }
301 ldrb b2, [v_c, #4] @ load b2
302 orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 }
304 ldrb b1, [v_b, #4] @ load b1
305 orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 }
307 vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 }
308 vsub.s16 d0_ab, x1_ab, x0_ab
310 ldrb b0, [v_a, #4] @ load b0
311 orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 }
313 vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 }
314 vsub.s16 d2_ab, x2_ab, x1_ab
316 vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 }
317 vsub.s16 d1_ab, y2_ab, y1_ab
319 orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 }
320 ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 }
322 ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 }
323 ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 }
325 vsub.s16 d3_ab, y1_ab, y0_ab
326 smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) -
327 @ ((x2 - X1) * (b1 - b0))
328 vmull.s16 ga_uvrg_x, d0_a, d1_a
329 smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) -
330 @ ((b2 - b1) * (y1 - y0))
331 vmlsl.s16 ga_uvrg_x, d2_a, d3_a
332 movs gs_bx, ga_bx, asr #31
334 vmull.s16 ga_uvrg_y, d0_b, d1_b
335 rsbmi ga_bx, ga_bx, #0
337 @ r12 = psx_gpu->uvrgb_phase
338 ldr r12, [psx_gpu, #psx_gpu_uvrgb_phase_offset]
340 vmlsl.s16 ga_uvrg_y, d2_b, d3_b
341 movs gs_by, ga_by, asr #31
343 vshr.u64 d0, d30, #22
344 add b_base, r12, b0, lsl #16
346 vdup.u32 uvrgb_phase, r12
348 rsbmi ga_by, ga_by, #0
349 vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0
351 @ r12 = psx_gpu->triangle_winding_offset
352 ldrb r12, [psx_gpu, #psx_gpu_triangle_winding_offset]
353 vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0
355 rsb r12, r12, #0 @ r12 = -(triangle->winding)
357 vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w }
358 sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS)
360 vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16
361 vdup.u32 r_shift, r14 @ r_shift = { shift, shift, shift, shift }
363 vadd.u32 uvrg_base, uvrgb_phase
364 vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x)
366 vmov area_r_s, s0 @ area_r_s = triangle_reciprocal
367 vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y)
369 vmull.u32 gw_rg_x, ga_rg_x, d0[0]
370 vmull.u32 gw_uv_x, ga_uv_x, d0[0]
371 vmull.u32 gw_rg_y, ga_rg_y, d0[0]
372 vmull.u32 gw_uv_y, ga_uv_y, d0[0]
374 vshl.u64 gw_rg_x, gw_rg_x, r_shift
375 vshl.u64 gw_uv_x, gw_uv_x, r_shift
376 vshl.u64 gw_rg_y, gw_rg_y, r_shift
377 vshl.u64 gw_uv_y, gw_uv_y, r_shift
379 veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask
380 vmovn.u64 g_uv_x, gw_uv_x
382 veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask
383 vmovn.u64 g_rg_x, gw_rg_x
385 veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
386 vmovn.u64 g_uv_y, gw_uv_y
388 vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
389 vmovn.u64 g_rg_y, gw_rg_y
391 veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
392 mov ga_bx, ga_bx, lsl #13
394 vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
395 mov ga_by, ga_by, lsl #13
398 umull gw_bx_l, gw_bx_h, ga_bx, area_r_s
400 vshl.u32 g_uvrg_x, g_uvrg_x, #4
401 vshl.u32 g_uvrg_y, g_uvrg_y, #4
403 umull gw_by_l, gw_by_h, ga_by, area_r_s
404 vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0]
406 eor gs_bx, gs_bx, r12
407 vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1
409 veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0
410 eor gs_by, gs_by, r12
412 rsb r11, r14, #0 @ r11 = negative shift for scalar lsr
413 add store_a, psx_gpu, #psx_gpu_uvrg_offset
415 sub r11, r11, #(32 - 13)
417 add store_b, store_a, #16
420 vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1
421 vst1.u32 { uvrg_base }, [store_a, :128], store_inc
423 vst1.u32 { uvrg_dx1 }, [store_b, :128], store_inc
424 mov g_bx, gw_bx_h, lsr r11
426 vst1.u32 { g_uvrg_y }, [store_a, :128], store_inc
427 mov g_by, gw_by_h, lsr r11
429 vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \
430 [store_b, :128], store_inc
431 eor g_bx, g_bx, gs_bx
433 vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \
434 [store_b, :128], store_inc
435 sub g_bx, g_bx, gs_bx
438 eor g_by, g_by, gs_by
440 mls b_base, g_bx, x0, b_base
441 sub g_by, g_by, gs_by
446 add g_bx2, g_bx, g_bx
447 add g_bx3, g_bx, g_bx2
449 stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by }
451 ldmia sp!, { r4 - r11, pc }
468 #define height_minor_a r7
469 #define height_minor_b r8
470 #define height_major r9
473 #define reciprocal_table_ptr r10
475 #define edge_alt_low r4
476 #define edge_alt_high r5
477 #define edge_dx_dy_alt r6
478 #define edge_shift_alt r10
480 #define edge_dx_dy_alt_low r4
481 #define edge_dx_dy_alt_high r5
483 #define span_edge_data r4
484 #define span_uvrg_offset r5
485 #define span_b_offset r6
493 #define alternate_x q0
494 #define alternate_dx_dy q1
495 #define alternate_x_32 q2
497 #define alternate_x_low d0
498 #define alternate_x_high d1
499 #define alternate_dx_dy_low d2
500 #define alternate_dx_dy_high d3
501 #define alternate_x_32_low d4
502 #define alternate_x_32_high d5
506 #define left_dx_dy q5
507 #define right_dx_dy q6
509 #define right_edge q8
511 #define left_x_low d6
512 #define left_x_high d7
513 #define right_x_low d8
514 #define right_x_high d9
515 #define left_dx_dy_low d10
516 #define left_dx_dy_high d11
517 #define right_dx_dy_low d12
518 #define right_dx_dy_high d13
519 #define left_edge_low d14
520 #define left_edge_high d15
521 #define right_edge_low d16
522 #define right_edge_high d17
524 #define y_mid_point d18
527 #define left_right_x_16 q11
528 #define span_shifts_y q12
531 #define span_shifts d24
536 #define left_right_x_16_low d22
537 #define left_right_x_16_high d23
542 #define alternate_x_16 d4
545 #define v_clip_low d6
547 #define right_x_32 q10
548 #define left_x_32 q11
549 #define alternate_select d24
551 #define right_x_32_low d20
552 #define right_x_32_high d21
553 #define left_x_32_low d22
554 #define left_x_32_high d23
557 #define edges_dx_dy d2
558 #define edge_shifts d3
559 #define edge_shifts_64 q2
561 #define edges_xy_left d0
562 #define edges_xy_right d1
564 #define height_reciprocals d6
572 #define heights_b d12
573 #define edges_dx_dy_64 q10
575 #define edges_dx_dy_64_left d20
576 #define edges_dx_dy_64_right d21
579 #define setup_spans_prologue() \
580 stmdb sp!, { r4 - r11, lr }; \
582 ldrsh x_a, [v_a, #8]; \
583 ldrsh x_b, [v_b, #8]; \
584 ldrsh x_c, [v_c, #8]; \
585 ldrsh y_a, [v_a, #10]; \
586 ldrsh y_b, [v_b, #10]; \
587 ldrsh y_c, [v_c, #10]; \
589 add temp, psx_gpu, #psx_gpu_uvrg_offset; \
590 vld1.32 { uvrg }, [temp]; \
591 add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \
592 vld1.32 { uvrg_dy }, [temp]; \
593 ldr reciprocal_table_ptr, [psx_gpu, #psx_gpu_reciprocal_table_ptr_offset]; \
595 vmov.u32 c_0x01, #0x01 \
597 #define setup_spans_load_b() \
598 ldr b, [psx_gpu, #psx_gpu_b_offset]; \
599 ldr b_dy, [psx_gpu, #psx_gpu_b_dy_offset] \
601 #define setup_spans_prologue_b() \
602 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
603 add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \
605 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
606 vmov.u16 c_0x0004, #0x0004; \
608 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
609 vmov.u16 c_0x0001, #0x0001; \
611 vld1.u16 { left_edge_low[], left_edge_high[] }, [temp]; \
612 add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \
614 vld1.u16 { right_edge_low[], right_edge_high[] }, [temp]; \
615 vadd.u16 right_edge, right_edge, c_0x0001; \
617 vmov.u16 c_0x0007, #0x0007; \
618 vmvn.u16 c_0xFFFE, #0x0001 \
621 #define compute_edge_delta_x2() \
622 ldr temp, [reciprocal_table_ptr, height, lsl #2]; \
624 vdup.u32 heights, height; \
625 vsub.u32 widths, x_ends, x_starts; \
627 vdup.u32 edge_shifts, temp; \
628 vsub.u32 heights_b, heights, c_0x01; \
629 vshr.u32 height_reciprocals, edge_shifts, #10; \
631 vmla.s32 heights_b, x_starts, heights; \
632 vbic.u16 edge_shifts, #0xE0; \
633 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
634 vmull.s32 edges_xy, heights_b, height_reciprocals \
637 #define height_reciprocal_alt r11
638 #define height_b_alt r12
640 #define compute_edge_delta_x3(start_c, height_a, height_b) \
641 vmov heights, height_a, height_b; \
642 ldr temp, [reciprocal_table_ptr, height_a, lsl #2]; \
643 vmov.u32 edge_shifts[0], temp; \
644 ldr temp, [reciprocal_table_ptr, height_b, lsl #2]; \
645 vmov.u32 edge_shifts[1], temp; \
646 ldr edge_shift_alt, [reciprocal_table_ptr, height_minor_b, lsl #2]; \
648 vsub.u32 widths, x_ends, x_starts; \
649 sub width_alt, x_c, start_c; \
651 vsub.u32 heights_b, heights, c_0x01; \
652 sub height_b_alt, height_minor_b, #1; \
654 vshr.u32 height_reciprocals, edge_shifts, #10; \
655 lsr height_reciprocal_alt, edge_shift_alt, #10; \
657 vmla.s32 heights_b, x_starts, heights; \
658 mla height_b_alt, height_minor_b, start_c, height_b_alt; \
660 vbic.u16 edge_shifts, #0xE0; \
661 and edge_shift_alt, edge_shift_alt, #0x1F; \
663 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
664 mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \
666 vmull.s32 edges_xy, heights_b, height_reciprocals; \
667 smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \
670 #define setup_spans_adjust_y_up() \
671 vsub.u32 y_x4, y_x4, c_0x0004 \
673 #define setup_spans_adjust_y_down() \
674 vadd.u32 y_x4, y_x4, c_0x0004 \
676 #define setup_spans_adjust_interpolants_up() \
677 vsub.u32 uvrg, uvrg, uvrg_dy; \
680 #define setup_spans_adjust_interpolants_down() \
681 vadd.u32 uvrg, uvrg, uvrg_dy; \
685 #define setup_spans_clip_interpolants_increment() \
686 mla b, b_dy, clip, b; \
687 vmla.s32 uvrg, uvrg_dy, v_clip \
689 #define setup_spans_clip_interpolants_decrement() \
690 mls b, b_dy, clip, b; \
691 vmls.s32 uvrg, uvrg_dy, v_clip \
693 #define setup_spans_clip_alternate_yes() \
694 smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \
696 #define setup_spans_clip_alternate_no() \
698 #define setup_spans_clip(direction, alternate_active) \
699 vdup.u32 v_clip, clip; \
700 setup_spans_clip_alternate_##alternate_active(); \
701 setup_spans_clip_interpolants_##direction(); \
702 vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \
705 #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \
706 vmovl.s32 edge_shifts_64, edge_shifts; \
707 vmovl.s32 edges_dx_dy_64, edges_dx_dy; \
709 vshl.s64 edges_xy, edges_xy, edge_shifts_64; \
710 vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \
712 vmov left_x_low, edges_xy_##left_index; \
713 vmov right_x_low, edges_xy_##right_index; \
715 vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \
716 vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \
717 vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \
718 vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \
720 vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \
721 vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \
723 vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \
724 vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \
727 #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \
728 setup_spans_adjust_edges_alternate_no(left_index, right_index); \
730 vdup.u16 y_mid_point, y_b; \
731 rsb temp, edge_shift_alt, #32; \
733 lsl edge_alt_high, edge_alt_high, edge_shift_alt; \
734 orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \
735 lsl edge_alt_low, edge_alt_low, edge_shift_alt; \
736 vmov alternate_x_low, edge_alt_low, edge_alt_high; \
738 asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \
739 lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \
740 vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \
741 vmov alternate_dx_dy_high, alternate_dx_dy_low; \
743 vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \
744 vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \
747 #define setup_spans_y_select_up() \
748 vclt.s16 alternate_select, y_x4, y_mid_point \
750 #define setup_spans_y_select_down() \
751 vcgt.s16 alternate_select, y_x4, y_mid_point \
754 #define setup_spans_alternate_select_left() \
755 vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \
757 #define setup_spans_alternate_select_right() \
758 vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \
761 #define setup_spans_set_x4_alternate_yes(alternate, direction) \
762 vshrn.s64 alternate_x_32_low, alternate_x, #32; \
763 vshrn.s64 left_x_32_low, left_x, #32; \
764 vshrn.s64 right_x_32_low, right_x, #32; \
766 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
767 vadd.u64 left_x, left_x, left_dx_dy; \
768 vadd.u64 right_x, right_x, right_dx_dy; \
770 vshrn.s64 alternate_x_32_high, alternate_x, #32; \
771 vshrn.s64 left_x_32_high, left_x, #32; \
772 vshrn.s64 right_x_32_high, right_x, #32; \
774 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
775 vadd.u64 left_x, left_x, left_dx_dy; \
776 vadd.u64 right_x, right_x, right_dx_dy; \
778 vmovn.u32 alternate_x_16, alternate_x_32; \
779 setup_spans_y_select_##direction(); \
780 vmovn.u32 left_right_x_16_low, left_x_32; \
782 vmovn.u32 left_right_x_16_high, right_x_32; \
783 setup_spans_alternate_select_##alternate(); \
785 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
786 str b, [span_b_offset], #4; \
787 setup_spans_adjust_interpolants_##direction(); \
789 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
791 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
792 str b, [span_b_offset], #4; \
793 setup_spans_adjust_interpolants_##direction(); \
795 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
797 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
798 str b, [span_b_offset], #4; \
799 setup_spans_adjust_interpolants_##direction(); \
801 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
802 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
803 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
805 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
806 str b, [span_b_offset], #4; \
807 setup_spans_adjust_interpolants_##direction(); \
809 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
810 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
812 vst4.u16 { left_right_x_16, span_shifts_y }, [span_edge_data]!; \
814 setup_spans_adjust_y_##direction() \
817 #define setup_spans_set_x4_alternate_no(alternate, direction) \
818 vshrn.s64 left_x_32_low, left_x, #32; \
819 vshrn.s64 right_x_32_low, right_x, #32; \
821 vadd.u64 left_x, left_x, left_dx_dy; \
822 vadd.u64 right_x, right_x, right_dx_dy; \
824 vshrn.s64 left_x_32_high, left_x, #32; \
825 vshrn.s64 right_x_32_high, right_x, #32; \
827 vadd.u64 left_x, left_x, left_dx_dy; \
828 vadd.u64 right_x, right_x, right_dx_dy; \
830 vmovn.u32 left_right_x_16_low, left_x_32; \
831 vmovn.u32 left_right_x_16_high, right_x_32; \
833 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
834 str b, [span_b_offset], #4; \
835 setup_spans_adjust_interpolants_##direction(); \
837 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
839 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
840 str b, [span_b_offset], #4; \
841 setup_spans_adjust_interpolants_##direction(); \
843 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
845 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
846 str b, [span_b_offset], #4; \
847 setup_spans_adjust_interpolants_##direction(); \
849 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
850 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
851 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
853 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
854 str b, [span_b_offset], #4; \
855 setup_spans_adjust_interpolants_##direction(); \
857 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
858 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
860 vst4.u16 { left_right_x_16, span_shifts_y }, [span_edge_data]!; \
862 setup_spans_adjust_y_##direction() \
865 #define edge_adjust_low r11
866 #define edge_adjust_high r12
868 #define setup_spans_alternate_adjust_yes() \
869 smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \
870 subs edge_alt_low, edge_alt_low, edge_adjust_low; \
871 sbc edge_alt_high, edge_alt_high, edge_adjust_high \
873 #define setup_spans_alternate_adjust_no() \
876 #define setup_spans_down(left_index, right_index, alternate, alternate_active) \
877 setup_spans_alternate_adjust_##alternate_active(); \
878 setup_spans_load_b(); \
880 ldrsh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]; \
881 subs y_c, y_c, temp; \
882 subgt height, height, y_c; \
883 addgt height, height, #1; \
885 ldrsh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]; \
886 subs clip, temp, y_a; \
889 sub height, height, clip; \
890 add y_a, y_a, clip; \
891 setup_spans_clip(increment, alternate_active); \
897 orr temp, y_a, y_a, lsl #16; \
898 add temp, temp, #(1 << 16); \
900 add y_a, y_a, #(2 << 16); \
901 vmov y_x4, temp, y_a; \
903 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
905 setup_spans_prologue_b(); \
907 strh height, [psx_gpu, #psx_gpu_num_spans_offset]; \
910 setup_spans_set_x4_alternate_##alternate_active(alternate, down); \
911 subs height, height, #4; \
917 #define setup_spans_alternate_pre_increment_yes() \
918 adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \
919 adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \
921 #define setup_spans_alternate_pre_increment_no() \
924 #define setup_spans_up_decrement_yes() \
925 suble height, height, #1 \
927 #define setup_spans_up_decrement_no() \
930 #define setup_spans_up(left_index, right_index, alternate, alternate_active) \
931 setup_spans_alternate_adjust_##alternate_active(); \
932 setup_spans_load_b(); \
935 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]; \
936 subs temp, temp, y_c; \
937 subgt height, height, temp; \
938 setup_spans_up_decrement_##alternate_active(); \
940 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]; \
941 subs clip, y_a, temp; \
944 sub height, height, clip; \
945 sub y_a, y_a, clip; \
946 setup_spans_clip(decrement, alternate_active); \
952 orr temp, y_a, y_a, lsl #16; \
953 sub temp, temp, #(1 << 16); \
955 sub y_a, y_a, #(2 << 16); \
956 vmov y_x4, temp, y_a; \
958 vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \
960 setup_spans_alternate_pre_increment_##alternate_active(); \
961 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
963 setup_spans_adjust_interpolants_up(); \
964 setup_spans_prologue_b(); \
966 strh height, [psx_gpu, #psx_gpu_num_spans_offset]; \
969 setup_spans_set_x4_alternate_##alternate_active(alternate, up); \
970 subs height, height, #4; \
976 #define setup_spans_epilogue() \
977 ldmia sp!, { r4 - r11, pc } \
980 #define setup_spans_up_up(minor, major) \
981 setup_spans_prologue(); \
982 sub height_minor_a, y_a, y_b; \
983 sub height_minor_b, y_b, y_c; \
984 sub height, y_a, y_c; \
986 vdup.u32 x_starts, x_a; \
987 vmov x_ends, x_c, x_b; \
989 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
990 setup_spans_up(major, minor, minor, yes); \
991 setup_spans_epilogue() \
993 function(setup_spans_up_left)
994 setup_spans_up_up(left, right)
996 function(setup_spans_up_right)
997 setup_spans_up_up(right, left)
999 #define setup_spans_down_down(minor, major) \
1000 setup_spans_prologue(); \
1001 sub height_minor_a, y_b, y_a; \
1002 sub height_minor_b, y_c, y_b; \
1003 sub height, y_c, y_a; \
1005 vdup.u32 x_starts, x_a; \
1006 vmov x_ends, x_c, x_b; \
1008 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
1009 setup_spans_down(major, minor, minor, yes); \
1010 setup_spans_epilogue() \
1012 function(setup_spans_down_left)
1013 setup_spans_down_down(left, right)
1015 function(setup_spans_down_right)
1016 setup_spans_down_down(right, left)
1019 #define setup_spans_up_flat() \
1020 sub height, y_a, y_c; \
1022 compute_edge_delta_x2(); \
1023 setup_spans_up(left, right, none, no); \
1024 setup_spans_epilogue() \
1026 function(setup_spans_up_a)
1027 setup_spans_prologue()
1029 vmov x_starts, x_a, x_b
1030 vdup.u32 x_ends, x_c
1032 setup_spans_up_flat()
1034 function(setup_spans_up_b)
1035 setup_spans_prologue()
1037 vdup.u32 x_starts, x_a
1038 vmov x_ends, x_b, x_c
1040 setup_spans_up_flat()
1042 #define setup_spans_down_flat() \
1043 sub height, y_c, y_a; \
1045 compute_edge_delta_x2(); \
1046 setup_spans_down(left, right, none, no); \
1047 setup_spans_epilogue() \
1049 function(setup_spans_down_a)
1050 setup_spans_prologue()
1052 vmov x_starts, x_a, x_b
1053 vdup.u32 x_ends, x_c
1055 setup_spans_down_flat()
1057 function(setup_spans_down_b)
1058 setup_spans_prologue()
1060 vdup.u32 x_starts, x_a
1061 vmov x_ends, x_b, x_c
1063 setup_spans_down_flat()
1068 #define edges_xy_b q11
1069 #define edges_dx_dy_b d26
1070 #define edge_shifts_b d27
1071 #define edges_dx_dy_and_shifts_b q13
1072 #define height_increment d20
1074 #define edges_dx_dy_and_shifts q1
1076 #define edges_xy_b_left d22
1077 #define edges_xy_b_right d23
1079 #define setup_spans_up_down_load_edge_set_b() \
1080 vmov edges_xy, edges_xy_b; \
1081 vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \
1084 function(setup_spans_up_down)
1085 setup_spans_prologue()
1087 // s32 middle_y = y_a;
1088 sub height_minor_a, y_a, y_b
1089 sub height_minor_b, y_c, y_a
1090 sub height_major, y_c, y_b
1092 vmov x_starts, x_a, x_c
1093 vdup.u32 x_ends, x_b
1095 compute_edge_delta_x3(x_a, height_minor_a, height_major)
1098 vmov height_increment, temp, height_minor_b
1099 vmlal.s32 edges_xy, edges_dx_dy, height_increment
1101 vmov edges_xy_b_left, edge_alt_low, edge_alt_high
1102 vmov edges_xy_b_right, edges_xy_right
1104 vmov edge_shifts_b, edge_shifts
1105 vmov.u32 edge_shifts_b[0], edge_shift_alt
1107 vneg.s32 edges_dx_dy_b, edges_dx_dy
1108 vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt
1112 setup_spans_load_b()
1115 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]
1116 subs temp, temp, y_b
1117 subgt height_minor_a, height_minor_a, temp
1119 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]
1120 subs clip, y_a, temp
1123 sub height_minor_a, height_minor_a, clip
1125 setup_spans_clip(decrement, no)
1128 cmp height_minor_a, #0
1131 orr temp, y_a, y_a, lsl #16
1132 sub temp, temp, #(1 << 16)
1134 sub y_a, y_a, #(2 << 16)
1135 vmov y_x4, temp, y_a
1137 vaddw.s32 edges_xy, edges_xy, edges_dx_dy
1139 strh height_minor_a, [psx_gpu, #psx_gpu_num_spans_offset]
1141 setup_spans_adjust_edges_alternate_no(left, right);
1142 setup_spans_adjust_interpolants_up()
1143 setup_spans_up_down_load_edge_set_b()
1145 setup_spans_prologue_b()
1149 setup_spans_set_x4_alternate_no(none, up)
1150 subs height_minor_a, height_minor_a, #4
1153 add span_edge_data, span_edge_data, height_minor_a, lsl #3
1154 add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4
1155 add span_b_offset, span_b_offset, height_minor_a, lsl #2
1158 add temp, psx_gpu, #psx_gpu_uvrg_offset
1159 vld1.32 { uvrg }, [temp]
1162 setup_spans_load_b()
1164 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]
1166 subgt height_minor_b, height_minor_b, y_c
1167 addgt height_minor_b, height_minor_b, #1
1169 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]
1170 subs clip, temp, y_a
1173 sub height_minor_b, height_minor_b, clip
1175 setup_spans_clip(increment, no)
1178 cmp height_minor_b, #0
1181 orr temp, y_a, y_a, lsl #16
1182 add temp, temp, #(1 << 16)
1184 add y_a, y_a, #(2 << 16)
1185 vmov y_x4, temp, y_a
1187 setup_spans_adjust_edges_alternate_no(left, right)
1189 ldrh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1190 add temp, temp, height_minor_b
1192 cmp temp, #MAX_SPANS
1195 strh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1198 setup_spans_set_x4_alternate_no(none, down)
1199 subs height_minor_b, height_minor_b, #4
1203 setup_spans_epilogue()
1206 setup_spans_up_down_load_edge_set_b()
1207 setup_spans_prologue_b()
1211 // FIXME: overflow corner case
1212 sub temp, temp, height_minor_b
1213 bics height_minor_b, #3
1214 add temp, temp, height_minor_b
1215 strh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1219 #undef span_uvrg_offset
1220 #undef span_edge_data
1221 #undef span_b_offset
1226 #define num_spans r1
1227 #define span_uvrg_offset r2
1228 #define span_edge_data r3
1229 #define span_b_offset r4
1231 #define span_num_blocks r6
1235 #define dither_offset_ptr r10
1236 #define block_ptr_a r11
1238 #define num_blocks r14
1240 #define uvrg_dx_ptr r2
1241 #define texture_mask_ptr r3
1242 #define dither_shift r8
1243 #define dither_row r10
1248 #define block_ptr_b r10
1250 #define block_span_ptr r10
1251 #define right_mask r8
1270 #define b_whole_8 d14
1271 #define fb_mask_ptrs d15
1284 #define u_whole_low d16
1285 #define u_whole_high d17
1286 #define v_whole_low d18
1287 #define v_whole_high d19
1288 #define r_whole_low d20
1289 #define r_whole_high d21
1290 #define g_whole_low d22
1291 #define g_whole_high d23
1292 #define b_whole_low d24
1293 #define b_whole_high d25
1298 #define u_whole_8 d26
1299 #define v_whole_8 d27
1300 #define u_whole_8b d24
1301 #define r_whole_8 d24
1302 #define g_whole_8 d25
1304 #define uv_whole_8 q13
1305 #define uv_whole_8b q14
1307 #define dither_offsets q14
1308 #define texture_mask q15
1309 #define texture_mask_u d30
1310 #define texture_mask_v d31
1312 #define dither_offsets_short d28
1316 #define block_span q10
1321 #define draw_mask q1
1322 #define draw_mask_edge q13
1323 #define test_mask q0
1329 #define setup_blocks_texture_swizzled() \
1330 vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \
1331 vsli.u8 u_whole_8, v_whole_8, #4; \
1332 vsri.u8 v_whole_8, u_whole_8b, #4 \
1334 #define setup_blocks_texture_unswizzled() \
1337 #define setup_blocks_shaded_textured_builder(swizzling) \
1340 function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \
1341 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
1342 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1344 vld1.u32 { uvrg_dx }, [uvrg_dx_ptr, :128]; \
1345 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1347 cmp num_spans, #0; \
1350 stmdb sp!, { r4 - r11, r14 }; \
1351 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1353 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
1354 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1356 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [texture_mask_ptr, :16]; \
1357 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1359 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1360 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1362 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
1363 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1365 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1368 vmov.u8 fb_mask_ptrs, #0; \
1370 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
1371 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1373 ldrh y, [span_edge_data, #edge_data_y_offset]; \
1374 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
1376 cmp span_num_blocks, #0; \
1379 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
1380 add num_blocks, span_num_blocks, num_blocks; \
1382 cmp num_blocks, #MAX_BLOCKS; \
1386 ldr b, [span_b_offset]; \
1387 add fb_ptr, fb_ptr, y, lsl #11; \
1389 vdup.u32 v_left_x, left_x; \
1392 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
1393 add fb_ptr, fb_ptr, left_x, lsl #1; \
1395 mla b, b_dx, left_x, b; \
1396 and dither_shift, left_x, #0x03; \
1398 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
1399 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1401 mov dither_shift, dither_shift, lsl #3; \
1402 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1405 subs span_num_blocks, span_num_blocks, #1; \
1407 mov dither_row, dither_row, ror dither_shift; \
1408 mov b_dx4, b_dx, lsl #2; \
1410 vdup.u32 dither_offsets_short, dither_row; \
1411 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1413 vdup.u32 b_block, b; \
1414 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1416 vdup.u32 u_block, uv[0]; \
1417 mov b_dx8, b_dx, lsl #3; \
1419 vdup.u32 v_block, uv[1]; \
1420 vdup.u32 r_block, rg[0]; \
1421 vdup.u32 g_block, rg[1]; \
1423 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1425 vadd.u32 u_block, u_block, block_span; \
1426 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1428 vadd.u32 v_block, v_block, block_span; \
1429 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1431 vadd.u32 r_block, r_block, block_span; \
1432 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1434 vadd.u32 g_block, g_block, block_span; \
1435 vld1.u32 { block_span }, [block_span_ptr, :128]; \
1437 vadd.u32 b_block, b_block, block_span; \
1438 add block_ptr_b, block_ptr_a, #16; \
1440 vshrn.u32 u_whole_low, u_block, #16; \
1441 vshrn.u32 v_whole_low, v_block, #16; \
1442 vshrn.u32 r_whole_low, r_block, #16; \
1443 vshrn.u32 g_whole_low, g_block, #16; \
1445 vdup.u32 dx4, uv_dx4[0]; \
1446 vshrn.u32 b_whole_low, b_block, #16; \
1448 vaddhn.u32 u_whole_high, u_block, dx4; \
1449 vdup.u32 dx4, uv_dx4[1]; \
1451 vaddhn.u32 v_whole_high, v_block, dx4; \
1452 vdup.u32 dx4, rg_dx4[0]; \
1454 vaddhn.u32 r_whole_high, r_block, dx4; \
1455 vdup.u32 dx4, rg_dx4[1]; \
1457 vaddhn.u32 g_whole_high, g_block, dx4; \
1458 vdup.u32 dx4, b_dx4; \
1460 vaddhn.u32 b_whole_high, b_block, dx4; \
1461 vdup.u32 dx8, uv_dx8[0]; \
1463 vadd.u32 u_block, u_block, dx8; \
1464 vdup.u32 dx8, uv_dx8[1]; \
1466 vadd.u32 v_block, v_block, dx8; \
1467 vdup.u32 dx8, rg_dx8[0]; \
1469 vadd.u32 r_block, r_block, dx8; \
1470 vdup.u32 dx8, rg_dx8[1]; \
1472 vadd.u32 g_block, g_block, dx8; \
1473 vdup.u32 dx8, b_dx8; \
1475 vadd.u32 b_block, b_block, dx8; \
1476 vmovn.u16 u_whole_8, u_whole; \
1478 vmovn.u16 v_whole_8, v_whole; \
1480 vmovn.u16 b_whole_8, b_whole; \
1482 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1484 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1485 setup_blocks_texture_##swizzling(); \
1487 vmovn.u16 r_whole_8, r_whole; \
1491 vmovn.u16 g_whole_8, g_whole; \
1492 vshrn.u32 u_whole_low, u_block, #16; \
1494 vst2.u8 { u_whole_8, v_whole_8 }, [block_ptr_a, :128], c_32; \
1495 vshrn.u32 v_whole_low, v_block, #16; \
1497 vst1.u32 { r_whole_8, g_whole_8 }, [block_ptr_b, :128], c_32; \
1498 vshrn.u32 r_whole_low, r_block, #16; \
1500 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1501 vshrn.u32 g_whole_low, g_block, #16; \
1503 vdup.u32 dx4, uv_dx4[0]; \
1504 vshrn.u32 b_whole_low, b_block, #16; \
1506 vaddhn.u32 u_whole_high, u_block, dx4; \
1507 vdup.u32 dx4, uv_dx4[1]; \
1509 vaddhn.u32 v_whole_high, v_block, dx4; \
1510 vdup.u32 dx4, rg_dx4[0]; \
1512 vaddhn.u32 r_whole_high, r_block, dx4; \
1513 vdup.u32 dx4, rg_dx4[1]; \
1515 vaddhn.u32 g_whole_high, g_block, dx4; \
1516 vdup.u32 dx4, b_dx4; \
1518 vaddhn.u32 b_whole_high, b_block, dx4; \
1519 vdup.u32 dx8, uv_dx8[0]; \
1521 vadd.u32 u_block, u_block, dx8; \
1522 vdup.u32 dx8, uv_dx8[1]; \
1524 vadd.u32 v_block, v_block, dx8; \
1525 vdup.u32 dx8, rg_dx8[0]; \
1527 vadd.u32 r_block, r_block, dx8; \
1528 vdup.u32 dx8, rg_dx8[1]; \
1530 vadd.u32 g_block, g_block, dx8; \
1531 vdup.u32 dx8, b_dx8; \
1533 vadd.u32 b_block, b_block, dx8; \
1534 vmovn.u16 u_whole_8, u_whole; \
1536 add fb_ptr, fb_ptr, #16; \
1537 vmovn.u16 v_whole_8, v_whole; \
1539 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1540 vmovn.u16 b_whole_8, b_whole; \
1544 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1545 subs span_num_blocks, span_num_blocks, #1; \
1547 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1548 setup_blocks_texture_##swizzling(); \
1550 vmovn.u16 r_whole_8, r_whole; \
1554 vmovn.u16 g_whole_8, g_whole; \
1555 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
1557 vld1.u32 { test_mask }, [psx_gpu, :128]; \
1558 vdup.u8 draw_mask, right_mask; \
1560 vmov.u32 fb_mask_ptrs[0], right_mask; \
1561 vtst.u16 draw_mask, draw_mask, test_mask; \
1562 vzip.u8 u_whole_8, v_whole_8; \
1564 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1565 vst1.u32 { r_whole_8, g_whole_8 }, [block_ptr_b, :128], c_32; \
1566 vst1.u32 { uv_whole_8 }, [block_ptr_a, :128], c_32; \
1567 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1568 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1571 add span_uvrg_offset, span_uvrg_offset, #16; \
1572 add span_b_offset, span_b_offset, #4; \
1574 add span_edge_data, span_edge_data, #8; \
1575 subs num_spans, num_spans, #1; \
1577 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1580 ldmia sp!, { r4 - r11, pc }; \
1583 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1584 vpush { texture_mask }; \
1585 vpush { uvrg_dx4 }; \
1587 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
1588 bl flush_render_block_buffer; \
1589 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
1591 vpop { uvrg_dx4 }; \
1592 vpop { texture_mask }; \
1594 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1595 vmov.u8 fb_mask_ptrs, #0; \
1597 mov num_blocks, span_num_blocks; \
1598 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1602 setup_blocks_shaded_textured_builder(swizzled)
1603 setup_blocks_shaded_textured_builder(unswizzled)
1606 #define setup_blocks_unshaded_textured_builder(swizzling) \
1609 function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \
1610 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
1611 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1613 vld1.u32 { uvrg_dx }, [uvrg_dx_ptr, :128]; \
1614 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1616 cmp num_spans, #0; \
1619 stmdb sp!, { r4 - r11, r14 }; \
1620 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1622 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1624 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [texture_mask_ptr, :16]; \
1625 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1627 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1628 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1630 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1632 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1635 vmov.u8 fb_mask_ptrs, #0; \
1637 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
1638 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1640 ldrh y, [span_edge_data, #edge_data_y_offset]; \
1641 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
1643 cmp span_num_blocks, #0; \
1646 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
1647 add num_blocks, span_num_blocks, num_blocks; \
1649 cmp num_blocks, #MAX_BLOCKS; \
1653 add fb_ptr, fb_ptr, y, lsl #11; \
1655 vdup.u32 v_left_x, left_x; \
1658 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
1659 add fb_ptr, fb_ptr, left_x, lsl #1; \
1661 and dither_shift, left_x, #0x03; \
1663 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
1664 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1666 mov dither_shift, dither_shift, lsl #3; \
1667 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1670 subs span_num_blocks, span_num_blocks, #1; \
1672 mov dither_row, dither_row, ror dither_shift; \
1674 vdup.u32 dither_offsets_short, dither_row; \
1675 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1677 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1679 vdup.u32 u_block, uv[0]; \
1681 vdup.u32 v_block, uv[1]; \
1682 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1684 vadd.u32 u_block, u_block, block_span; \
1685 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1687 vadd.u32 v_block, v_block, block_span; \
1688 add block_ptr_b, block_ptr_a, #16; \
1690 vshrn.u32 u_whole_low, u_block, #16; \
1691 vshrn.u32 v_whole_low, v_block, #16; \
1693 vdup.u32 dx4, uv_dx4[0]; \
1695 vaddhn.u32 u_whole_high, u_block, dx4; \
1696 vdup.u32 dx4, uv_dx4[1]; \
1698 vaddhn.u32 v_whole_high, v_block, dx4; \
1699 vdup.u32 dx8, uv_dx8[0]; \
1701 vadd.u32 u_block, u_block, dx8; \
1702 vdup.u32 dx8, uv_dx8[1]; \
1704 vadd.u32 v_block, v_block, dx8; \
1705 vmovn.u16 u_whole_8, u_whole; \
1707 vmovn.u16 v_whole_8, v_whole; \
1710 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1712 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1713 setup_blocks_texture_##swizzling(); \
1718 vshrn.u32 u_whole_low, u_block, #16; \
1720 vst2.u8 { u_whole_8, v_whole_8 }, [block_ptr_a, :128], c_32; \
1721 vshrn.u32 v_whole_low, v_block, #16; \
1723 add block_ptr_b, block_ptr_b, #32; \
1724 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1726 vdup.u32 dx4, uv_dx4[0]; \
1727 vaddhn.u32 u_whole_high, u_block, dx4; \
1728 vdup.u32 dx4, uv_dx4[1]; \
1730 vaddhn.u32 v_whole_high, v_block, dx4; \
1731 vdup.u32 dx8, uv_dx8[0]; \
1733 vadd.u32 u_block, u_block, dx8; \
1734 vdup.u32 dx8, uv_dx8[1]; \
1736 vadd.u32 v_block, v_block, dx8; \
1737 vmovn.u16 u_whole_8, u_whole; \
1739 add fb_ptr, fb_ptr, #16; \
1740 vmovn.u16 v_whole_8, v_whole; \
1742 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1745 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1746 subs span_num_blocks, span_num_blocks, #1; \
1748 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1749 setup_blocks_texture_##swizzling(); \
1754 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
1756 vld1.u32 { test_mask }, [psx_gpu, :128]; \
1757 vdup.u8 draw_mask, right_mask; \
1759 vmov.u32 fb_mask_ptrs[0], right_mask; \
1760 vtst.u16 draw_mask, draw_mask, test_mask; \
1761 vzip.u8 u_whole_8, v_whole_8; \
1763 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1764 add block_ptr_b, block_ptr_b, #32; \
1765 vst1.u32 { uv_whole_8 }, [block_ptr_a, :128], c_32; \
1766 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1767 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1770 add span_uvrg_offset, span_uvrg_offset, #16; \
1771 add span_edge_data, span_edge_data, #8; \
1772 subs num_spans, num_spans, #1; \
1774 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1777 ldmia sp!, { r4 - r11, pc }; \
1780 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1781 vpush { texture_mask }; \
1782 vpush { uvrg_dx4 }; \
1784 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
1785 bl flush_render_block_buffer; \
1786 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
1788 vpop { uvrg_dx4 }; \
1789 vpop { texture_mask }; \
1791 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1792 vmov.u8 fb_mask_ptrs, #0; \
1794 mov num_blocks, span_num_blocks; \
1795 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1799 setup_blocks_unshaded_textured_builder(swizzled)
1800 setup_blocks_unshaded_textured_builder(unswizzled)
1805 function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect)
1806 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]
1807 veor.u32 draw_mask, draw_mask, draw_mask
1812 stmdb sp!, { r4 - r11, r14 }
1813 vld1.u32 { test_mask }, [psx_gpu, :128]
1815 ldr color, [psx_gpu, #psx_gpu_triangle_color_offset]
1817 ubfx color_r, color, #3, #5
1818 ubfx color_g, color, #11, #5
1819 ubfx color_b, color, #19, #5
1821 orr color, color_r, color_b, lsl #10
1822 orr color, color, color_g, lsl #5
1824 vdup.u16 colors, color
1826 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
1827 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1829 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1830 add block_ptr_a, block_ptr_a, num_blocks, lsl #6
1833 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]
1834 ldrh y, [span_edge_data, #edge_data_y_offset]
1836 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
1838 cmp span_num_blocks, #0
1841 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]
1842 add num_blocks, span_num_blocks, num_blocks
1844 cmp num_blocks, #MAX_BLOCKS
1848 add fb_ptr, fb_ptr, y, lsl #11
1851 add fb_ptr, fb_ptr, left_x, lsl #1
1854 subs span_num_blocks, span_num_blocks, #1
1856 add block_ptr_b, block_ptr_a, #16
1859 vmov.u32 fb_mask_ptrs[1], fb_ptr
1863 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_32
1864 vst1.u32 { colors }, [block_ptr_b, :128], c_32
1865 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32
1867 add fb_ptr, fb_ptr, #16
1868 add block_ptr_b, block_ptr_b, #32
1872 vmov.u32 fb_mask_ptrs[1], fb_ptr
1873 subs span_num_blocks, span_num_blocks, #1
1878 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]
1880 vdup.u8 draw_mask_edge, right_mask
1881 vtst.u16 draw_mask_edge, draw_mask_edge, test_mask
1883 vst1.u32 { colors }, [block_ptr_b, :128], c_32
1884 vst1.u32 { draw_mask_edge }, [block_ptr_a, :128], c_32
1885 add block_ptr_b, block_ptr_b, #32
1886 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32
1889 add span_edge_data, span_edge_data, #8
1890 subs num_spans, num_spans, #1
1892 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
1895 ldmia sp!, { r4 - r11, pc }
1900 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
1901 bl flush_render_block_buffer
1902 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
1906 vld1.u32 { test_mask }, [psx_gpu, :128]
1907 veor.u32 draw_mask, draw_mask, draw_mask
1909 mov num_blocks, span_num_blocks
1910 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1914 #define mask_msb_scalar r14
1916 #define msb_mask q15
1918 #define pixels_low d16
1920 #define msb_mask_low d30
1921 #define msb_mask_high d31
1926 function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct)
1927 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]
1932 stmdb sp!, { r4 - r11, r14 }
1934 ldr color, [psx_gpu, #psx_gpu_triangle_color_offset]
1936 ubfx color_r, color, #3, #5
1937 ubfx color_g, color, #11, #5
1939 ldrh mask_msb_scalar, [psx_gpu, #psx_gpu_mask_msb_offset]
1940 ubfx color_b, color, #19, #5
1942 orr color, color_r, color_b, lsl #10
1943 orr color, color, color_g, lsl #5
1944 orr color, color, mask_msb_scalar
1946 vdup.u16 colors, color
1948 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1949 orr color, color, color, lsl #16
1953 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]
1954 ldrh y, [span_edge_data, #edge_data_y_offset]
1956 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
1958 cmp span_num_blocks, #0
1961 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]
1963 add fb_ptr, fb_ptr, y, lsl #11
1964 subs span_num_blocks, span_num_blocks, #1
1966 add fb_ptr, fb_ptr, left_x, lsl #1
1970 vst1.u32 { colors }, [fb_ptr]!
1971 subs span_num_blocks, span_num_blocks, #1
1976 ldrb right_mask, [span_edge_data, #edge_data_right_mask_offset]
1978 cmp right_mask, #0x0
1981 tst right_mask, #0xF
1982 streq color, [fb_ptr], #4
1983 moveq right_mask, right_mask, lsr #4
1984 streq color, [fb_ptr], #4
1986 tst right_mask, #0x3
1987 streq color, [fb_ptr], #4
1988 moveq right_mask, right_mask, lsr #2
1990 tst right_mask, #0x1
1991 strheq color, [fb_ptr]
1994 add span_edge_data, span_edge_data, #8
1995 subs num_spans, num_spans, #1
1998 ldmia sp!, { r4 - r11, pc }
2001 vst1.u32 { colors }, [fb_ptr]
2008 #define rg_dx_ptr r2
2026 #undef dither_offsets
2046 #define r_whole_low d6
2047 #define r_whole_high d7
2048 #define g_whole_low d8
2049 #define g_whole_high d9
2050 #define b_whole_low d10
2051 #define b_whole_high d11
2053 #define gb_whole_8 q6
2055 #define g_whole_8 d12
2056 #define b_whole_8 d13
2058 #define r_whole_8 d14
2070 #define block_span q5
2078 #define d128_0x7 q13
2082 #define dither_offsets q14
2083 #define draw_mask q15
2085 #define dither_offsets_low d28
2088 #define test_mask q10
2091 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2092 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2093 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2095 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2096 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2097 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2099 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2101 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2104 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2107 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2108 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
2109 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2111 vld1.u32 { rg_dx }, [rg_dx_ptr, :64]; \
2113 cmp num_spans, #0; \
2116 stmdb sp!, { r4 - r11, r14 }; \
2117 vshl.u32 rg_dx4, rg_dx, #2; \
2119 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
2120 vshl.u32 rg_dx8, rg_dx, #3; \
2122 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2124 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
2125 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2127 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2128 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2130 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2131 vmov.u8 d64_1, #1; \
2133 vmov.u8 d128_4, #4; \
2134 vmov.u8 d64_128, #128; \
2136 vmov.u8 d128_0x7, #0x7; \
2139 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
2140 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2142 ldrh y, [span_edge_data, #edge_data_y_offset]; \
2143 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
2145 cmp span_num_blocks, #0; \
2148 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
2149 add num_blocks, span_num_blocks, num_blocks; \
2151 cmp num_blocks, #MAX_BLOCKS; \
2155 ldr b, [span_b_offset]; \
2156 add fb_ptr, fb_ptr, y, lsl #11; \
2158 vdup.u32 v_left_x, left_x; \
2161 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
2162 add fb_ptr, fb_ptr, left_x, lsl #1; \
2164 mla b, b_dx, left_x, b; \
2165 and dither_shift, left_x, #0x03; \
2167 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
2168 vshr.u32 rg_dx, rg_dx4, #2; \
2170 mov dither_shift, dither_shift, lsl #3; \
2171 vmla.u32 rg, rg_dx, v_left_x; \
2174 subs span_num_blocks, span_num_blocks, #1; \
2176 mov dither_row, dither_row, ror dither_shift; \
2177 mov b_dx4, b_dx, lsl #2; \
2179 vdup.u32 dither_offsets, dither_row; \
2180 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2182 vdup.u32 b_block, b; \
2183 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2185 mov b_dx8, b_dx, lsl #3; \
2186 vdup.u32 r_block, rg[0]; \
2187 vdup.u32 g_block, rg[1]; \
2189 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2191 vadd.u32 r_block, r_block, block_span; \
2192 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2194 vadd.u32 g_block, g_block, block_span; \
2195 vld1.u32 { block_span }, [block_span_ptr, :128]; \
2197 vadd.u32 b_block, b_block, block_span; \
2198 add block_ptr_b, block_ptr_a, #16; \
2200 vshrn.u32 r_whole_low, r_block, #16; \
2201 vshrn.u32 g_whole_low, g_block, #16; \
2202 vshrn.u32 b_whole_low, b_block, #16; \
2203 vdup.u32 dx4, rg_dx4[0]; \
2205 vaddhn.u32 r_whole_high, r_block, dx4; \
2206 vdup.u32 dx4, rg_dx4[1]; \
2208 vaddhn.u32 g_whole_high, g_block, dx4; \
2209 vdup.u32 dx4, b_dx4; \
2211 vaddhn.u32 b_whole_high, b_block, dx4; \
2212 vdup.u32 dx8, rg_dx8[0]; \
2214 vadd.u32 r_block, r_block, dx8; \
2215 vdup.u32 dx8, rg_dx8[1]; \
2217 vadd.u32 g_block, g_block, dx8; \
2218 vdup.u32 dx8, b_dx8; \
2220 vadd.u32 b_block, b_block, dx8; \
2222 vmovn.u16 r_whole_8, r_whole; \
2223 vmovn.u16 g_whole_8, g_whole; \
2224 vmovn.u16 b_whole_8, b_whole; \
2227 veor.u32 draw_mask, draw_mask, draw_mask; \
2230 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2231 vshrn.u32 r_whole_low, r_block, #16; \
2233 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2234 vshrn.u32 g_whole_low, g_block, #16; \
2236 vshrn.u32 b_whole_low, b_block, #16; \
2237 str fb_ptr, [block_ptr_a, #44]; \
2239 vdup.u32 dx4, rg_dx4[0]; \
2240 vshr.u8 r_whole_8, r_whole_8, #3; \
2241 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2243 vaddhn.u32 r_whole_high, r_block, dx4; \
2244 vdup.u32 dx4, rg_dx4[1]; \
2246 vaddhn.u32 g_whole_high, g_block, dx4; \
2247 vdup.u32 dx4, b_dx4; \
2249 vaddhn.u32 b_whole_high, b_block, dx4; \
2250 vdup.u32 dx8, rg_dx8[0]; \
2252 vmull.u8 pixels, r_whole_8, d64_1; \
2253 vmlal.u8 pixels, g_whole_8, d64_4; \
2254 vmlal.u8 pixels, b_whole_8, d64_128; \
2256 vadd.u32 r_block, r_block, dx8; \
2257 vdup.u32 dx8, rg_dx8[1]; \
2259 vadd.u32 g_block, g_block, dx8; \
2260 vdup.u32 dx8, b_dx8; \
2262 vadd.u32 b_block, b_block, dx8; \
2263 add fb_ptr, fb_ptr, #16; \
2265 vmovn.u16 r_whole_8, r_whole; \
2266 vmovn.u16 g_whole_8, g_whole; \
2267 vmovn.u16 b_whole_8, b_whole; \
2269 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_64; \
2270 vst1.u32 { pixels }, [block_ptr_b, :128], c_64; \
2274 subs span_num_blocks, span_num_blocks, #1; \
2278 str fb_ptr, [block_ptr_a, #44]; \
2279 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2281 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
2282 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2284 vshr.u8 r_whole_8, r_whole_8, #3; \
2285 vdup.u8 draw_mask, right_mask; \
2287 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2288 vld1.u32 { test_mask }, [psx_gpu, :128]; \
2290 vtst.u16 draw_mask, draw_mask, test_mask; \
2292 vmull.u8 pixels, r_whole_8, d64_1; \
2293 vmlal.u8 pixels, g_whole_8, d64_4; \
2294 vmlal.u8 pixels, b_whole_8, d64_128; \
2296 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_64; \
2297 vst1.u32 { pixels }, [block_ptr_b, :128], c_64; \
2300 add span_uvrg_offset, span_uvrg_offset, #16; \
2301 add span_b_offset, span_b_offset, #4; \
2303 add span_edge_data, span_edge_data, #8; \
2304 subs num_spans, num_spans, #1; \
2306 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
2309 ldmia sp!, { r4 - r11, pc }; \
2312 /* TODO: Load from psx_gpu instead of saving/restoring these */\
2315 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
2316 bl flush_render_block_buffer; \
2317 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
2321 vmov.u8 d64_1, #1; \
2322 vmov.u8 d128_4, #4; \
2323 vmov.u8 d64_128, #128; \
2324 vmov.u8 d128_0x7, #0x7; \
2326 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2328 mov num_blocks, span_num_blocks; \
2329 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2333 setup_blocks_shaded_untextured_indirect_builder(undithered)
2334 setup_blocks_shaded_untextured_indirect_builder(dithered)
2339 #define mask_msb_ptr r14
2341 #define draw_mask q0
2342 #define pixels_low d16
2343 #define pixels_high d17
2347 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2350 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2351 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
2352 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2354 vld1.u32 { rg_dx }, [rg_dx_ptr, :64]; \
2356 cmp num_spans, #0; \
2359 stmdb sp!, { r4 - r11, r14 }; \
2360 vshl.u32 rg_dx4, rg_dx, #2; \
2362 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
2363 vshl.u32 rg_dx8, rg_dx, #3; \
2365 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2366 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2368 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2369 vmov.u8 d64_1, #1; \
2371 vmov.u8 d128_4, #4; \
2372 vmov.u8 d64_128, #128; \
2374 vmov.u8 d128_0x7, #0x7; \
2375 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2376 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
2379 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
2380 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2382 ldrh y, [span_edge_data, #edge_data_y_offset]; \
2383 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
2385 cmp span_num_blocks, #0; \
2388 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
2389 add fb_ptr, fb_ptr, y, lsl #11; \
2391 ldr b, [span_b_offset]; \
2392 vdup.u32 v_left_x, left_x; \
2395 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
2396 add fb_ptr, fb_ptr, left_x, lsl #1; \
2398 mla b, b_dx, left_x, b; \
2399 and dither_shift, left_x, #0x03; \
2401 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
2402 vshr.u32 rg_dx, rg_dx4, #2; \
2404 mov dither_shift, dither_shift, lsl #3; \
2405 vmla.u32 rg, rg_dx, v_left_x; \
2407 subs span_num_blocks, span_num_blocks, #1; \
2409 mov dither_row, dither_row, ror dither_shift; \
2410 mov b_dx4, b_dx, lsl #2; \
2412 vdup.u32 dither_offsets, dither_row; \
2413 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2415 vdup.u32 b_block, b; \
2416 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2418 mov b_dx8, b_dx, lsl #3; \
2419 vdup.u32 r_block, rg[0]; \
2420 vdup.u32 g_block, rg[1]; \
2422 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2424 vadd.u32 r_block, r_block, block_span; \
2425 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2427 vadd.u32 g_block, g_block, block_span; \
2428 vld1.u32 { block_span }, [block_span_ptr, :128]; \
2430 vadd.u32 b_block, b_block, block_span; \
2431 add block_ptr_b, block_ptr_a, #16; \
2433 vshrn.u32 r_whole_low, r_block, #16; \
2434 vshrn.u32 g_whole_low, g_block, #16; \
2435 vshrn.u32 b_whole_low, b_block, #16; \
2436 vdup.u32 dx4, rg_dx4[0]; \
2438 vaddhn.u32 r_whole_high, r_block, dx4; \
2439 vdup.u32 dx4, rg_dx4[1]; \
2441 vaddhn.u32 g_whole_high, g_block, dx4; \
2442 vdup.u32 dx4, b_dx4; \
2444 vaddhn.u32 b_whole_high, b_block, dx4; \
2445 vdup.u32 dx8, rg_dx8[0]; \
2447 vadd.u32 r_block, r_block, dx8; \
2448 vdup.u32 dx8, rg_dx8[1]; \
2450 vadd.u32 g_block, g_block, dx8; \
2451 vdup.u32 dx8, b_dx8; \
2453 vadd.u32 b_block, b_block, dx8; \
2455 vmovn.u16 r_whole_8, r_whole; \
2456 vmovn.u16 g_whole_8, g_whole; \
2457 vmovn.u16 b_whole_8, b_whole; \
2462 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2463 vshrn.u32 r_whole_low, r_block, #16; \
2465 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2466 vshrn.u32 g_whole_low, g_block, #16; \
2468 vshrn.u32 b_whole_low, b_block, #16; \
2470 vdup.u32 dx4, rg_dx4[0]; \
2471 vshr.u8 r_whole_8, r_whole_8, #3; \
2472 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2474 vaddhn.u32 r_whole_high, r_block, dx4; \
2475 vdup.u32 dx4, rg_dx4[1]; \
2477 vmov pixels, msb_mask; \
2478 vaddhn.u32 g_whole_high, g_block, dx4; \
2479 vdup.u32 dx4, b_dx4; \
2481 vaddhn.u32 b_whole_high, b_block, dx4; \
2482 vdup.u32 dx8, rg_dx8[0]; \
2484 vmlal.u8 pixels, r_whole_8, d64_1; \
2485 vmlal.u8 pixels, g_whole_8, d64_4; \
2486 vmlal.u8 pixels, b_whole_8, d64_128; \
2488 vadd.u32 r_block, r_block, dx8; \
2489 vdup.u32 dx8, rg_dx8[1]; \
2491 vadd.u32 g_block, g_block, dx8; \
2492 vdup.u32 dx8, b_dx8; \
2494 vadd.u32 b_block, b_block, dx8; \
2496 vmovn.u16 r_whole_8, r_whole; \
2497 vmovn.u16 g_whole_8, g_whole; \
2498 vmovn.u16 b_whole_8, b_whole; \
2500 vst1.u32 { pixels }, [fb_ptr]!; \
2501 subs span_num_blocks, span_num_blocks, #1; \
2505 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2507 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
2508 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2510 vshr.u8 r_whole_8, r_whole_8, #3; \
2511 rbit right_mask, right_mask; \
2512 vmov pixels, msb_mask; \
2513 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2514 clz right_mask, right_mask; \
2516 vmlal.u8 pixels, r_whole_8, d64_1; \
2517 vmlal.u8 pixels, g_whole_8, d64_4; \
2518 vmlal.u8 pixels, b_whole_8, d64_128; \
2520 JT_OP_REL(100f, right_mask, temp); \
2521 JT_OP(ldr pc, [pc, right_mask, lsl #2]); \
2525 .word JTE(100b, 4f); \
2526 .word JTE(100b, 5f); \
2527 .word JTE(100b, 6f); \
2528 .word JTE(100b, 7f); \
2529 .word JTE(100b, 8f); \
2530 .word JTE(100b, 9f); \
2531 .word JTE(100b, 10f); \
2532 .word JTE(100b, 11f); \
2535 vst1.u16 { pixels_low[0] }, [fb_ptr]; \
2539 vst1.u32 { pixels_low[0] }, [fb_ptr]; \
2543 vst1.u32 { pixels_low[0] }, [fb_ptr]!; \
2544 vst1.u16 { pixels_low[2] }, [fb_ptr]; \
2548 vst1.u32 { pixels_low }, [fb_ptr]; \
2552 vst1.u32 { pixels_low }, [fb_ptr]!; \
2553 vst1.u16 { pixels_high[0] }, [fb_ptr]; \
2557 vst1.u32 { pixels_low }, [fb_ptr]!; \
2558 vst1.u32 { pixels_high[0] }, [fb_ptr]!; \
2562 vst1.u32 { pixels_low }, [fb_ptr]!; \
2563 vst1.u32 { pixels_high[0] }, [fb_ptr]!; \
2564 vst1.u16 { pixels_high[2] }, [fb_ptr]; \
2568 vst1.u32 { pixels }, [fb_ptr]; \
2572 add span_uvrg_offset, span_uvrg_offset, #16; \
2573 add span_b_offset, span_b_offset, #4; \
2575 add span_edge_data, span_edge_data, #8; \
2576 subs num_spans, num_spans, #1; \
2580 ldmia sp!, { r4 - r11, pc } \
2582 setup_blocks_shaded_untextured_direct_builder(undithered)
2583 setup_blocks_shaded_untextured_direct_builder(dithered)
2592 #define block_ptr r1
2593 #define num_blocks r2
2606 #define texture_ptr r11
2620 #define pixels_d r10
2624 #define clut_ptr r12
2625 #define current_texture_mask r5
2626 #define dirty_textures_mask r6
2630 #define clut_low_a d2
2631 #define clut_low_b d3
2632 #define clut_high_a d4
2633 #define clut_high_b d5
2638 #define texels_low d6
2639 #define texels_high d7
2643 function(texture_blocks_untextured)
2649 function(texture_blocks_4bpp)
2650 stmdb sp!, { r3 - r11, r14 }
2651 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2653 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2654 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2656 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
2657 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]
2659 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
2660 vuzp.u8 clut_a, clut_b
2662 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
2663 tst dirty_textures_mask, current_texture_mask
2669 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2671 uxtah uv_0, texture_ptr, uv_01
2672 uxtah uv_1, texture_ptr, uv_01, ror #16
2674 uxtah uv_2, texture_ptr, uv_23
2675 uxtah uv_3, texture_ptr, uv_23, ror #16
2677 uxtah uv_4, texture_ptr, uv_45
2678 ldrb pixel_0, [uv_0]
2680 uxtah uv_5, texture_ptr, uv_45, ror #16
2681 ldrb pixel_1, [uv_1]
2683 uxtah uv_6, texture_ptr, uv_67
2684 ldrb pixel_2, [uv_2]
2686 uxtah uv_7, texture_ptr, uv_67, ror #16
2687 ldrb pixel_3, [uv_3]
2689 ldrb pixel_4, [uv_4]
2690 subs num_blocks, num_blocks, #1
2692 ldrb pixel_5, [uv_5]
2693 orr pixels_a, pixel_0, pixel_1, lsl #8
2695 ldrb pixel_6, [uv_6]
2696 orr pixels_b, pixel_4, pixel_5, lsl #8
2698 ldrb pixel_7, [uv_7]
2699 orr pixels_a, pixels_a, pixel_2, lsl #16
2701 orr pixels_b, pixels_b, pixel_6, lsl #16
2702 orr pixels_a, pixels_a, pixel_3, lsl #24
2704 orr pixels_b, pixels_b, pixel_7, lsl #24
2705 vmov texels, pixels_a, pixels_b
2707 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2708 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2710 vst2.u8 { texels_low, texels_high }, [block_ptr, :128], c_64
2713 ldmia sp!, { r3 - r11, pc }
2716 stmdb sp!, { r1 - r2 }
2717 bl update_texture_4bpp_cache
2720 ldmia sp!, { r1 - r2 }
2726 function(texture_blocks_8bpp)
2727 stmdb sp!, { r3 - r11, r14 }
2728 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2730 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2731 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2733 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
2734 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
2736 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset]
2737 tst dirty_textures_mask, current_texture_mask
2743 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2745 uxtah uv_0, texture_ptr, uv_01
2746 uxtah uv_1, texture_ptr, uv_01, ror #16
2748 uxtah uv_2, texture_ptr, uv_23
2749 uxtah uv_3, texture_ptr, uv_23, ror #16
2751 uxtah uv_4, texture_ptr, uv_45
2752 ldrb pixel_0, [uv_0]
2754 uxtah uv_5, texture_ptr, uv_45, ror #16
2755 ldrb pixel_1, [uv_1]
2757 uxtah uv_6, texture_ptr, uv_67
2758 ldrb pixel_2, [uv_2]
2760 uxtah uv_7, texture_ptr, uv_67, ror #16
2761 ldrb pixel_3, [uv_3]
2763 ldrb pixel_4, [uv_4]
2764 add pixel_0, pixel_0, pixel_0
2766 ldrb pixel_5, [uv_5]
2767 add pixel_1, pixel_1, pixel_1
2769 ldrb pixel_6, [uv_6]
2770 add pixel_2, pixel_2, pixel_2
2772 ldrb pixel_7, [uv_7]
2773 add pixel_3, pixel_3, pixel_3
2775 ldrh pixel_0, [clut_ptr, pixel_0]
2776 add pixel_4, pixel_4, pixel_4
2778 ldrh pixel_1, [clut_ptr, pixel_1]
2779 add pixel_5, pixel_5, pixel_5
2781 ldrh pixel_2, [clut_ptr, pixel_2]
2782 add pixel_6, pixel_6, pixel_6
2784 ldrh pixel_3, [clut_ptr, pixel_3]
2785 add pixel_7, pixel_7, pixel_7
2787 ldrh pixel_4, [clut_ptr, pixel_4]
2788 orr pixels_a, pixel_0, pixel_1, lsl #16
2790 ldrh pixel_5, [clut_ptr, pixel_5]
2791 orr pixels_c, pixel_2, pixel_3, lsl #16
2793 ldrh pixel_6, [clut_ptr, pixel_6]
2794 subs num_blocks, num_blocks, #1
2796 ldrh pixel_7, [clut_ptr, pixel_7]
2797 orr pixels_b, pixel_4, pixel_5, lsl #16
2799 orr pixels_d, pixel_6, pixel_7, lsl #16
2800 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2802 add block_ptr, block_ptr, #64
2805 ldmia sp!, { r3 - r11, pc }
2808 stmdb sp!, { r1 - r2, EXTRA_UNSAVED_REGS r12 }
2810 bl update_texture_8bpp_cache
2812 ldmia sp!, { r1 - r2, EXTRA_UNSAVED_REGS r12 }
2842 #define block_ptr r1
2843 #define num_blocks r2
2887 #define texture_ptr r12
2892 function(texture_blocks_16bpp)
2893 stmdb sp!, { r3 - r11, r14 }
2894 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2896 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2897 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2900 ldrh uv_0, [block_ptr]
2901 subs num_blocks, num_blocks, #1
2903 ldrh uv_1, [block_ptr, #2]
2905 and v_0, uv_0, #0xFF00
2906 and v_1, uv_1, #0xFF00
2908 and u_0, uv_0, #0xFF
2909 and u_1, uv_1, #0xFF
2911 add uv_0, u_0, v_0, lsl #2
2912 ldrh uv_2, [block_ptr, #4]
2914 add uv_1, u_1, v_1, lsl #2
2915 ldrh uv_3, [block_ptr, #6]
2917 add uv_0, uv_0, uv_0
2918 add uv_1, uv_1, uv_1
2920 and v_2, uv_2, #0xFF00
2921 and v_3, uv_3, #0xFF00
2923 and u_2, uv_2, #0xFF
2924 and u_3, uv_3, #0xFF
2926 add uv_2, u_2, v_2, lsl #2
2927 ldrh uv_4, [block_ptr, #8]
2929 add uv_3, u_3, v_3, lsl #2
2930 ldrh uv_5, [block_ptr, #10]
2932 add uv_2, uv_2, uv_2
2933 add uv_3, uv_3, uv_3
2935 and v_4, uv_4, #0xFF00
2936 and v_5, uv_5, #0xFF00
2938 and u_4, uv_4, #0xFF
2939 and u_5, uv_5, #0xFF
2941 add uv_4, u_4, v_4, lsl #2
2942 ldrh uv_6, [block_ptr, #12]
2944 add uv_5, u_5, v_5, lsl #2
2945 ldrh uv_7, [block_ptr, #14]
2947 add uv_4, uv_4, uv_4
2948 ldrh pixel_0, [texture_ptr, uv_0]
2950 add uv_5, uv_5, uv_5
2951 ldrh pixel_1, [texture_ptr, uv_1]
2953 and v_6, uv_6, #0xFF00
2954 ldrh pixel_2, [texture_ptr, uv_2]
2956 and v_7, uv_7, #0xFF00
2957 ldrh pixel_3, [texture_ptr, uv_3]
2959 and u_6, uv_6, #0xFF
2960 ldrh pixel_4, [texture_ptr, uv_4]
2962 and u_7, uv_7, #0xFF
2963 ldrh pixel_5, [texture_ptr, uv_5]
2965 add uv_6, u_6, v_6, lsl #2
2966 add uv_7, u_7, v_7, lsl #2
2968 add uv_6, uv_6, uv_6
2969 add uv_7, uv_7, uv_7
2971 orr pixels_a, pixel_0, pixel_1, lsl #16
2972 orr pixels_b, pixel_2, pixel_3, lsl #16
2974 ldrh pixel_6, [texture_ptr, uv_6]
2975 orr pixels_c, pixel_4, pixel_5, lsl #16
2977 ldrh pixel_7, [texture_ptr, uv_7]
2978 orr pixels_d, pixel_6, pixel_7, lsl #16
2980 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
2981 add block_ptr, block_ptr, #64
2985 ldmia sp!, { r3 - r11, pc }
3000 #undef msb_mask_high
3008 #define num_blocks r1
3009 #define color_ptr r2
3010 #define colors_scalar r2
3011 #define colors_scalar_compare r3
3012 #define mask_msb_ptr r2
3014 #define block_ptr_load_a r0
3015 #define block_ptr_store r3
3016 #define block_ptr_load_b r12
3021 #define draw_mask_bits_scalar r5
3023 #define d128_0x07 q0
3024 #define d128_0x1F q1
3025 #define d128_0x8000 q2
3026 #define test_mask q3
3028 #define colors_rg q5
3029 #define colors_b_dm_bits q6
3030 #define texels_rg q7
3033 #define pixels_b q10
3035 #define zero_mask q4
3036 #define draw_mask q12
3037 #define msb_mask q13
3039 #define fb_pixels q8
3041 #define pixels_gb_low q9
3043 #define colors_r d10
3044 #define colors_g d11
3045 #define colors_b d12
3046 #define draw_mask_bits d13
3047 #define texels_r d14
3048 #define texels_g d15
3049 #define pixels_r_low d16
3050 #define pixels_g_low d18
3051 #define pixels_b_low d19
3052 #define msb_mask_low d26
3053 #define msb_mask_high d27
3058 #define texels_b d31
3060 #define shade_blocks_textured_modulated_prologue_indirect() \
3062 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3064 #define shade_blocks_textured_modulated_prologue_direct() \
3065 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3066 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16] \
3069 #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \
3071 #define shade_blocks_textured_false_modulation_check_undithered(target) \
3072 ldr colors_scalar, [psx_gpu, #psx_gpu_triangle_color_offset]; \
3073 movw colors_scalar_compare, #0x8080; \
3075 movt colors_scalar_compare, #0x80; \
3076 cmp colors_scalar, colors_scalar_compare; \
3077 beq shade_blocks_textured_unmodulated_##target \
3079 #define shade_blocks_textured_false_modulation_check_dithered(target) \
3081 #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \
3082 shade_blocks_textured_false_modulation_check_##dithering(target); \
3083 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3084 vld1.u32 { colors_r[] }, [color_ptr, :32]; \
3085 vdup.u8 colors_g, colors_r[1]; \
3086 vdup.u8 colors_b, colors_r[2]; \
3087 vdup.u8 colors_r, colors_r[0] \
3090 #define shade_blocks_textured_modulated_load_dithered(target) \
3091 vld1.u32 { target }, [block_ptr_load_b, :128] \
3093 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3094 vld1.u32 { target }, [block_ptr_load_b, :128], c_32 \
3096 #define shade_blocks_textured_modulated_load_undithered(target) \
3098 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3099 add block_ptr_load_b, block_ptr_load_b, #32 \
3101 #define shade_blocks_textured_modulate_dithered(channel) \
3102 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3104 #define shade_blocks_textured_modulate_undithered(channel) \
3105 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3108 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3109 vst1.u32 { draw_mask }, [block_ptr_store, :128]! \
3111 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3112 ldr fb_ptr, [block_ptr_load_b, #(offset - 64)]; \
3113 vld1.u32 { fb_pixels }, [fb_ptr]; \
3114 vbit.u16 pixels, fb_pixels, draw_mask \
3116 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3117 vst1.u32 { pixels }, [block_ptr_store, :128], c_48 \
3119 #define shade_blocks_textured_modulated_store_pixels_direct() \
3120 vst1.u32 { pixels }, [fb_ptr] \
3123 #define shade_blocks_textured_modulated_load_rg_shaded() \
3124 vld1.u32 { colors_r, colors_g }, [block_ptr_load_b, :128], c_32 \
3126 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3127 add block_ptr_load_b, block_ptr_load_b, #32 \
3129 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3130 vld1.u32 { colors_b, draw_mask_bits }, [block_ptr_load_a, :128], c_32 \
3132 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3133 ldr draw_mask_bits_scalar, [block_ptr_load_a, #8]; \
3134 add block_ptr_load_a, block_ptr_load_a, #32 \
3136 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3137 vdup.u16 draw_mask, draw_mask_bits[0] \
3139 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3140 vdup.u16 draw_mask, draw_mask_bits_scalar \
3143 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3145 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3146 vorr.u16 pixels, pixels, msb_mask \
3149 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3152 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3153 shade_blocks_textured_modulated_prologue_##shading(dithering, target); \
3154 stmdb sp!, { r4 - r5, lr }; \
3155 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3157 vld1.u32 { test_mask }, [psx_gpu, :128]; \
3159 shade_blocks_textured_modulated_prologue_##target(); \
3161 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3164 add block_ptr_load_b, block_ptr_load_a, #16; \
3165 vmov.u8 d64_1, #1; \
3166 vmov.u8 d64_4, #4; \
3167 vmov.u8 d64_128, #128; \
3169 vld1.u32 { texels }, [block_ptr_load_a, :128], c_32; \
3170 vmov.u8 d128_0x07, #0x07; \
3172 shade_blocks_textured_modulated_load_rg_##shading(); \
3173 vmov.u8 d128_0x1F, #0x1F; \
3175 shade_blocks_textured_modulated_load_bdm_##shading(); \
3176 vmov.u16 d128_0x8000, #0x8000; \
3178 vmovn.u16 texels_r, texels; \
3179 vshrn.u16 texels_g, texels, #5; \
3181 vshrn.u16 texels_b, texels, #7; \
3182 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3184 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3185 vtst.u16 draw_mask, draw_mask, test_mask; \
3187 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3188 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3190 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3191 vshr.u8 texels_b, texels_b, #3; \
3193 shade_blocks_textured_modulate_##dithering(r); \
3194 shade_blocks_textured_modulate_##dithering(g); \
3195 shade_blocks_textured_modulate_##dithering(b); \
3197 vand.u16 pixels, texels, d128_0x8000; \
3198 vceq.u16 zero_mask, texels, #0; \
3200 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3201 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3202 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3204 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3205 vorr.u16 draw_mask, draw_mask, zero_mask; \
3206 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3207 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3209 subs num_blocks, num_blocks, #1; \
3215 vld1.u32 { texels }, [block_ptr_load_a, :128], c_32; \
3216 shade_blocks_textured_modulated_load_rg_##shading(); \
3217 vshrn.u16 texels_g, texels, #5; \
3219 shade_blocks_textured_modulated_load_bdm_##shading(); \
3220 vshrn.u16 texels_b, texels, #7; \
3222 pld [block_ptr_load_a]; \
3223 vmovn.u16 texels_r, texels; \
3224 vmlal.u8 pixels, pixels_r_low, d64_1; \
3226 vmlal.u8 pixels, pixels_g_low, d64_4; \
3227 vmlal.u8 pixels, pixels_b_low, d64_128; \
3228 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3230 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3231 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3233 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3234 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3236 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3237 vtst.u16 draw_mask, draw_mask, test_mask; \
3239 shade_blocks_textured_modulated_store_pixels_##target(); \
3240 vshr.u8 texels_b, texels_b, #3; \
3242 shade_blocks_textured_modulate_##dithering(r); \
3243 shade_blocks_textured_modulate_##dithering(g); \
3244 shade_blocks_textured_modulate_##dithering(b); \
3246 vand.u16 pixels, texels, d128_0x8000; \
3247 vceq.u16 zero_mask, texels, #0; \
3249 subs num_blocks, num_blocks, #1; \
3251 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3252 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3253 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3255 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3256 vorr.u16 draw_mask, draw_mask, zero_mask; \
3257 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3258 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3263 vmlal.u8 pixels, pixels_r_low, d64_1; \
3264 vmlal.u8 pixels, pixels_g_low, d64_4; \
3265 vmlal.u8 pixels, pixels_b_low, d64_128; \
3267 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3268 shade_blocks_textured_modulated_store_pixels_##target(); \
3270 ldmia sp!, { r4 - r5, pc } \
3273 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3274 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3275 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3276 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3278 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3279 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3280 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3281 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3299 #undef msb_mask_high
3302 #define num_blocks r1
3303 #define mask_msb_ptr r2
3304 #define color_ptr r3
3306 #define block_ptr_load r0
3307 #define draw_mask_store_ptr r3
3308 #define draw_mask_bits_ptr r12
3309 #define draw_mask_ptr r12
3310 #define pixel_store_ptr r14
3312 #define fb_ptr_cmp r4
3315 #define fb_ptr_next r14
3319 #define test_mask q0
3321 #define draw_mask q2
3322 #define zero_mask q3
3323 #define draw_mask_combined q4
3324 #define fb_pixels q5
3325 #define fb_pixels_next q6
3328 #define draw_mask_low d4
3329 #define draw_mask_high d5
3330 #define msb_mask_low d14
3331 #define msb_mask_high d15
3334 function(shade_blocks_textured_unmodulated_indirect)
3336 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3338 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3339 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3341 vld1.u32 { test_mask }, [psx_gpu, :128]
3342 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3345 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3347 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3348 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3349 [draw_mask_bits_ptr, :16], c_64
3350 vceq.u16 zero_mask, pixels, #0
3352 vtst.u16 draw_mask, draw_mask, test_mask
3353 vst1.u32 { pixels }, [pixel_store_ptr, :128], c_64
3355 subs num_blocks, num_blocks, #1
3359 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3360 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3362 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3363 [draw_mask_bits_ptr, :16], c_64
3364 vceq.u16 zero_mask, pixels, #0
3366 vtst.u16 draw_mask, draw_mask, test_mask
3367 vst1.u32 { pixels }, [pixel_store_ptr, :128], c_64
3369 vst1.u32 { draw_mask_combined }, [draw_mask_store_ptr, :128], c_64
3370 subs num_blocks, num_blocks, #1
3375 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3376 vst1.u32 { draw_mask_combined }, [draw_mask_store_ptr, :128], c_64
3383 function(shade_blocks_textured_unmodulated_direct)
3384 stmdb sp!, { r4, r14 }
3385 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3387 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3388 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3390 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
3393 vld1.u32 { test_mask }, [psx_gpu, :128]
3394 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3396 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3397 [draw_mask_bits_ptr, :16], c_64
3398 ldr fb_ptr_next, [block_ptr_load, #44]
3400 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3401 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3402 vceq.u16 zero_mask, pixels, #0
3403 vtst.u16 draw_mask, draw_mask, test_mask
3405 subs num_blocks, num_blocks, #1
3409 mov fb_ptr, fb_ptr_next
3410 ldr fb_ptr_next, [block_ptr_load, #44]
3412 vorr.u16 pixels, pixels, msb_mask
3414 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3415 vmov fb_pixels, fb_pixels_next
3417 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3418 [draw_mask_bits_ptr, :16], c_64
3419 vbif.u16 fb_pixels, pixels, draw_mask_combined
3421 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3422 pld [fb_ptr_next, #64]
3424 add fb_ptr_cmp, fb_ptr_cmp, #14
3425 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3430 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3431 vceq.u16 zero_mask, pixels, #0
3433 vst1.u16 { fb_pixels }, [fb_ptr]
3434 vtst.u16 draw_mask, draw_mask, test_mask
3437 subs num_blocks, num_blocks, #1
3441 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3442 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3444 vst1.u16 { fb_pixels_next }, [fb_ptr_next]
3446 ldmia sp!, { r4, pc }
3449 vst1.u16 { fb_pixels }, [fb_ptr]
3450 vceq.u16 zero_mask, pixels, #0
3452 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3453 vtst.u16 draw_mask, draw_mask, test_mask
3458 function(shade_blocks_unshaded_untextured_indirect)
3463 function(shade_blocks_unshaded_untextured_direct)
3464 stmdb sp!, { r4, r14 }
3465 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3467 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3468 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3470 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
3471 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3473 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3474 vld1.u16 { pixels }, [color_ptr, :128]
3477 vld1.u16 { draw_mask }, [draw_mask_ptr, :128], c_64
3479 vorr.u16 pixels, pixels, msb_mask
3480 subs num_blocks, num_blocks, #1
3482 ldr fb_ptr_next, [block_ptr_load], #64
3484 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3488 vmov fb_pixels, fb_pixels_next
3489 mov fb_ptr, fb_ptr_next
3490 ldr fb_ptr_next, [block_ptr_load], #64
3492 vbif.u16 fb_pixels, pixels, draw_mask
3493 vld1.u16 { draw_mask }, [draw_mask_ptr, :128], c_64
3495 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3496 add fb_ptr_cmp, fb_ptr_cmp, #14
3500 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3501 vst1.u16 { fb_pixels }, [fb_ptr]
3504 subs num_blocks, num_blocks, #1
3508 vbif.u16 fb_pixels_next, pixels, draw_mask
3509 vst1.u16 { fb_pixels_next }, [fb_ptr_next]
3511 ldmia sp!, { r4, pc }
3514 vst1.u16 { fb_pixels }, [fb_ptr]
3515 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3519 #undef draw_mask_ptr
3526 #define num_blocks r1
3527 #define msb_mask_ptr r2
3528 #define pixel_ptr r3
3529 #define draw_mask_ptr r0
3532 #define fb_ptr_next r14
3533 #define fb_ptr_cmp r4
3541 #undef msb_mask_high
3542 #undef draw_mask_next
3545 #undef fb_pixels_next
3548 #define draw_mask q1
3550 #define fb_pixels q3
3551 #define blend_pixels q4
3552 #define pixels_no_msb q5
3553 #define blend_mask q6
3554 #define fb_pixels_no_msb q7
3555 #define d128_0x8000 q8
3556 #define d128_0x0421 q9
3557 #define fb_pixels_next q10
3558 #define blend_pixels_next q11
3559 #define pixels_next q12
3560 #define draw_mask_next q13
3561 #define write_mask q14
3563 #define pixels_rb q5
3564 #define pixels_mg q7
3566 #define d128_0x7C1F q8
3567 #define d128_0x03E0 q9
3568 #define fb_pixels_rb q10
3569 #define fb_pixels_g q11
3570 #define fb_pixels_masked q11
3571 #define d128_0x83E0 q15
3572 #define pixels_fourth q7
3573 #define d128_0x1C07 q12
3574 #define d128_0x00E0 q13
3575 #define d128_0x80E0 q13
3577 #define msb_mask_low d0
3578 #define msb_mask_high d1
3580 #define blend_blocks_average_set_blend_mask_textured(source) \
3581 vclt.s16 blend_mask, source, #0 \
3583 #define blend_blocks_average_set_stp_bit_textured() \
3584 vorr.u16 blend_pixels, #0x8000 \
3586 #define blend_blocks_average_combine_textured(source) \
3587 vbif.u16 blend_pixels, source, blend_mask \
3589 #define blend_blocks_average_set_blend_mask_untextured(source) \
3591 #define blend_blocks_average_set_stp_bit_untextured() \
3593 #define blend_blocks_average_combine_untextured(source) \
3595 #define blend_blocks_average_mask_set_on() \
3596 vclt.s16 write_mask, fb_pixels_next, #0 \
3598 #define blend_blocks_average_mask_copy_on() \
3599 vorr.u16 draw_mask, draw_mask_next, write_mask \
3601 #define blend_blocks_average_mask_copy_b_on() \
3602 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3604 #define blend_blocks_average_mask_set_off() \
3606 #define blend_blocks_average_mask_copy_off() \
3607 vmov draw_mask, draw_mask_next \
3609 #define blend_blocks_average_mask_copy_b_off() \
3611 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3614 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3615 stmdb sp!, { r4, r14 }; \
3616 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3617 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3619 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3620 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3622 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3625 vmov.u16 d128_0x8000, #0x8000; \
3626 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
3627 ldr fb_ptr_next, [pixel_ptr, #28]; \
3629 vmov.u16 d128_0x0421, #0x0400; \
3630 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
3632 vorr.u16 d128_0x0421, #0x0021; \
3633 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3635 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3636 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3637 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3638 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3639 blend_blocks_average_mask_set_##mask_evaluate(); \
3640 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3642 subs num_blocks, num_blocks, #1; \
3646 mov fb_ptr, fb_ptr_next; \
3647 ldr fb_ptr_next, [pixel_ptr, #28]; \
3649 vmov pixels, pixels_next; \
3650 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
3652 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3654 blend_blocks_average_mask_copy_##mask_evaluate(); \
3655 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
3657 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3658 blend_blocks_average_set_stp_bit_##texturing(); \
3659 vmov fb_pixels, fb_pixels_next; \
3660 blend_blocks_average_combine_##texturing(pixels); \
3662 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3663 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3664 cmp fb_ptr_cmp, #28; \
3667 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3668 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3670 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3671 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3673 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3674 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3676 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3677 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3678 blend_blocks_average_mask_set_##mask_evaluate(); \
3679 vst1.u16 { fb_pixels }, [fb_ptr]; \
3682 subs num_blocks, num_blocks, #1; \
3686 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3687 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3689 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3690 blend_blocks_average_set_stp_bit_##texturing(); \
3691 blend_blocks_average_combine_##texturing(pixels_next); \
3693 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3694 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3695 vst1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3697 ldmia sp!, { r4, pc }; \
3700 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3701 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3702 vst1.u16 { fb_pixels }, [fb_ptr]; \
3704 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3705 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3706 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3707 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3708 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3709 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3713 blend_blocks_average_builder(textured, off)
3714 blend_blocks_average_builder(untextured, off)
3715 blend_blocks_average_builder(textured, on)
3716 blend_blocks_average_builder(untextured, on)
3719 #define blend_blocks_add_mask_set_on() \
3720 vclt.s16 write_mask, fb_pixels, #0 \
3722 #define blend_blocks_add_mask_copy_on() \
3723 vorr.u16 draw_mask, draw_mask, write_mask \
3725 #define blend_blocks_add_mask_set_off() \
3727 #define blend_blocks_add_mask_copy_off() \
3730 #define blend_blocks_add_textured_builder(mask_evaluate) \
3733 function(blend_blocks_textured_add_##mask_evaluate) \
3734 stmdb sp!, { r4, r14 }; \
3735 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3736 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3738 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3739 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3741 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3744 vmov.u16 d128_0x7C1F, #0x7C00; \
3745 vmov.u16 d128_0x03E0, #0x0300; \
3746 vmov.u16 d128_0x83E0, #0x8000; \
3747 vorr.u16 d128_0x03E0, #0x00E0; \
3748 vorr.u16 d128_0x7C1F, #0x001F; \
3749 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3751 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3752 ldr fb_ptr_next, [pixel_ptr, #28]; \
3753 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3754 vclt.s16 blend_mask, pixels, #0; \
3755 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3756 blend_blocks_add_mask_set_##mask_evaluate(); \
3757 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3759 blend_blocks_add_mask_copy_##mask_evaluate(); \
3760 vorr.u16 pixels, pixels, msb_mask; \
3761 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3762 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3763 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3764 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3765 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3766 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3767 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3768 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3770 subs num_blocks, num_blocks, #1; \
3774 mov fb_ptr, fb_ptr_next; \
3776 ldr fb_ptr_next, [pixel_ptr, #28]; \
3778 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3779 vclt.s16 blend_mask, pixels, #0; \
3781 vorr.u16 pixels, pixels, msb_mask; \
3782 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3783 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3785 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3786 pld [fb_ptr_next, #64]; \
3788 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3789 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3791 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3792 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3794 cmp fb_ptr_cmp, #28; \
3797 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3798 blend_blocks_add_mask_set_##mask_evaluate(); \
3799 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3800 blend_blocks_add_mask_copy_##mask_evaluate(); \
3801 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3802 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3803 vst1.u16 { blend_pixels }, [fb_ptr]; \
3806 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3807 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3808 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3809 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3810 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3812 subs num_blocks, num_blocks, #1; \
3816 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3817 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3818 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
3820 ldmia sp!, { r4, pc }; \
3823 vst1.u16 { blend_pixels }, [fb_ptr]; \
3824 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3826 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3827 blend_blocks_add_mask_set_##mask_evaluate(); \
3828 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3829 blend_blocks_add_mask_copy_##mask_evaluate(); \
3830 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3834 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3837 function(blend_blocks_untextured_add_##mask_evaluate) \
3838 stmdb sp!, { r4, r14 }; \
3839 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3840 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3842 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3843 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3845 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3848 vmov.u16 d128_0x7C1F, #0x7C00; \
3849 vmov.u16 d128_0x03E0, #0x0300; \
3850 vorr.u16 d128_0x7C1F, #0x001F; \
3851 vorr.u16 d128_0x03E0, #0x00E0; \
3853 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3854 ldr fb_ptr_next, [pixel_ptr, #28]; \
3855 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3856 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3857 blend_blocks_add_mask_set_##mask_evaluate(); \
3858 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3860 blend_blocks_add_mask_copy_##mask_evaluate(); \
3861 vand.u16 pixels_g, pixels, d128_0x03E0; \
3862 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
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 mov fb_ptr, fb_ptr_next; \
3875 ldr fb_ptr_next, [pixel_ptr, #28]; \
3877 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3879 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3880 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3881 vand.u16 pixels_g, pixels, d128_0x03E0; \
3883 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3884 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3886 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3887 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3888 cmp fb_ptr_cmp, #28; \
3891 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3892 blend_blocks_add_mask_set_##mask_evaluate(); \
3893 blend_blocks_add_mask_copy_##mask_evaluate(); \
3894 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3895 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3896 vst1.u16 { blend_pixels }, [fb_ptr]; \
3899 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3900 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3901 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3902 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3903 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3905 subs num_blocks, num_blocks, #1; \
3909 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3910 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3911 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3912 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
3914 ldmia sp!, { r4, pc }; \
3917 vst1.u16 { blend_pixels }, [fb_ptr]; \
3918 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3920 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3921 blend_blocks_add_mask_set_##mask_evaluate(); \
3922 blend_blocks_add_mask_copy_##mask_evaluate(); \
3923 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3927 blend_blocks_add_textured_builder(off)
3928 blend_blocks_add_textured_builder(on)
3929 blend_blocks_add_untextured_builder(off)
3930 blend_blocks_add_untextured_builder(on)
3932 #define blend_blocks_subtract_set_blend_mask_textured() \
3933 vclt.s16 blend_mask, pixels_next, #0 \
3935 #define blend_blocks_subtract_combine_textured() \
3936 vbif.u16 blend_pixels, pixels, blend_mask \
3938 #define blend_blocks_subtract_set_stp_textured() \
3939 vorr.u16 blend_pixels, #0x8000 \
3941 #define blend_blocks_subtract_msb_mask_textured() \
3942 vorr.u16 pixels, pixels_next, msb_mask \
3944 #define blend_blocks_subtract_set_blend_mask_untextured() \
3946 #define blend_blocks_subtract_combine_untextured() \
3948 #define blend_blocks_subtract_set_stp_untextured() \
3949 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3951 #define blend_blocks_subtract_msb_mask_untextured() \
3954 #define blend_blocks_subtract_mask_set_on() \
3955 vclt.s16 write_mask, fb_pixels, #0 \
3957 #define blend_blocks_subtract_mask_copy_on() \
3958 vorr.u16 draw_mask, draw_mask_next, write_mask \
3960 #define blend_blocks_subtract_mask_set_off() \
3962 #define blend_blocks_subtract_mask_copy_off() \
3963 vmov draw_mask, draw_mask_next \
3966 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
3969 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
3970 stmdb sp!, { r4, r14 }; \
3971 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3972 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3974 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3975 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3977 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3980 vmov.u16 d128_0x7C1F, #0x7C00; \
3981 vmov.u16 d128_0x03E0, #0x0300; \
3982 vorr.u16 d128_0x7C1F, #0x001F; \
3983 vorr.u16 d128_0x03E0, #0x00E0; \
3985 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
3986 ldr fb_ptr_next, [pixel_ptr, #28]; \
3987 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
3988 blend_blocks_subtract_set_blend_mask_##texturing(); \
3989 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3990 blend_blocks_subtract_mask_set_##mask_evaluate(); \
3991 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
3993 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
3994 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3995 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3996 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3997 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3999 subs num_blocks, num_blocks, #1; \
4003 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4004 mov fb_ptr, fb_ptr_next; \
4005 ldr fb_ptr_next, [pixel_ptr, #28]; \
4007 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
4008 blend_blocks_subtract_msb_mask_##texturing(); \
4010 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
4011 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4012 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4013 blend_blocks_subtract_set_stp_##texturing(); \
4014 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4015 blend_blocks_subtract_combine_##texturing(); \
4016 blend_blocks_subtract_set_blend_mask_##texturing(); \
4017 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4019 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4020 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4021 cmp fb_ptr_cmp, #28; \
4024 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4025 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4026 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4027 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4028 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4029 vst1.u16 { blend_pixels }, [fb_ptr]; \
4030 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4033 subs num_blocks, num_blocks, #1; \
4037 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4039 blend_blocks_subtract_msb_mask_##texturing(); \
4040 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4041 blend_blocks_subtract_set_stp_##texturing(); \
4042 blend_blocks_subtract_combine_##texturing(); \
4043 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4044 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4046 ldmia sp!, { r4, pc }; \
4049 vst1.u16 { blend_pixels }, [fb_ptr]; \
4050 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4051 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4052 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4053 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4054 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4055 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4059 blend_blocks_subtract_builder(textured, off)
4060 blend_blocks_subtract_builder(textured, on)
4061 blend_blocks_subtract_builder(untextured, off)
4062 blend_blocks_subtract_builder(untextured, on)
4065 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4068 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4069 stmdb sp!, { r4, r14 }; \
4070 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4071 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
4073 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4074 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
4076 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4079 vmov.u16 d128_0x7C1F, #0x7C00; \
4080 vmov.u16 d128_0x03E0, #0x0300; \
4081 vmov.u16 d128_0x1C07, #0x1C00; \
4082 vmov.u16 d128_0x00E0, #0x00E0; \
4083 vorr.u16 d128_0x7C1F, #0x001F; \
4084 vorr.u16 d128_0x03E0, #0x00E0; \
4085 vorr.u16 d128_0x1C07, #0x0007; \
4087 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4088 ldr fb_ptr_next, [pixel_ptr, #28]; \
4089 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4090 vclt.s16 blend_mask, pixels, #0; \
4091 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4092 blend_blocks_add_mask_set_##mask_evaluate(); \
4093 vshr.s16 pixels_fourth, pixels, #2; \
4094 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4096 blend_blocks_add_mask_copy_##mask_evaluate(); \
4097 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4098 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4099 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4100 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4101 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4102 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4103 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4105 subs num_blocks, num_blocks, #1; \
4109 mov fb_ptr, fb_ptr_next; \
4110 ldr fb_ptr_next, [pixel_ptr, #28]; \
4112 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4113 vorr.u16 blend_pixels, #0x8000; /* stp */ \
4114 vbif.u16 blend_pixels, pixels, blend_mask; \
4116 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4117 vclt.s16 blend_mask, pixels, #0; \
4118 vshr.s16 pixels_fourth, pixels, #2; \
4119 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4120 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4122 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4123 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4125 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4126 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4127 cmp fb_ptr_cmp, #28; \
4130 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4131 blend_blocks_add_mask_set_##mask_evaluate(); \
4132 blend_blocks_add_mask_copy_##mask_evaluate(); \
4133 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4134 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4135 vst1.u16 { blend_pixels }, [fb_ptr]; \
4138 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4139 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4140 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4141 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4142 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4144 subs num_blocks, num_blocks, #1; \
4148 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4149 vorr.u16 blend_pixels, #0x8000; /* stp */ \
4150 vbif.u16 blend_pixels, pixels, blend_mask; \
4151 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4152 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4153 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4155 ldmia sp!, { r4, pc }; \
4158 vst1.u16 { blend_pixels }, [fb_ptr]; \
4159 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4161 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4162 blend_blocks_add_mask_set_##mask_evaluate(); \
4163 blend_blocks_add_mask_copy_##mask_evaluate(); \
4164 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4169 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4172 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4173 stmdb sp!, { r4, r14 }; \
4174 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4175 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
4177 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4178 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
4180 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4183 vmov.u16 d128_0x7C1F, #0x7C00; \
4184 vmov.u16 d128_0x03E0, #0x0300; \
4185 vmov.u16 d128_0x1C07, #0x1C00; \
4186 vmov.u16 d128_0x00E0, #0x00E0; \
4187 vorr.u16 d128_0x7C1F, #0x001F; \
4188 vorr.u16 d128_0x03E0, #0x00E0; \
4189 vorr.u16 d128_0x1C07, #0x0007; \
4191 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4192 ldr fb_ptr_next, [pixel_ptr, #28]; \
4193 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4194 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4195 blend_blocks_add_mask_set_##mask_evaluate(); \
4196 vshr.s16 pixels_fourth, pixels, #2; \
4197 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4199 blend_blocks_add_mask_copy_##mask_evaluate(); \
4200 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4201 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4202 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4203 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4204 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4205 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4206 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4208 subs num_blocks, num_blocks, #1; \
4212 mov fb_ptr, fb_ptr_next; \
4213 ldr fb_ptr_next, [pixel_ptr, #28]; \
4215 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4217 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4218 vshr.s16 pixels_fourth, pixels, #2; \
4219 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4220 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4222 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4223 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4225 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4226 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4227 cmp fb_ptr_cmp, #28; \
4230 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4231 blend_blocks_add_mask_set_##mask_evaluate(); \
4232 blend_blocks_add_mask_copy_##mask_evaluate(); \
4233 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4234 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4235 vst1.u16 { blend_pixels }, [fb_ptr]; \
4238 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4239 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4240 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4241 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4242 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4244 subs num_blocks, num_blocks, #1; \
4248 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4249 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4250 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4251 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4253 ldmia sp!, { r4, pc }; \
4256 vst1.u16 { blend_pixels }, [fb_ptr]; \
4257 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4259 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4260 blend_blocks_add_mask_set_##mask_evaluate(); \
4261 blend_blocks_add_mask_copy_##mask_evaluate(); \
4262 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4266 blend_blocks_add_fourth_textured_builder(off)
4267 blend_blocks_add_fourth_textured_builder(on)
4268 blend_blocks_add_fourth_untextured_builder(off)
4269 blend_blocks_add_fourth_untextured_builder(on)
4271 // TODO: Optimize this more. Need a scene that actually uses it for
4276 function(blend_blocks_textured_unblended_on)
4277 stmdb sp!, { r4, r14 }
4278 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4279 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
4281 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4282 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
4284 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4287 ldr fb_ptr, [pixel_ptr, #28]
4288 vld1.u16 { fb_pixels }, [fb_ptr]
4289 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64
4290 vclt.s16 write_mask, fb_pixels, #0
4291 vld1.u32 { pixels }, [pixel_ptr, :128], c_64
4293 subs num_blocks, num_blocks, #1
4297 vorr.u16 pixels, pixels, msb_mask
4298 vorr.u16 draw_mask, draw_mask, write_mask
4299 vbif.u16 fb_pixels, pixels, draw_mask
4300 vst1.u16 { fb_pixels }, [fb_ptr]
4302 ldr fb_ptr, [pixel_ptr, #28]
4303 vld1.u16 { fb_pixels }, [fb_ptr]
4304 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64
4305 vclt.s16 write_mask, fb_pixels, #0
4306 vld1.u32 { pixels }, [pixel_ptr, :128], c_64
4308 subs num_blocks, num_blocks, #1
4312 vorr.u16 pixels, pixels, msb_mask
4313 vorr.u16 draw_mask, draw_mask, write_mask
4314 vbif.u16 fb_pixels, pixels, draw_mask
4315 vst1.u16 { fb_pixels }, [fb_ptr]
4317 ldmia sp!, { r4, pc }
4320 function(blend_blocks_textured_unblended_off)
4330 vld1.u32 { u_whole_8, v_whole_8 }, [r1, :128], r3
4350 #define num_width r12
4360 function(render_block_fill_body)
4361 vdup.u16 colors_a, color
4364 vmov colors_b, colors_a
4365 sub pitch, pitch, width, lsl #1
4367 mov num_width, width
4370 vst1.u32 { colors_a, colors_b }, [vram_ptr, :256]!
4372 subs num_width, num_width, #16
4375 add vram_ptr, vram_ptr, pitch
4376 mov num_width, width
4378 subs height, height, #1
4392 #undef dirty_textures_mask
4394 #undef current_texture_mask
4405 #define offset_u_right r10
4406 #define width_rounded r11
4407 #define height_rounded r12
4409 #define texture_offset_base r1
4410 #define tile_width r2
4411 #define tile_height r3
4412 #define num_blocks r4
4414 #define sub_tile_height r6
4416 #define texture_mask r8
4417 #define column_data r9
4418 #define texture_offset r10
4419 #define tiles_remaining r11
4420 #define fb_ptr_advance_column r12
4421 #define texture_block_ptr r14
4425 #define texture_page_ptr r3
4426 #define left_block_mask r4
4427 #define right_block_mask r5
4428 #define texture_mask_rev r10
4429 #define control_mask r11
4431 #define dirty_textures_mask r4
4433 #define current_texture_mask r6
4447 #define draw_masks_fb_ptrs q1
4449 #define draw_mask_fb_ptr_left d2
4450 #define draw_mask_fb_ptr_right d3
4452 #define draw_mask_fb_ptr_left_a d2
4453 #define draw_mask_fb_ptr_left_b d3
4454 #define draw_mask_fb_ptr_right_a d10
4455 #define draw_mask_fb_ptr_right_b d11
4456 #define draw_masks_fb_ptrs2 q5
4458 #define clut_low_a d4
4459 #define clut_low_b d5
4460 #define clut_high_a d6
4461 #define clut_high_b d7
4463 #define block_masks d8
4464 #define block_masks_shifted d9
4469 #define texels_low d12
4470 #define texels_high d13
4472 #define texels_wide_low d14
4473 #define texels_wide_high d15
4474 #define texels_wide q7
4477 setup_sprite_flush_blocks:
4480 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
4481 bl flush_render_block_buffer
4482 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
4486 add block, psx_gpu, #psx_gpu_blocks_offset
4490 setup_sprite_update_texture_4bpp_cache:
4491 stmdb sp!, { r0 - r3, r14 }
4492 bl update_texture_4bpp_cache
4493 ldmia sp!, { r0 - r3, pc }
4496 setup_sprite_update_texture_8bpp_cache:
4497 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r14 }
4498 bl update_texture_8bpp_cache
4499 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS pc }
4502 #define setup_sprite_tiled_initialize_4bpp() \
4503 ldr dirty_textures_mask, \
4504 [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]; \
4505 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]; \
4507 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]; \
4508 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]; \
4510 tst current_texture_mask, dirty_textures_mask; \
4511 vuzp.u8 clut_a, clut_b; \
4513 blne setup_sprite_update_texture_4bpp_cache \
4515 #define setup_sprite_tiled_initialize_8bpp() \
4516 ldr dirty_textures_mask, \
4517 [psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset]; \
4518 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]; \
4520 tst current_texture_mask, dirty_textures_mask; \
4521 blne setup_sprite_update_texture_8bpp_cache \
4524 #define setup_sprite_block_count_single() \
4527 #define setup_sprite_block_count_double() \
4528 sub_tile_height, lsl #1 \
4530 #define setup_sprite_tile_add_blocks(type) \
4531 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4532 cmp num_blocks, #MAX_BLOCKS; \
4534 movgt num_blocks, setup_sprite_block_count_##type(); \
4535 blgt setup_sprite_flush_blocks \
4538 #define setup_sprite_tile_full_4bpp(edge) \
4539 setup_sprite_tile_add_blocks(double); \
4542 and texture_block_ptr, texture_offset, texture_mask; \
4543 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4546 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4547 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4549 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4550 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4552 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4553 add texture_block_ptr, texture_offset, #8; \
4555 and texture_block_ptr, texture_block_ptr, texture_mask; \
4556 add block, block, #40; \
4558 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4559 add fb_ptr, fb_ptr, #16; \
4561 vst1.u32 { draw_mask_fb_ptr_left }, [block, :64]; \
4562 add block, block, #24; \
4564 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4565 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4568 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4569 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4571 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4572 add block, block, #40; \
4574 add texture_offset, texture_offset, #0x10; \
4575 add fb_ptr, fb_ptr, #(2048 - 16); \
4577 vst1.u32 { draw_mask_fb_ptr_right }, [block, :64]; \
4578 add block, block, #24; \
4580 subs sub_tile_height, sub_tile_height, #1; \
4583 add texture_offset, texture_offset, #0xF00; \
4584 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4587 #define setup_sprite_tile_half_4bpp(edge) \
4588 setup_sprite_tile_add_blocks(single); \
4591 and texture_block_ptr, texture_offset, texture_mask; \
4592 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4595 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4596 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4598 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4599 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4601 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4602 add block, block, #40; \
4604 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4605 vst1.u32 { draw_mask_fb_ptr_##edge }, [block, :64]; \
4607 add block, block, #24; \
4608 add texture_offset, texture_offset, #0x10; \
4610 add fb_ptr, fb_ptr, #2048; \
4611 subs sub_tile_height, sub_tile_height, #1; \
4615 add texture_offset, texture_offset, #0xF00; \
4616 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4619 #define setup_sprite_tile_full_8bpp(edge) \
4620 setup_sprite_tile_add_blocks(double); \
4621 add block, block, #16; \
4624 and texture_block_ptr, texture_offset, texture_mask; \
4625 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4628 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4629 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4631 add texture_block_ptr, texture_offset, #8; \
4632 vst1.u32 { texels }, [block, :64]; \
4634 and texture_block_ptr, texture_block_ptr, texture_mask; \
4635 add block, block, #24; \
4637 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4639 add fb_ptr, fb_ptr, #16; \
4640 vst1.u32 { draw_mask_fb_ptr_left }, [block, :64]; \
4642 add block, block, #40; \
4643 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4646 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4647 vst1.u32 { texels }, [block, :64]; \
4648 add block, block, #24; \
4650 add texture_offset, texture_offset, #0x10; \
4651 add fb_ptr, fb_ptr, #(2048 - 16); \
4653 vst1.u32 { draw_mask_fb_ptr_right }, [block, :64]; \
4654 add block, block, #40; \
4656 subs sub_tile_height, sub_tile_height, #1; \
4659 sub block, block, #16; \
4660 add texture_offset, texture_offset, #0xF00; \
4661 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4664 #define setup_sprite_tile_half_8bpp(edge) \
4665 setup_sprite_tile_add_blocks(single); \
4666 add block, block, #16; \
4669 and texture_block_ptr, texture_offset, texture_mask; \
4670 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4673 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4674 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4676 vst1.u32 { texels }, [block, :64]; \
4677 add block, block, #24; \
4679 vst1.u32 { draw_mask_fb_ptr_##edge }, [block, :64]; \
4680 add block, block, #40; \
4682 add texture_offset, texture_offset, #0x10; \
4683 add fb_ptr, fb_ptr, #2048; \
4685 subs sub_tile_height, sub_tile_height, #1; \
4688 sub block, block, #16; \
4689 add texture_offset, texture_offset, #0xF00; \
4690 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4693 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4694 add texture_offset, texture_offset_base, #8; \
4695 add fb_ptr, fb_ptr, #16 \
4697 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4698 mov texture_offset, texture_offset_base \
4700 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4701 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4703 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4704 mov texture_offset, texture_offset_base \
4706 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4707 sub fb_ptr, fb_ptr, #16 \
4709 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4711 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4712 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4714 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4717 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \
4719 mov sub_tile_height, column_data; \
4720 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4721 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4722 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4724 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \
4726 and sub_tile_height, column_data, #0xFF; \
4727 mov tiles_remaining, column_data, lsr #16; \
4728 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4729 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4731 subs tiles_remaining, tiles_remaining, #1; \
4735 mov sub_tile_height, #16; \
4736 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4737 subs tiles_remaining, tiles_remaining, #1; \
4741 uxtb sub_tile_height, column_data, ror #8; \
4742 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4743 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4746 #define setup_sprite_column_data_single() \
4747 mov column_data, height; \
4748 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset] \
4750 #define setup_sprite_column_data_multi() \
4751 and height_rounded, height_rounded, #0xF; \
4752 rsb column_data, offset_v, #16; \
4754 add height_rounded, height_rounded, #1; \
4755 sub tile_height, tile_height, #1; \
4757 orr column_data, column_data, tile_height, lsl #16; \
4758 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]; \
4760 orr column_data, column_data, height_rounded, lsl #8 \
4762 #define setup_sprite_setup_left_draw_mask_fb_ptr() \
4763 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4764 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4766 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column() \
4767 mov fb_ptr_advance_column, #32; \
4768 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4770 sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11; \
4771 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4773 #define setup_sprite_setup_right_draw_mask_fb_ptr() \
4774 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4775 vdup.u8 draw_mask_fb_ptr_right, block_masks[5] \
4777 #define setup_sprite_tile_column_width_single(tm, multi_height, edge_mode, \
4779 setup_sprite_##tm##_single_##multi_height##_##edge_mode##_##edge##x4mode: \
4780 setup_sprite_column_data_##multi_height(); \
4781 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4782 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4783 setup_sprite_setup_left_draw_mask_fb_ptr##x4mode(); \
4785 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, tm, x4mode); \
4786 ldmia sp!, { r4 - r11, pc } \
4788 #define setup_sprite_tiled_advance_column() \
4789 add texture_offset_base, texture_offset_base, #0x100; \
4790 tst texture_offset_base, #0xF00; \
4791 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4793 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4794 right_mode, x4mode) \
4795 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode##x4mode:\
4796 setup_sprite_column_data_##multi_height(); \
4798 setup_sprite_setup_left_draw_mask_fb_ptr_advance_column##x4mode(); \
4800 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm, x4mode);\
4802 subs tile_width, tile_width, #2; \
4803 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4807 vmov.u8 draw_masks_fb_ptrs, #0; \
4808 vmov.u8 draw_masks_fb_ptrs2, #0; \
4811 setup_sprite_tiled_advance_column(); \
4812 setup_sprite_tile_column_height_##multi_height(full, none, tm, x4mode); \
4813 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4814 subs tile_width, tile_width, #1; \
4818 setup_sprite_setup_right_draw_mask_fb_ptr##x4mode(); \
4820 setup_sprite_tiled_advance_column(); \
4821 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm, x4mode);\
4822 ldmia sp!, { r4 - r11, pc } \
4825 #define setup_sprite_offset_u_adjust() \
4827 #define setup_sprite_get_left_block_mask() \
4828 and left_block_mask, left_block_mask, #0xFF \
4830 #define setup_sprite_compare_left_block_mask() \
4831 cmp left_block_mask, #0xFF \
4833 #define setup_sprite_get_right_block_mask() \
4834 uxtb right_block_mask, right_block_mask, ror #8 \
4836 #define setup_sprite_compare_right_block_mask() \
4837 cmp right_block_mask, #0xFF \
4842 #define fb_ptr2 column_data
4844 #define setup_sprite_offset_u_adjust_4x() \
4845 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4846 lsl offset_u_right, #1; \
4848 add offset_u_right, #1 \
4850 #define setup_sprite_get_left_block_mask_4x() \
4851 sxth left_block_mask, left_block_mask \
4853 #define setup_sprite_compare_left_block_mask_4x() \
4854 cmp left_block_mask, #0xFFFFFFFF \
4856 #define setup_sprite_get_right_block_mask_4x() \
4857 sxth right_block_mask, right_block_mask, ror #16 \
4859 #define setup_sprite_compare_right_block_mask_4x() \
4860 cmp right_block_mask, #0xFFFFFFFF \
4863 #define widen_texels_16bpp(texels_) \
4864 vmov texels_wide_low, texels_; \
4865 vmov texels_wide_high, texels_; \
4866 vzip.16 texels_wide_low, texels_wide_high \
4868 #define widen_texels_8bpp(texels_) \
4869 vmov texels_wide_low, texels_; \
4870 vmov texels_wide_high, texels_; \
4871 vzip.8 texels_wide_low, texels_wide_high \
4873 #define write_block_16bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4874 vst1.u32 { texels_ }, [block_, :128]; \
4875 add block_, block_, #40; \
4877 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4878 vst1.u32 { draw_mask_fb_ptr_ }, [block_, :64]; \
4879 add block_, block_, #24 \
4881 /* assumes 16-byte offset already added to block_ */
4882 #define write_block_8bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4883 vst1.u32 { texels_ }, [block_, :64]; \
4884 add block_, block_, #24; \
4886 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4887 vst1.u32 { draw_mask_fb_ptr_ }, [block_, :64]; \
4888 add block_, block_, #40 \
4890 #define do_texture_block_16bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4891 draw_mask_fb_ptr_b_) \
4892 widen_texels_16bpp(texels_low); \
4893 add fb_ptr_tmp, fb_ptr, #1024*2; \
4895 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr); \
4897 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4898 widen_texels_16bpp(texels_high); \
4900 add fb_ptr_tmp, fb_ptr, #8*2; \
4901 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4903 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4904 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4906 #define do_texture_block_8bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4907 draw_mask_fb_ptr_b_) \
4908 widen_texels_8bpp(texels); \
4909 add fb_ptr_tmp, fb_ptr, #1024*2; \
4911 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr); \
4912 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4914 add fb_ptr_tmp, fb_ptr, #8*2; \
4915 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4917 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4918 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4921 #define setup_sprite_tiled_initialize_4bpp_4x() \
4922 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]; \
4923 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]; \
4925 vuzp.u8 clut_a, clut_b \
4927 #define setup_sprite_tiled_initialize_8bpp_4x() \
4930 #define setup_sprite_block_count_single_4x() \
4931 sub_tile_height, lsl #2 \
4933 #define setup_sprite_block_count_double_4x() \
4934 sub_tile_height, lsl #(1+2) \
4936 #define setup_sprite_tile_full_4bpp_4x(edge) \
4937 setup_sprite_tile_add_blocks(double_4x); \
4938 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4941 and texture_block_ptr, texture_offset, texture_mask; \
4944 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4945 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4947 add texture_block_ptr, texture_offset, #8; \
4948 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4950 and texture_block_ptr, texture_block_ptr, texture_mask; \
4951 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4953 vzip.8 texels_low, texels_high; \
4954 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
4955 draw_mask_fb_ptr_left_b); \
4957 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4958 pld [fb_ptr, #2048]; \
4960 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4961 add fb_ptr, fb_ptr, #16*2; \
4963 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4964 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4966 vzip.8 texels_low, texels_high; \
4967 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
4968 draw_mask_fb_ptr_right_b); \
4970 add texture_offset, texture_offset, #0x10; \
4971 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
4973 subs sub_tile_height, sub_tile_height, #1; \
4976 ldr column_data, [sp], #8; /* fb_ptr2 */ \
4977 add texture_offset, texture_offset, #0xF00; \
4978 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4981 #define setup_sprite_tile_half_4bpp_4x(edge) \
4982 setup_sprite_tile_add_blocks(single_4x); \
4983 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4986 and texture_block_ptr, texture_offset, texture_mask; \
4989 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4990 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4992 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4993 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4995 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4996 add texture_offset, texture_offset, #0x10; \
4998 vzip.8 texels_low, texels_high; \
4999 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5000 draw_mask_fb_ptr_##edge##_b); \
5002 pld [fb_ptr, #2048]; \
5003 add fb_ptr, fb_ptr, #2048 * 2; \
5005 subs sub_tile_height, sub_tile_height, #1; \
5008 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5009 add texture_offset, texture_offset, #0xF00; \
5010 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5013 #define setup_sprite_tile_full_8bpp_4x(edge) \
5014 setup_sprite_tile_add_blocks(double_4x); \
5015 add block, block, #16; \
5016 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5019 and texture_block_ptr, texture_offset, texture_mask; \
5022 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5023 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5025 add texture_block_ptr, texture_offset, #8; \
5026 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
5027 draw_mask_fb_ptr_left_b); \
5029 pld [fb_ptr, #2048]; \
5030 and texture_block_ptr, texture_block_ptr, texture_mask; \
5032 add fb_ptr, fb_ptr, #16*2; \
5033 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5035 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5037 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
5038 draw_mask_fb_ptr_right_b); \
5040 add texture_offset, texture_offset, #0x10; \
5041 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
5043 subs sub_tile_height, sub_tile_height, #1; \
5046 sub block, block, #16; \
5047 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5048 add texture_offset, texture_offset, #0xF00; \
5049 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5052 #define setup_sprite_tile_half_8bpp_4x(edge) \
5053 setup_sprite_tile_add_blocks(single_4x); \
5054 add block, block, #16; \
5055 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5058 and texture_block_ptr, texture_offset, texture_mask; \
5061 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5062 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5064 pld [fb_ptr, #2048]; \
5065 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5066 draw_mask_fb_ptr_##edge##_b); \
5068 add texture_offset, texture_offset, #0x10; \
5069 add fb_ptr, fb_ptr, #2048 * 2; \
5071 subs sub_tile_height, sub_tile_height, #1; \
5074 sub block, block, #16; \
5075 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5076 add texture_offset, texture_offset, #0xF00; \
5077 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5080 #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \
5081 add texture_offset, texture_offset_base, #8; \
5082 add fb_ptr, fb_ptr, #16 * 2 \
5084 #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \
5085 mov texture_offset, texture_offset_base \
5087 #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \
5088 setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \
5090 #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \
5091 mov texture_offset, texture_offset_base \
5093 #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \
5094 sub fb_ptr, fb_ptr, #16 * 2 \
5096 #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \
5098 #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \
5099 setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \
5101 #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \
5104 #define setup_sprite_setup_left_draw_mask_fb_ptr_4x() \
5105 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5106 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5107 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5108 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5110 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column_4x() \
5111 mov fb_ptr_advance_column, #32 * 2; \
5112 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5113 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5114 sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11 + 1; \
5115 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5116 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5118 #define setup_sprite_setup_right_draw_mask_fb_ptr_4x() \
5119 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[4]; \
5120 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[5]; \
5121 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[6]; \
5122 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[7] \
5132 // [sp + 12]: color (unused)
5134 #define setup_sprite_tiled_builder(texture_mode, x4mode) \
5136 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \
5138 setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \
5140 setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \
5142 setup_sprite_tile_column_width_single(texture_mode, single, full, none, \
5144 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \
5146 setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \
5148 setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \
5150 setup_sprite_tile_column_width_single(texture_mode, single, half, right, \
5152 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \
5154 setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \
5156 setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \
5158 setup_sprite_tile_column_width_single(texture_mode, single, half, left, \
5160 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \
5162 setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \
5167 function(setup_sprite_##texture_mode##x4mode) \
5168 stmdb sp!, { r4 - r11, r14 }; \
5169 setup_sprite_tiled_initialize_##texture_mode##x4mode(); \
5172 and offset_u, u, #0xF; \
5174 ldr width, [sp, #40]; \
5175 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
5177 ldr height, [sp, #44]; \
5178 add fb_ptr, fb_ptr, y, lsl #11; \
5180 add fb_ptr, fb_ptr, x, lsl #1; \
5181 and offset_v, v, #0xF; \
5183 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
5184 add width_rounded, offset_u, width; \
5186 add height_rounded, offset_v, height; \
5187 add width_rounded, width_rounded, #15; \
5189 add height_rounded, height_rounded, #15; \
5190 mov tile_width, width_rounded, lsr #4; \
5192 /* texture_offset_base = VH-VL-00-00 */\
5193 mov texture_offset_base, v, lsl #8; \
5194 and offset_u_right, width_rounded, #0xF; \
5196 /* texture_offset_base = VH-UH-UL-00 */\
5197 bfi texture_offset_base, u, #4, #8; \
5198 mov right_block_mask, #0xFFFFFFFE; \
5200 setup_sprite_offset_u_adjust##x4mode(); \
5202 /* texture_offset_base = VH-UH-VL-00 */\
5203 bfi texture_offset_base, v, #4, #4; \
5204 mov left_block_mask, #0xFFFFFFFF; \
5206 mov tile_height, height_rounded, lsr #4; \
5207 mvn left_block_mask, left_block_mask, lsl offset_u; \
5209 /* texture_mask = HH-HL-WH-WL */\
5210 ldrh texture_mask, [psx_gpu, #psx_gpu_texture_mask_width_offset]; \
5211 mov right_block_mask, right_block_mask, lsl offset_u_right; \
5213 /* texture_mask_rev = WH-WL-HH-HL */\
5214 rev16 texture_mask_rev, texture_mask; \
5215 vmov block_masks, left_block_mask, right_block_mask; \
5217 /* texture_mask = HH-HL-HL-WL */\
5218 bfi texture_mask, texture_mask_rev, #4, #4; \
5219 /* texture_mask_rev = 00-00-00-WH */\
5220 mov texture_mask_rev, texture_mask_rev, lsr #12; \
5222 /* texture_mask = HH-WH-HL-WL */\
5223 bfi texture_mask, texture_mask_rev, #8, #4; \
5224 setup_sprite_get_left_block_mask##x4mode(); \
5226 mov control_mask, #0; \
5227 setup_sprite_compare_left_block_mask##x4mode(); \
5229 setup_sprite_get_right_block_mask##x4mode(); \
5230 orreq control_mask, control_mask, #0x4; \
5232 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
5233 setup_sprite_compare_right_block_mask##x4mode(); \
5235 orreq control_mask, control_mask, #0x8; \
5236 cmp tile_width, #1; \
5238 add block, psx_gpu, #psx_gpu_blocks_offset; \
5239 orreq control_mask, control_mask, #0x1; \
5241 cmp tile_height, #1; \
5242 add block, block, num_blocks, lsl #6; \
5244 orreq control_mask, control_mask, #0x2; \
5245 JT_OP_REL(9f, control_mask, temp); \
5246 JT_OP(ldr pc, [pc, control_mask, lsl #2]); \
5250 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_full_full##x4mode); \
5251 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_full_none##x4mode); \
5252 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_full_full##x4mode); \
5253 .word JTE(9b, setup_sprite_##texture_mode##_single_single_full_none##x4mode); \
5254 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_half_full##x4mode); \
5255 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_half_right##x4mode); \
5256 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_half_full##x4mode); \
5257 .word JTE(9b, setup_sprite_##texture_mode##_single_single_half_right##x4mode);\
5258 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_full_half##x4mode); \
5259 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_half_left##x4mode); \
5260 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_full_half##x4mode); \
5261 .word JTE(9b, setup_sprite_##texture_mode##_single_single_half_left##x4mode); \
5262 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_half_half##x4mode); \
5264 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_half_half##x4mode); \
5267 setup_sprite_tiled_builder(4bpp,);
5268 setup_sprite_tiled_builder(8bpp,);
5270 #undef draw_mask_fb_ptr_left
5271 #undef draw_mask_fb_ptr_right
5273 setup_sprite_tiled_builder(4bpp, _4x);
5274 setup_sprite_tiled_builder(8bpp, _4x);
5282 #define block_ptr r0
5283 #define num_blocks r1
5285 #define texel_shift_mask r3
5286 #define block_pixels_a r4
5287 #define block_pixels_b r5
5296 #define texels_01 r6
5297 #define texels_23 r7
5298 #define texels_45 r8
5299 #define texels_67 r9
5301 function(texture_sprite_blocks_8bpp)
5302 stmdb sp!, { r4 - r11, r14 }
5303 movw texel_shift_mask, #(0xFF << 1)
5305 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5306 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
5308 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
5309 ldr block_pixels_a, [block_ptr, #16]
5312 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
5313 ldr block_pixels_b, [block_ptr, #20]
5315 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
5316 ldrh texel_0, [clut_ptr, texel_0]
5318 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
5319 ldrh texel_1, [clut_ptr, texel_1]
5321 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
5322 ldr block_pixels_a, [block_ptr, #(64 + 16)]
5324 ldrh texel_2, [clut_ptr, texel_2]
5325 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
5327 ldrh texel_3, [clut_ptr, texel_3]
5328 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
5330 ldrh texel_4, [clut_ptr, texel_4]
5331 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
5333 ldrh texel_5, [clut_ptr, texel_5]
5334 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
5336 ldrh texel_6, [clut_ptr, texel_6]
5337 orr texels_01, texel_0, texel_1, lsl #16
5339 ldrh texel_7, [clut_ptr, texel_7]
5340 orr texels_23, texel_2, texel_3, lsl #16
5342 orr texels_45, texel_4, texel_5, lsl #16
5343 str texels_01, [block_ptr, #0]
5345 orr texels_67, texel_6, texel_7, lsl #16
5346 str texels_23, [block_ptr, #4]
5348 subs num_blocks, num_blocks, #1
5349 str texels_45, [block_ptr, #8]
5351 str texels_67, [block_ptr, #12]
5352 add block_ptr, block_ptr, #64
5356 ldmia sp!, { r4 - r11, pc }
5359 #undef width_rounded
5362 #undef texture_offset
5365 #undef texels_wide_low
5366 #undef texels_wide_high
5378 #define left_offset r8
5379 #define width_rounded r9
5380 #define right_width r10
5382 #define block_width r11
5384 #define texture_offset_base r1
5385 #define texture_mask r2
5386 #define texture_page_ptr r3
5387 #define num_blocks r4
5390 #define texture_offset r8
5391 #define blocks_remaining r9
5393 #define fb_ptr_pitch r12
5394 #define texture_block_ptr r14
5396 #define texture_mask_width r2
5397 #define texture_mask_height r3
5398 #define left_mask_bits r4
5399 #define right_mask_bits r5
5403 #undef block_masks_shifted
5406 #define block_masks d0
5407 #define block_masks_shifted d1
5408 #define draw_mask_fb_ptr d2
5411 #define draw_mask_fb_ptr_a d2
5412 #define draw_mask_fb_ptr_b d3
5413 #define texels_low d4
5414 #define texels_high d5
5415 #define texels_wide_low d6
5416 #define texels_wide_high d7
5417 #define texels_wide q3
5420 setup_sprites_16bpp_flush:
5423 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
5424 bl flush_render_block_buffer
5425 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
5429 add block, psx_gpu, #psx_gpu_blocks_offset
5430 mov num_blocks, block_width
5434 function(setup_sprite_16bpp)
5435 stmdb sp!, { r4 - r11, r14 }
5436 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5439 add fb_ptr, fb_ptr, y, lsl #11
5441 ldr width, [sp, #40]
5442 add fb_ptr, fb_ptr, x, lsl #1
5444 ldr height, [sp, #44]
5445 and left_offset, u, #0x7
5447 add texture_offset_base, u, u
5448 add width_rounded, width, #7
5450 add texture_offset_base, texture_offset_base, v, lsl #11
5451 mov left_mask_bits, #0xFF
5453 ldrb texture_mask_width, [psx_gpu, #psx_gpu_texture_mask_width_offset]
5454 add width_rounded, width_rounded, left_offset
5456 ldrb texture_mask_height, [psx_gpu, #psx_gpu_texture_mask_height_offset]
5457 sub fb_ptr, fb_ptr, left_offset, lsl #1
5459 add texture_mask, texture_mask_width, texture_mask_width
5460 mov right_mask_bits, #0xFE
5462 and right_width, width_rounded, #0x7
5463 mvn left_mask_bits, left_mask_bits, lsl left_offset
5465 add texture_mask, texture_mask, texture_mask_height, lsl #11
5466 mov block_width, width_rounded, lsr #3
5468 mov right_mask_bits, right_mask_bits, lsl right_width
5469 movw fb_ptr_pitch, #(2048 + 16)
5471 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5472 vmov block_masks, left_mask_bits, right_mask_bits
5474 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5475 add block, psx_gpu, #psx_gpu_blocks_offset
5477 bic texture_offset_base, texture_offset_base, #0xF
5480 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
5481 add block, block, num_blocks, lsl #6
5485 vext.32 block_masks_shifted, block_masks, block_masks, #1
5486 vorr.u32 block_masks, block_masks, block_masks_shifted
5487 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5490 add num_blocks, num_blocks, #1
5491 cmp num_blocks, #MAX_BLOCKS
5492 blgt setup_sprites_16bpp_flush
5494 and texture_block_ptr, texture_offset_base, texture_mask
5495 subs height, height, #1
5497 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5498 vld1.u32 { texels }, [texture_block_ptr, :128]
5500 vst1.u32 { texels }, [block, :128]
5501 add block, block, #40
5503 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5506 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5508 add block, block, #24
5509 add texture_offset_base, texture_offset_base, #2048
5510 add fb_ptr, fb_ptr, #2048
5511 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5514 ldmia sp!, { r4 - r11, pc }
5517 add num_blocks, num_blocks, block_width
5518 mov texture_offset, texture_offset_base
5520 cmp num_blocks, #MAX_BLOCKS
5521 blgt setup_sprites_16bpp_flush
5523 add texture_offset_base, texture_offset_base, #2048
5524 and texture_block_ptr, texture_offset, texture_mask
5526 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5527 vld1.u32 { texels }, [texture_block_ptr, :128]
5529 vst1.u32 { texels }, [block, :128]
5530 add block, block, #40
5532 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5533 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5536 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5537 subs blocks_remaining, block_width, #2
5539 add texture_offset, texture_offset, #16
5540 add fb_ptr, fb_ptr, #16
5542 vmov.u8 draw_mask_fb_ptr, #0
5544 add block, block, #24
5548 and texture_block_ptr, texture_offset, texture_mask
5549 subs blocks_remaining, blocks_remaining, #1
5551 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5552 vld1.u32 { texels }, [texture_block_ptr, :128]
5554 vst1.u32 { texels }, [block, :128]
5555 add block, block, #40
5557 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5560 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5562 add texture_offset, texture_offset, #16
5563 add fb_ptr, fb_ptr, #16
5565 add block, block, #24
5569 and texture_block_ptr, texture_offset, texture_mask
5570 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5572 vld1.u32 { texels }, [texture_block_ptr, :128]
5573 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5575 vst1.u32 { texels }, [block, :128]
5576 add block, block, #40
5578 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5579 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5581 add block, block, #24
5582 subs height, height, #1
5584 add fb_ptr, fb_ptr, fb_ptr_pitch
5585 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5589 ldmia sp!, { r4 - r11, pc }
5593 // FIXME: duplicate code with normal version :(
5594 #undef draw_mask_fb_ptr
5596 function(setup_sprite_16bpp_4x)
5597 stmdb sp!, { r4 - r11, r14 }
5598 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5601 add fb_ptr, fb_ptr, y, lsl #11
5603 ldr width, [sp, #40]
5604 add fb_ptr, fb_ptr, x, lsl #1
5606 ldr height, [sp, #44]
5607 and left_offset, u, #0x7
5609 add texture_offset_base, u, u
5610 add width_rounded, width, #7
5612 add texture_offset_base, texture_offset_base, v, lsl #11
5613 movw left_mask_bits, #0xFFFF
5615 ldrb texture_mask_width, [psx_gpu, #psx_gpu_texture_mask_width_offset]
5616 add width_rounded, width_rounded, left_offset
5620 ldrb texture_mask_height, [psx_gpu, #psx_gpu_texture_mask_height_offset]
5621 sub fb_ptr, fb_ptr, left_offset, lsl #1
5623 add texture_mask, texture_mask_width, texture_mask_width
5624 movw right_mask_bits, #0xFFFC
5626 and right_width, width_rounded, #0x7
5627 mvn left_mask_bits, left_mask_bits, lsl left_offset
5631 add texture_mask, texture_mask, texture_mask_height, lsl #11
5632 mov block_width, width_rounded, lsr #3
5634 mov right_mask_bits, right_mask_bits, lsl right_width
5635 movw fb_ptr_pitch, #(2048 + 16) * 2
5637 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4+1
5638 vmov block_masks, left_mask_bits, right_mask_bits
5640 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5641 add block, psx_gpu, #psx_gpu_blocks_offset
5643 bic texture_offset_base, texture_offset_base, #0xF
5646 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
5647 add block, block, num_blocks, lsl #6
5652 vext.32 block_masks_shifted, block_masks, block_masks, #1
5653 vorr.u32 block_masks, block_masks, block_masks_shifted
5654 vdup.u8 draw_mask_fb_ptr_a, block_masks[0]
5655 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5658 add num_blocks, num_blocks, block_width
5659 cmp num_blocks, #MAX_BLOCKS
5660 blgt setup_sprites_16bpp_flush
5662 and texture_block_ptr, texture_offset_base, texture_mask
5663 subs height, height, #1
5665 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5666 vld1.u32 { texels }, [texture_block_ptr, :128]
5668 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5670 add texture_offset_base, texture_offset_base, #2048
5671 add fb_ptr, fb_ptr, #2048*2
5672 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5675 ldmia sp!, { r4 - r11, pc }
5678 add num_blocks, num_blocks, block_width
5679 mov texture_offset, texture_offset_base
5681 vdup.u8 draw_mask_fb_ptr_a, block_masks[0] // left_mask_bits
5682 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5684 cmp num_blocks, #MAX_BLOCKS
5685 blgt setup_sprites_16bpp_flush
5687 add texture_offset_base, texture_offset_base, #2048
5688 and texture_block_ptr, texture_offset, texture_mask
5690 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5691 vld1.u32 { texels }, [texture_block_ptr, :128]
5693 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5695 subs blocks_remaining, block_width, #2*4
5696 add texture_offset, texture_offset, #16
5698 vmov.u8 draw_mask_fb_ptr_a, #0
5699 vmov.u8 draw_mask_fb_ptr_b, #0
5701 add fb_ptr, fb_ptr, #16*2
5705 and texture_block_ptr, texture_offset, texture_mask
5706 subs blocks_remaining, blocks_remaining, #4
5708 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5709 vld1.u32 { texels }, [texture_block_ptr, :128]
5711 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5712 add texture_offset, texture_offset, #16
5714 add fb_ptr, fb_ptr, #16*2
5718 vdup.u8 draw_mask_fb_ptr_a, block_masks[4] // right_mask_bits
5719 vdup.u8 draw_mask_fb_ptr_b, block_masks[5]
5721 and texture_block_ptr, texture_offset, texture_mask
5722 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5724 vld1.u32 { texels }, [texture_block_ptr, :128]
5726 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5727 subs height, height, #1
5729 add fb_ptr, fb_ptr, fb_ptr_pitch
5730 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5734 ldmia sp!, { r4 - r11, pc }
5739 #undef right_mask_bits
5742 #undef blocks_remaining
5752 #define right_width r5
5753 #define right_mask_bits r6
5757 #define fb_ptr_pitch r12
5759 // referenced by setup_sprites_16bpp_flush
5760 #define num_blocks r4
5762 #define block_width r11
5767 #define blocks_remaining r6
5770 #define right_mask q1
5771 #define test_mask q2
5772 #define draw_mask q2
5773 #define draw_mask_bits_fb_ptr d6
5778 function(setup_sprite_untextured)
5779 ldrh r12, [psx_gpu, #psx_gpu_render_state_offset]
5780 tst r12, #(RENDER_STATE_MASK_EVALUATE | RENDER_FLAGS_MODULATE_TEXELS \
5781 | RENDER_FLAGS_BLEND)
5782 ldrbeq r12, [psx_gpu, #psx_gpu_render_mode_offset]
5783 tsteq r12, #RENDER_INTERLACE_ENABLED
5784 beq setup_sprite_untextured_simple
5786 stmdb sp!, { r4 - r11, r14 }
5788 ldr width, [sp, #40]
5789 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5791 ldr height, [sp, #44]
5792 add fb_ptr, fb_ptr, y, lsl #11
5794 add fb_ptr, fb_ptr, x, lsl #1
5795 sub right_width, width, #1
5797 ldr color, [sp, #48]
5800 add block_width, width, #7
5804 mov right_mask_bits, #0xff
5806 sub fb_ptr_pitch, block_width, #1
5807 lsl right_mask_bits, right_width
5809 lsl fb_ptr_pitch, #3+1
5810 ubfx color_r, color, #3, #5
5812 rsb fb_ptr_pitch, #1024*2
5813 ubfx color_g, color, #11, #5
5815 vld1.u32 { test_mask }, [psx_gpu, :128]
5816 ubfx color_b, color, #19, #5
5818 vdup.u16 right_mask, right_mask_bits
5819 orr color, color_r, color_b, lsl #10
5821 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5822 orr color, color, color_g, lsl #5
5824 vtst.u16 right_mask, right_mask, test_mask
5825 add block, psx_gpu, #psx_gpu_blocks_offset
5827 vdup.u16 colors, color
5828 add block, block, num_blocks, lsl #6
5831 setup_sprite_untextured_height_loop:
5832 add num_blocks, block_width
5833 sub blocks_remaining, block_width, #1
5835 cmp num_blocks, #MAX_BLOCKS
5836 blgt setup_sprites_16bpp_flush
5838 cmp blocks_remaining, #0
5841 vmov.u8 draw_mask, #0 /* zero_mask */
5842 vmov.u8 draw_mask_bits_fb_ptr, #0
5845 vst1.u32 { draw_mask }, [block, :128]!
5846 subs blocks_remaining, #1
5848 vst1.u32 { colors }, [block, :128]
5849 add block, block, #24
5851 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5852 vst1.u32 { draw_mask_bits_fb_ptr }, [block, :64]
5854 add block, block, #24
5859 vst1.u32 { right_mask }, [block, :128]!
5862 vst1.u32 { colors }, [block, :128]
5863 add block, block, #24
5865 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5866 vst1.u32 { draw_mask_bits_fb_ptr }, [block, :64]
5868 add block, block, #24
5869 add fb_ptr, fb_ptr_pitch
5871 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5872 bgt setup_sprite_untextured_height_loop
5874 ldmia sp!, { r4 - r11, pc }
5878 #undef texture_page_ptr
5880 #undef dirty_textures_mask
5881 #undef current_texture_mask
5884 #define current_texture_page r1
5885 #define texture_page_ptr r2
5886 #define vram_ptr_a r3
5887 #define current_texture_page_x r12
5888 #define current_texture_page_y r4
5889 #define dirty_textures_mask r5
5893 #define current_texture_mask r9
5895 #define vram_ptr_b r11
5897 #define texel_block_a d0
5898 #define texel_block_b d1
5899 #define texel_block_expanded_a q1
5900 #define texel_block_expanded_b q2
5901 #define texel_block_expanded_ab q2
5902 #define texel_block_expanded_c q3
5903 #define texel_block_expanded_d q4
5904 #define texel_block_expanded_cd q3
5906 function(update_texture_4bpp_cache)
5907 stmdb sp!, { r4 - r11, r14 }
5910 ldrb current_texture_page, [psx_gpu, #psx_gpu_current_texture_page_offset]
5912 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_base_offset]
5913 ldr vram_ptr_a, [psx_gpu, #psx_gpu_vram_ptr_offset]
5915 and current_texture_page_x, current_texture_page, #0xF
5916 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
5918 mov current_texture_page_y, current_texture_page, lsr #4
5919 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
5921 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5924 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5925 bic dirty_textures_mask, current_texture_mask
5928 str dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
5933 add vram_ptr_b, vram_ptr_a, #2048
5936 vld1.u32 { texel_block_a }, [vram_ptr_a, :64], c_4096
5937 vld1.u32 { texel_block_b }, [vram_ptr_b, :64], c_4096
5939 vmovl.u8 texel_block_expanded_a, texel_block_a
5940 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5941 vmovl.u8 texel_block_expanded_c, texel_block_b
5942 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5944 vbic.u16 texel_block_expanded_a, #0x00F0
5945 vbic.u16 texel_block_expanded_b, #0x00F0
5946 vbic.u16 texel_block_expanded_c, #0x00F0
5947 vbic.u16 texel_block_expanded_d, #0x00F0
5949 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
5950 texel_block_expanded_b
5951 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
5952 texel_block_expanded_d
5954 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
5955 [texture_page_ptr, :256]!
5957 subs sub_y, sub_y, #1
5961 add vram_ptr_a, vram_ptr_a, #8
5962 add vram_ptr_b, vram_ptr_b, #8
5964 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5965 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5967 subs tile_x, tile_x, #1
5971 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5972 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5974 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5975 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5977 subs tile_y, tile_y, #1
5981 ldmia sp!, { r4 - r11, pc }
5984 #undef current_texture_page
5987 #define texture_page r1
5988 #define texture_page_ptr r2
5989 #define vram_ptr_a r3
5990 #define texture_page_x r12
5991 #define texture_page_y r4
5992 #define current_texture_page r5
5997 #define vram_ptr_b r11
6009 function(update_texture_8bpp_cache_slice)
6010 stmdb sp!, { r4 - r11, r14 }
6013 ldrb current_texture_page, [psx_gpu, #psx_gpu_current_texture_page_offset]
6014 ldr vram_ptr_a, [psx_gpu, #psx_gpu_vram_ptr_offset]
6016 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_base_offset]
6019 and texture_page_x, texture_page, #0xF
6020 mov texture_page_y, texture_page, lsr #4
6022 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
6025 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
6026 eor current_texture_page, current_texture_page, texture_page
6028 ands current_texture_page, current_texture_page, #0x1
6031 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6034 add vram_ptr_b, vram_ptr_a, #2048
6037 vld1.u32 { texels_a }, [vram_ptr_a, :128], c_4096
6038 vld1.u32 { texels_b }, [vram_ptr_b, :128], c_4096
6039 vld1.u32 { texels_c }, [vram_ptr_a, :128], c_4096
6040 vld1.u32 { texels_d }, [vram_ptr_b, :128], c_4096
6042 vst1.u32 { texels_a, texels_b }, [texture_page_ptr, :256]!
6043 vst1.u32 { texels_c, texels_d }, [texture_page_ptr, :256]!
6045 subs sub_y, sub_y, #1
6050 add vram_ptr_a, vram_ptr_a, #16
6051 add vram_ptr_b, vram_ptr_b, #16
6053 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
6054 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
6056 subs tile_x, tile_x, #1
6061 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
6062 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
6064 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
6065 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
6067 subs tile_y, tile_y, #1
6068 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6073 ldmia sp!, { r4 - r11, pc }
6076 /* void scale2x_tiles8(void *dst, const void *src, int w8, int h) */
6077 function(scale2x_tiles8)
6081 add r12, r0, #1024*2
6085 vld1.u16 { q0 }, [r1, :128]!
6086 vld1.u16 { q2 }, [r1, :128]!
6092 vst1.u16 { q0, q1 }, [r0, :128]!
6093 vst1.u16 { q0, q1 }, [r12, :128]!
6095 vst1.u16 { q2, q3 }, [r0, :128]!
6096 vst1.u16 { q2, q3 }, [r12, :128]!
6103 sub r0, r0, r2, lsl #4+1
6105 add r12, r0, #1024*2
6111 // vim:filetype=armasm