2 * Copyright (C) 2011 Gilead Kutnick "Exophase" <exophase@gmail.com>
3 * Copyright (C) 2012 GraÅžvydas Ignotas "notaz" <notasas@gmail.com>
5 * This program is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU General Public License as
7 * published by the Free Software Foundation; either version 2 of
8 * the License, or (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License for more details.
18 #define MAX_BLOCKS_PER_ROW 128
20 #define RENDER_STATE_MASK_EVALUATE 0x20
21 #define RENDER_FLAGS_MODULATE_TEXELS 0x1
22 #define RENDER_FLAGS_BLEND 0x2
23 #define RENDER_INTERLACE_ENABLED 0x1
25 #include "psx_gpu_offsets.h"
27 #define psx_gpu_b_dx_offset (psx_gpu_b_block_span_offset + 4)
29 #define edge_data_left_x_offset 0
30 #define edge_data_num_blocks_offset 2
31 #define edge_data_right_mask_offset 4
32 #define edge_data_y_offset 6
78 #define gw_by_l gw_bx_l
103 #define uvrg_xxxx1 q4
108 #define uvrg_xxxx2 q5
113 #define yyyy_uvrg0 q6
118 #define yyyy_uvrg1 q7
123 #define yyyy_uvrg2 q8
150 #define uvrg_base q11
193 #define uvrgb_phase q13
199 #define function(name) \
201 .type name, %function; \
204 #define JT_OP_REL(table_label, index_reg, temp)
205 #define JT_OP(x...) x
206 #define JTE(start, target) target
208 #define EXTRA_UNSAVED_REGS
212 #define function(name) \
217 #define JT_OP_REL(table_label, index_reg, temp) \
218 adr temp, table_label; \
219 ldr temp, [temp, index_reg, lsl #2]; \
223 #define JTE(start, target) (target - start)
225 // r7 is preserved, but add it for EABI alignment..
226 #define EXTRA_UNSAVED_REGS r7, r9,
228 #define flush_render_block_buffer _flush_render_block_buffer
229 #define setup_sprite_untextured_simple _setup_sprite_untextured_simple
230 #define update_texture_8bpp_cache _update_texture_8bpp_cache
239 function(compute_all_gradients)
240 // First compute the triangle area reciprocal and shift. The division will
241 // happen concurrently with much of the work which follows.
242 @ r12 = psx_gpu->triangle_area
243 ldr r12, [psx_gpu, #psx_gpu_triangle_area_offset]
244 stmdb sp!, { r4 - r11, lr }
246 @ load exponent of 62 into upper half of double
248 clz r14, r12 @ r14 = shift
250 movt r4, #((62 + 1023) << 4)
251 mov r12, r12, lsl r14 @ r12 = triangle_area_normalized
253 @ load area normalized into lower half of double
255 vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n
257 movt r4, #((1022 + 31) << 4)
260 add r4, r4, r12, lsr #11
263 vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n
265 // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) =
266 // ( d0 * d1 ) - ( d2 * d3 ) =
267 // ( m0 ) - ( m1 ) = gradient
269 // This is split to do 12 elements at a time over three sets: a, b, and c.
270 // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so
271 // two of the slots are unused.
273 // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as
276 // First type is: uvrg bxxx xxxx
277 // Second type is: yyyy ybyy uvrg
278 // Since x_a and y_c are the same the same variable is used for both.
280 vld1.u32 { v0 }, [v_a, :128] @ v0 = { uvrg0, b0, x0, y0 }
281 ldrsh x0, [v_a, #8] @ load x0
283 vld1.u32 { v1 }, [v_b, :128] @ v1 = { uvrg1, b1, x1, y1}
284 ldrh x1, [v_b, #8] @ load x1
286 vld1.u32 { v2 }, [v_c, :128] @ v2 = { uvrg2, b2, x2, y2 }
287 ldrh x2, [v_c, #8] @ load x2
289 vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- }
290 ldrh y0, [v_a, #10] @ load y0
292 vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- }
293 ldrh y1, [v_b, #10] @ load y1
295 vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- }
296 ldrh y2, [v_c, #10] @ load y2
298 vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 }
299 vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 }
301 orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 }
302 pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 }
304 vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 }
305 vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 }
307 vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 }
308 vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 }
310 ldrb b2, [v_c, #4] @ load b2
311 orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 }
313 ldrb b1, [v_b, #4] @ load b1
314 orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 }
316 vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 }
317 vsub.s16 d0_ab, x1_ab, x0_ab
319 ldrb b0, [v_a, #4] @ load b0
320 orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 }
322 vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 }
323 vsub.s16 d2_ab, x2_ab, x1_ab
325 vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 }
326 vsub.s16 d1_ab, y2_ab, y1_ab
328 orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 }
329 ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 }
331 ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 }
332 ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 }
334 vsub.s16 d3_ab, y1_ab, y0_ab
335 smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) -
336 @ ((x2 - X1) * (b1 - b0))
337 vmull.s16 ga_uvrg_x, d0_a, d1_a
338 smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) -
339 @ ((b2 - b1) * (y1 - y0))
340 vmlsl.s16 ga_uvrg_x, d2_a, d3_a
341 movs gs_bx, ga_bx, asr #31
343 vmull.s16 ga_uvrg_y, d0_b, d1_b
344 rsbmi ga_bx, ga_bx, #0
346 @ r12 = psx_gpu->uvrgb_phase
347 ldr r12, [psx_gpu, #psx_gpu_uvrgb_phase_offset]
349 vmlsl.s16 ga_uvrg_y, d2_b, d3_b
350 movs gs_by, ga_by, asr #31
352 vshr.u64 d0, d30, #22
353 add b_base, r12, b0, lsl #16
355 vdup.u32 uvrgb_phase, r12
357 rsbmi ga_by, ga_by, #0
358 vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0
360 @ r12 = psx_gpu->triangle_winding_offset
361 ldrb r12, [psx_gpu, #psx_gpu_triangle_winding_offset]
362 vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0
364 rsb r12, r12, #0 @ r12 = -(triangle->winding)
366 vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w }
367 sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS)
369 vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16
370 vdup.u32 r_shift, r14 @ r_shift = { shift, shift, shift, shift }
372 vadd.u32 uvrg_base, uvrgb_phase
373 vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x)
375 vmov area_r_s, s0 @ area_r_s = triangle_reciprocal
376 vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y)
378 vmull.u32 gw_rg_x, ga_rg_x, d0[0]
379 vmull.u32 gw_uv_x, ga_uv_x, d0[0]
380 vmull.u32 gw_rg_y, ga_rg_y, d0[0]
381 vmull.u32 gw_uv_y, ga_uv_y, d0[0]
383 vshl.u64 gw_rg_x, gw_rg_x, r_shift
384 vshl.u64 gw_uv_x, gw_uv_x, r_shift
385 vshl.u64 gw_rg_y, gw_rg_y, r_shift
386 vshl.u64 gw_uv_y, gw_uv_y, r_shift
388 veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask
389 vmovn.u64 g_uv_x, gw_uv_x
391 veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask
392 vmovn.u64 g_rg_x, gw_rg_x
394 veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
395 vmovn.u64 g_uv_y, gw_uv_y
397 vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
398 vmovn.u64 g_rg_y, gw_rg_y
400 veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
401 mov ga_bx, ga_bx, lsl #13
403 vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
404 mov ga_by, ga_by, lsl #13
407 umull gw_bx_l, gw_bx_h, ga_bx, area_r_s
409 vshl.u32 g_uvrg_x, g_uvrg_x, #4
410 vshl.u32 g_uvrg_y, g_uvrg_y, #4
412 umull gw_by_l, gw_by_h, ga_by, area_r_s
413 vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0]
415 eor gs_bx, gs_bx, r12
416 vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1
418 veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0
419 eor gs_by, gs_by, r12
421 rsb r11, r14, #0 @ r11 = negative shift for scalar lsr
422 add store_a, psx_gpu, #psx_gpu_uvrg_offset
424 sub r11, r11, #(32 - 13)
426 add store_b, store_a, #16
429 vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1
430 vst1.u32 { uvrg_base }, [store_a, :128], store_inc
432 vst1.u32 { uvrg_dx1 }, [store_b, :128], store_inc
433 mov g_bx, gw_bx_h, lsr r11
435 vst1.u32 { g_uvrg_y }, [store_a, :128], store_inc
436 mov g_by, gw_by_h, lsr r11
438 vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \
439 [store_b, :128], store_inc
440 eor g_bx, g_bx, gs_bx
442 vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \
443 [store_b, :128], store_inc
444 sub g_bx, g_bx, gs_bx
447 eor g_by, g_by, gs_by
449 mls b_base, g_bx, x0, b_base
450 sub g_by, g_by, gs_by
455 add g_bx2, g_bx, g_bx
456 add g_bx3, g_bx, g_bx2
458 stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by }
460 ldmia sp!, { r4 - r11, pc }
477 #define height_minor_a r7
478 #define height_minor_b r8
479 #define height_major r9
482 #define reciprocal_table_ptr r10
484 #define edge_alt_low r4
485 #define edge_alt_high r5
486 #define edge_dx_dy_alt r6
487 #define edge_shift_alt r10
489 #define edge_dx_dy_alt_low r4
490 #define edge_dx_dy_alt_high r5
492 #define span_edge_data r4
493 #define span_uvrg_offset r5
494 #define span_b_offset r6
502 #define alternate_x q0
503 #define alternate_dx_dy q1
504 #define alternate_x_32 q2
506 #define alternate_x_low d0
507 #define alternate_x_high d1
508 #define alternate_dx_dy_low d2
509 #define alternate_dx_dy_high d3
510 #define alternate_x_32_low d4
511 #define alternate_x_32_high d5
515 #define left_dx_dy q5
516 #define right_dx_dy q6
518 #define right_edge q8
520 #define left_x_low d6
521 #define left_x_high d7
522 #define right_x_low d8
523 #define right_x_high d9
524 #define left_dx_dy_low d10
525 #define left_dx_dy_high d11
526 #define right_dx_dy_low d12
527 #define right_dx_dy_high d13
528 #define left_edge_low d14
529 #define left_edge_high d15
530 #define right_edge_low d16
531 #define right_edge_high d17
533 #define y_mid_point d18
536 #define left_right_x_16 q11
537 #define span_shifts_y q12
540 #define span_shifts d24
545 #define left_right_x_16_low d22
546 #define left_right_x_16_high d23
551 #define alternate_x_16 d4
554 #define v_clip_low d6
556 #define right_x_32 q10
557 #define left_x_32 q11
558 #define alternate_select d24
560 #define right_x_32_low d20
561 #define right_x_32_high d21
562 #define left_x_32_low d22
563 #define left_x_32_high d23
566 #define edges_dx_dy d2
567 #define edge_shifts d3
568 #define edge_shifts_64 q2
570 #define edges_xy_left d0
571 #define edges_xy_right d1
573 #define height_reciprocals d6
581 #define heights_b d12
582 #define edges_dx_dy_64 q10
584 #define edges_dx_dy_64_left d20
585 #define edges_dx_dy_64_right d21
588 #define setup_spans_prologue() \
589 stmdb sp!, { r4 - r11, lr }; \
591 ldrsh x_a, [v_a, #8]; \
592 ldrsh x_b, [v_b, #8]; \
593 ldrsh x_c, [v_c, #8]; \
594 ldrsh y_a, [v_a, #10]; \
595 ldrsh y_b, [v_b, #10]; \
596 ldrsh y_c, [v_c, #10]; \
598 add temp, psx_gpu, #psx_gpu_uvrg_offset; \
599 vld1.32 { uvrg }, [temp]; \
600 add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \
601 vld1.32 { uvrg_dy }, [temp]; \
602 ldr reciprocal_table_ptr, [psx_gpu, #psx_gpu_reciprocal_table_ptr_offset]; \
604 vmov.u32 c_0x01, #0x01 \
606 #define setup_spans_load_b() \
607 ldr b, [psx_gpu, #psx_gpu_b_offset]; \
608 ldr b_dy, [psx_gpu, #psx_gpu_b_dy_offset] \
610 #define setup_spans_prologue_b() \
611 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
612 add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \
614 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
615 vmov.u16 c_0x0004, #0x0004; \
617 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
618 vmov.u16 c_0x0001, #0x0001; \
620 vld1.u16 { left_edge_low[], left_edge_high[] }, [temp]; \
621 add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \
623 vld1.u16 { right_edge_low[], right_edge_high[] }, [temp]; \
624 vadd.u16 right_edge, right_edge, c_0x0001; \
626 vmov.u16 c_0x0007, #0x0007; \
627 vmvn.u16 c_0xFFFE, #0x0001 \
630 #define compute_edge_delta_x2() \
631 ldr temp, [reciprocal_table_ptr, height, lsl #2]; \
633 vdup.u32 heights, height; \
634 vsub.u32 widths, x_ends, x_starts; \
636 vdup.u32 edge_shifts, temp; \
637 vsub.u32 heights_b, heights, c_0x01; \
638 vshr.u32 height_reciprocals, edge_shifts, #10; \
640 vmla.s32 heights_b, x_starts, heights; \
641 vbic.u16 edge_shifts, #0xE0; \
642 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
643 vmull.s32 edges_xy, heights_b, height_reciprocals \
646 #define height_reciprocal_alt r11
647 #define height_b_alt r12
649 #define compute_edge_delta_x3(start_c, height_a, height_b) \
650 vmov heights, height_a, height_b; \
651 ldr temp, [reciprocal_table_ptr, height_a, lsl #2]; \
652 vmov.u32 edge_shifts[0], temp; \
653 ldr temp, [reciprocal_table_ptr, height_b, lsl #2]; \
654 vmov.u32 edge_shifts[1], temp; \
655 ldr edge_shift_alt, [reciprocal_table_ptr, height_minor_b, lsl #2]; \
657 vsub.u32 widths, x_ends, x_starts; \
658 sub width_alt, x_c, start_c; \
660 vsub.u32 heights_b, heights, c_0x01; \
661 sub height_b_alt, height_minor_b, #1; \
663 vshr.u32 height_reciprocals, edge_shifts, #10; \
664 lsr height_reciprocal_alt, edge_shift_alt, #10; \
666 vmla.s32 heights_b, x_starts, heights; \
667 mla height_b_alt, height_minor_b, start_c, height_b_alt; \
669 vbic.u16 edge_shifts, #0xE0; \
670 and edge_shift_alt, edge_shift_alt, #0x1F; \
672 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
673 mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \
675 vmull.s32 edges_xy, heights_b, height_reciprocals; \
676 smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \
679 #define setup_spans_adjust_y_up() \
680 vsub.u32 y_x4, y_x4, c_0x0004 \
682 #define setup_spans_adjust_y_down() \
683 vadd.u32 y_x4, y_x4, c_0x0004 \
685 #define setup_spans_adjust_interpolants_up() \
686 vsub.u32 uvrg, uvrg, uvrg_dy; \
689 #define setup_spans_adjust_interpolants_down() \
690 vadd.u32 uvrg, uvrg, uvrg_dy; \
694 #define setup_spans_clip_interpolants_increment() \
695 mla b, b_dy, clip, b; \
696 vmla.s32 uvrg, uvrg_dy, v_clip \
698 #define setup_spans_clip_interpolants_decrement() \
699 mls b, b_dy, clip, b; \
700 vmls.s32 uvrg, uvrg_dy, v_clip \
702 #define setup_spans_clip_alternate_yes() \
703 smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \
705 #define setup_spans_clip_alternate_no() \
707 #define setup_spans_clip(direction, alternate_active) \
708 vdup.u32 v_clip, clip; \
709 setup_spans_clip_alternate_##alternate_active(); \
710 setup_spans_clip_interpolants_##direction(); \
711 vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \
714 #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \
715 vmovl.s32 edge_shifts_64, edge_shifts; \
716 vmovl.s32 edges_dx_dy_64, edges_dx_dy; \
718 vshl.s64 edges_xy, edges_xy, edge_shifts_64; \
719 vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \
721 vmov left_x_low, edges_xy_##left_index; \
722 vmov right_x_low, edges_xy_##right_index; \
724 vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \
725 vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \
726 vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \
727 vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \
729 vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \
730 vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \
732 vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \
733 vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \
736 #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \
737 setup_spans_adjust_edges_alternate_no(left_index, right_index); \
739 vdup.u16 y_mid_point, y_b; \
740 rsb temp, edge_shift_alt, #32; \
742 lsl edge_alt_high, edge_alt_high, edge_shift_alt; \
743 orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \
744 lsl edge_alt_low, edge_alt_low, edge_shift_alt; \
745 vmov alternate_x_low, edge_alt_low, edge_alt_high; \
747 asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \
748 lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \
749 vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \
750 vmov alternate_dx_dy_high, alternate_dx_dy_low; \
752 vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \
753 vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \
756 #define setup_spans_y_select_up() \
757 vclt.s16 alternate_select, y_x4, y_mid_point \
759 #define setup_spans_y_select_down() \
760 vcgt.s16 alternate_select, y_x4, y_mid_point \
763 #define setup_spans_alternate_select_left() \
764 vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \
766 #define setup_spans_alternate_select_right() \
767 vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \
770 #define setup_spans_set_x4_alternate_yes(alternate, direction) \
771 vshrn.s64 alternate_x_32_low, alternate_x, #32; \
772 vshrn.s64 left_x_32_low, left_x, #32; \
773 vshrn.s64 right_x_32_low, right_x, #32; \
775 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
776 vadd.u64 left_x, left_x, left_dx_dy; \
777 vadd.u64 right_x, right_x, right_dx_dy; \
779 vshrn.s64 alternate_x_32_high, alternate_x, #32; \
780 vshrn.s64 left_x_32_high, left_x, #32; \
781 vshrn.s64 right_x_32_high, right_x, #32; \
783 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
784 vadd.u64 left_x, left_x, left_dx_dy; \
785 vadd.u64 right_x, right_x, right_dx_dy; \
787 vmovn.u32 alternate_x_16, alternate_x_32; \
788 setup_spans_y_select_##direction(); \
789 vmovn.u32 left_right_x_16_low, left_x_32; \
791 vmovn.u32 left_right_x_16_high, right_x_32; \
792 setup_spans_alternate_select_##alternate(); \
794 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
795 str b, [span_b_offset], #4; \
796 setup_spans_adjust_interpolants_##direction(); \
798 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
800 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
801 str b, [span_b_offset], #4; \
802 setup_spans_adjust_interpolants_##direction(); \
804 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
806 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
807 str b, [span_b_offset], #4; \
808 setup_spans_adjust_interpolants_##direction(); \
810 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
811 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
812 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
814 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
815 str b, [span_b_offset], #4; \
816 setup_spans_adjust_interpolants_##direction(); \
818 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
819 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
821 vst4.u16 { left_right_x_16, span_shifts_y }, [span_edge_data]!; \
823 setup_spans_adjust_y_##direction() \
826 #define setup_spans_set_x4_alternate_no(alternate, direction) \
827 vshrn.s64 left_x_32_low, left_x, #32; \
828 vshrn.s64 right_x_32_low, right_x, #32; \
830 vadd.u64 left_x, left_x, left_dx_dy; \
831 vadd.u64 right_x, right_x, right_dx_dy; \
833 vshrn.s64 left_x_32_high, left_x, #32; \
834 vshrn.s64 right_x_32_high, right_x, #32; \
836 vadd.u64 left_x, left_x, left_dx_dy; \
837 vadd.u64 right_x, right_x, right_dx_dy; \
839 vmovn.u32 left_right_x_16_low, left_x_32; \
840 vmovn.u32 left_right_x_16_high, right_x_32; \
842 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
843 str b, [span_b_offset], #4; \
844 setup_spans_adjust_interpolants_##direction(); \
846 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
848 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
849 str b, [span_b_offset], #4; \
850 setup_spans_adjust_interpolants_##direction(); \
852 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
854 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
855 str b, [span_b_offset], #4; \
856 setup_spans_adjust_interpolants_##direction(); \
858 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
859 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
860 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
862 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
863 str b, [span_b_offset], #4; \
864 setup_spans_adjust_interpolants_##direction(); \
866 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
867 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
869 vst4.u16 { left_right_x_16, span_shifts_y }, [span_edge_data]!; \
871 setup_spans_adjust_y_##direction() \
874 #define edge_adjust_low r11
875 #define edge_adjust_high r12
877 #define setup_spans_alternate_adjust_yes() \
878 smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \
879 subs edge_alt_low, edge_alt_low, edge_adjust_low; \
880 sbc edge_alt_high, edge_alt_high, edge_adjust_high \
882 #define setup_spans_alternate_adjust_no() \
885 #define setup_spans_down(left_index, right_index, alternate, alternate_active) \
886 setup_spans_alternate_adjust_##alternate_active(); \
887 setup_spans_load_b(); \
889 ldrsh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]; \
890 subs y_c, y_c, temp; \
891 subgt height, height, y_c; \
892 addgt height, height, #1; \
894 ldrsh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]; \
895 subs clip, temp, y_a; \
898 sub height, height, clip; \
899 add y_a, y_a, clip; \
900 setup_spans_clip(increment, alternate_active); \
906 orr temp, y_a, y_a, lsl #16; \
907 add temp, temp, #(1 << 16); \
909 add y_a, y_a, #(2 << 16); \
910 vmov y_x4, temp, y_a; \
912 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
914 setup_spans_prologue_b(); \
916 strh height, [psx_gpu, #psx_gpu_num_spans_offset]; \
919 setup_spans_set_x4_alternate_##alternate_active(alternate, down); \
920 subs height, height, #4; \
926 #define setup_spans_alternate_pre_increment_yes() \
927 adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \
928 adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \
930 #define setup_spans_alternate_pre_increment_no() \
933 #define setup_spans_up_decrement_yes() \
934 suble height, height, #1 \
936 #define setup_spans_up_decrement_no() \
939 #define setup_spans_up(left_index, right_index, alternate, alternate_active) \
940 setup_spans_alternate_adjust_##alternate_active(); \
941 setup_spans_load_b(); \
944 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]; \
945 subs temp, temp, y_c; \
946 subgt height, height, temp; \
947 setup_spans_up_decrement_##alternate_active(); \
949 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]; \
950 subs clip, y_a, temp; \
953 sub height, height, clip; \
954 sub y_a, y_a, clip; \
955 setup_spans_clip(decrement, alternate_active); \
961 orr temp, y_a, y_a, lsl #16; \
962 sub temp, temp, #(1 << 16); \
964 sub y_a, y_a, #(2 << 16); \
965 vmov y_x4, temp, y_a; \
967 vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \
969 setup_spans_alternate_pre_increment_##alternate_active(); \
970 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
972 setup_spans_adjust_interpolants_up(); \
973 setup_spans_prologue_b(); \
975 strh height, [psx_gpu, #psx_gpu_num_spans_offset]; \
978 setup_spans_set_x4_alternate_##alternate_active(alternate, up); \
979 subs height, height, #4; \
985 #define setup_spans_epilogue() \
986 ldmia sp!, { r4 - r11, pc } \
989 #define setup_spans_up_up(minor, major) \
990 setup_spans_prologue(); \
991 sub height_minor_a, y_a, y_b; \
992 sub height_minor_b, y_b, y_c; \
993 sub height, y_a, y_c; \
995 vdup.u32 x_starts, x_a; \
996 vmov x_ends, x_c, x_b; \
998 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
999 setup_spans_up(major, minor, minor, yes); \
1000 setup_spans_epilogue() \
1002 function(setup_spans_up_left)
1003 setup_spans_up_up(left, right)
1005 function(setup_spans_up_right)
1006 setup_spans_up_up(right, left)
1008 #define setup_spans_down_down(minor, major) \
1009 setup_spans_prologue(); \
1010 sub height_minor_a, y_b, y_a; \
1011 sub height_minor_b, y_c, y_b; \
1012 sub height, y_c, y_a; \
1014 vdup.u32 x_starts, x_a; \
1015 vmov x_ends, x_c, x_b; \
1017 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
1018 setup_spans_down(major, minor, minor, yes); \
1019 setup_spans_epilogue() \
1021 function(setup_spans_down_left)
1022 setup_spans_down_down(left, right)
1024 function(setup_spans_down_right)
1025 setup_spans_down_down(right, left)
1028 #define setup_spans_up_flat() \
1029 sub height, y_a, y_c; \
1031 compute_edge_delta_x2(); \
1032 setup_spans_up(left, right, none, no); \
1033 setup_spans_epilogue() \
1035 function(setup_spans_up_a)
1036 setup_spans_prologue()
1038 vmov x_starts, x_a, x_b
1039 vdup.u32 x_ends, x_c
1041 setup_spans_up_flat()
1043 function(setup_spans_up_b)
1044 setup_spans_prologue()
1046 vdup.u32 x_starts, x_a
1047 vmov x_ends, x_b, x_c
1049 setup_spans_up_flat()
1051 #define setup_spans_down_flat() \
1052 sub height, y_c, y_a; \
1054 compute_edge_delta_x2(); \
1055 setup_spans_down(left, right, none, no); \
1056 setup_spans_epilogue() \
1058 function(setup_spans_down_a)
1059 setup_spans_prologue()
1061 vmov x_starts, x_a, x_b
1062 vdup.u32 x_ends, x_c
1064 setup_spans_down_flat()
1066 function(setup_spans_down_b)
1067 setup_spans_prologue()
1069 vdup.u32 x_starts, x_a
1070 vmov x_ends, x_b, x_c
1072 setup_spans_down_flat()
1077 #define edges_xy_b q11
1078 #define edges_dx_dy_b d26
1079 #define edge_shifts_b d27
1080 #define edges_dx_dy_and_shifts_b q13
1081 #define height_increment d20
1083 #define edges_dx_dy_and_shifts q1
1085 #define edges_xy_b_left d22
1086 #define edges_xy_b_right d23
1088 #define setup_spans_up_down_load_edge_set_b() \
1089 vmov edges_xy, edges_xy_b; \
1090 vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \
1093 function(setup_spans_up_down)
1094 setup_spans_prologue()
1096 // s32 middle_y = y_a;
1097 sub height_minor_a, y_a, y_b
1098 sub height_minor_b, y_c, y_a
1099 sub height_major, y_c, y_b
1101 vmov x_starts, x_a, x_c
1102 vdup.u32 x_ends, x_b
1104 compute_edge_delta_x3(x_a, height_minor_a, height_major)
1107 vmov height_increment, temp, height_minor_b
1108 vmlal.s32 edges_xy, edges_dx_dy, height_increment
1110 vmov edges_xy_b_left, edge_alt_low, edge_alt_high
1111 vmov edges_xy_b_right, edges_xy_right
1113 vmov edge_shifts_b, edge_shifts
1114 vmov.u32 edge_shifts_b[0], edge_shift_alt
1116 vneg.s32 edges_dx_dy_b, edges_dx_dy
1117 vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt
1121 setup_spans_load_b()
1124 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]
1125 subs temp, temp, y_b
1126 subgt height_minor_a, height_minor_a, temp
1128 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]
1129 subs clip, y_a, temp
1132 sub height_minor_a, height_minor_a, clip
1134 setup_spans_clip(decrement, no)
1137 cmp height_minor_a, #0
1140 orr temp, y_a, y_a, lsl #16
1141 sub temp, temp, #(1 << 16)
1143 sub y_a, y_a, #(2 << 16)
1144 vmov y_x4, temp, y_a
1146 vaddw.s32 edges_xy, edges_xy, edges_dx_dy
1148 strh height_minor_a, [psx_gpu, #psx_gpu_num_spans_offset]
1150 setup_spans_adjust_edges_alternate_no(left, right);
1151 setup_spans_adjust_interpolants_up()
1152 setup_spans_up_down_load_edge_set_b()
1154 setup_spans_prologue_b()
1158 setup_spans_set_x4_alternate_no(none, up)
1159 subs height_minor_a, height_minor_a, #4
1162 add span_edge_data, span_edge_data, height_minor_a, lsl #3
1163 add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4
1164 add span_b_offset, span_b_offset, height_minor_a, lsl #2
1167 add temp, psx_gpu, #psx_gpu_uvrg_offset
1168 vld1.32 { uvrg }, [temp]
1171 setup_spans_load_b()
1173 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]
1175 subgt height_minor_b, height_minor_b, y_c
1176 addgt height_minor_b, height_minor_b, #1
1178 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]
1179 subs clip, temp, y_a
1182 sub height_minor_b, height_minor_b, clip
1184 setup_spans_clip(increment, no)
1187 cmp height_minor_b, #0
1190 orr temp, y_a, y_a, lsl #16
1191 add temp, temp, #(1 << 16)
1193 add y_a, y_a, #(2 << 16)
1194 vmov y_x4, temp, y_a
1196 setup_spans_adjust_edges_alternate_no(left, right)
1198 ldrh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1199 add temp, temp, height_minor_b
1201 cmp temp, #MAX_SPANS
1204 strh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1207 setup_spans_set_x4_alternate_no(none, down)
1208 subs height_minor_b, height_minor_b, #4
1212 setup_spans_epilogue()
1215 setup_spans_up_down_load_edge_set_b()
1216 setup_spans_prologue_b()
1220 // FIXME: overflow corner case
1221 sub temp, temp, height_minor_b
1222 bics height_minor_b, #3
1223 add temp, temp, height_minor_b
1224 strh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1228 #undef span_uvrg_offset
1229 #undef span_edge_data
1230 #undef span_b_offset
1235 #define num_spans r1
1236 #define span_uvrg_offset r2
1237 #define span_edge_data r3
1238 #define span_b_offset r4
1240 #define span_num_blocks r6
1244 #define dither_offset_ptr r10
1245 #define block_ptr_a r11
1247 #define num_blocks r14
1249 #define uvrg_dx_ptr r2
1250 #define texture_mask_ptr r3
1251 #define dither_shift r8
1252 #define dither_row r10
1257 #define block_ptr_b r10
1259 #define block_span_ptr r10
1260 #define right_mask r8
1279 #define b_whole_8 d14
1280 #define fb_mask_ptrs d15
1293 #define u_whole_low d16
1294 #define u_whole_high d17
1295 #define v_whole_low d18
1296 #define v_whole_high d19
1297 #define r_whole_low d20
1298 #define r_whole_high d21
1299 #define g_whole_low d22
1300 #define g_whole_high d23
1301 #define b_whole_low d24
1302 #define b_whole_high d25
1307 #define u_whole_8 d26
1308 #define v_whole_8 d27
1309 #define u_whole_8b d24
1310 #define r_whole_8 d24
1311 #define g_whole_8 d25
1313 #define uv_whole_8 q13
1314 #define uv_whole_8b q14
1316 #define dither_offsets q14
1317 #define texture_mask q15
1318 #define texture_mask_u d30
1319 #define texture_mask_v d31
1321 #define dither_offsets_short d28
1325 #define block_span q10
1330 #define draw_mask q1
1331 #define draw_mask_edge q13
1332 #define test_mask q0
1338 #define setup_blocks_texture_swizzled() \
1339 vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \
1340 vsli.u8 u_whole_8, v_whole_8, #4; \
1341 vsri.u8 v_whole_8, u_whole_8b, #4 \
1343 #define setup_blocks_texture_unswizzled() \
1346 #define setup_blocks_shaded_textured_builder(swizzling) \
1349 function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \
1350 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
1351 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1353 vld1.u32 { uvrg_dx }, [uvrg_dx_ptr, :128]; \
1354 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1356 cmp num_spans, #0; \
1359 stmdb sp!, { r4 - r11, r14 }; \
1360 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1362 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
1363 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1365 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [texture_mask_ptr, :16]; \
1366 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1368 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1369 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1371 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
1372 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1374 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1377 vmov.u8 fb_mask_ptrs, #0; \
1379 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
1380 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1382 ldrh y, [span_edge_data, #edge_data_y_offset]; \
1383 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
1385 cmp span_num_blocks, #0; \
1388 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
1389 add num_blocks, span_num_blocks, num_blocks; \
1391 cmp num_blocks, #MAX_BLOCKS; \
1395 ldr b, [span_b_offset]; \
1396 add fb_ptr, fb_ptr, y, lsl #11; \
1398 vdup.u32 v_left_x, left_x; \
1401 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
1402 add fb_ptr, fb_ptr, left_x, lsl #1; \
1404 mla b, b_dx, left_x, b; \
1405 and dither_shift, left_x, #0x03; \
1407 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
1408 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1410 mov dither_shift, dither_shift, lsl #3; \
1411 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1414 subs span_num_blocks, span_num_blocks, #1; \
1416 mov dither_row, dither_row, ror dither_shift; \
1417 mov b_dx4, b_dx, lsl #2; \
1419 vdup.u32 dither_offsets_short, dither_row; \
1420 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1422 vdup.u32 b_block, b; \
1423 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1425 vdup.u32 u_block, uv[0]; \
1426 mov b_dx8, b_dx, lsl #3; \
1428 vdup.u32 v_block, uv[1]; \
1429 vdup.u32 r_block, rg[0]; \
1430 vdup.u32 g_block, rg[1]; \
1432 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1434 vadd.u32 u_block, u_block, block_span; \
1435 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1437 vadd.u32 v_block, v_block, block_span; \
1438 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1440 vadd.u32 r_block, r_block, block_span; \
1441 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1443 vadd.u32 g_block, g_block, block_span; \
1444 vld1.u32 { block_span }, [block_span_ptr, :128]; \
1446 vadd.u32 b_block, b_block, block_span; \
1447 add block_ptr_b, block_ptr_a, #16; \
1449 vshrn.u32 u_whole_low, u_block, #16; \
1450 vshrn.u32 v_whole_low, v_block, #16; \
1451 vshrn.u32 r_whole_low, r_block, #16; \
1452 vshrn.u32 g_whole_low, g_block, #16; \
1454 vdup.u32 dx4, uv_dx4[0]; \
1455 vshrn.u32 b_whole_low, b_block, #16; \
1457 vaddhn.u32 u_whole_high, u_block, dx4; \
1458 vdup.u32 dx4, uv_dx4[1]; \
1460 vaddhn.u32 v_whole_high, v_block, dx4; \
1461 vdup.u32 dx4, rg_dx4[0]; \
1463 vaddhn.u32 r_whole_high, r_block, dx4; \
1464 vdup.u32 dx4, rg_dx4[1]; \
1466 vaddhn.u32 g_whole_high, g_block, dx4; \
1467 vdup.u32 dx4, b_dx4; \
1469 vaddhn.u32 b_whole_high, b_block, dx4; \
1470 vdup.u32 dx8, uv_dx8[0]; \
1472 vadd.u32 u_block, u_block, dx8; \
1473 vdup.u32 dx8, uv_dx8[1]; \
1475 vadd.u32 v_block, v_block, dx8; \
1476 vdup.u32 dx8, rg_dx8[0]; \
1478 vadd.u32 r_block, r_block, dx8; \
1479 vdup.u32 dx8, rg_dx8[1]; \
1481 vadd.u32 g_block, g_block, dx8; \
1482 vdup.u32 dx8, b_dx8; \
1484 vadd.u32 b_block, b_block, dx8; \
1485 vmovn.u16 u_whole_8, u_whole; \
1487 vmovn.u16 v_whole_8, v_whole; \
1489 vmovn.u16 b_whole_8, b_whole; \
1491 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1493 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1494 setup_blocks_texture_##swizzling(); \
1496 vmovn.u16 r_whole_8, r_whole; \
1500 vmovn.u16 g_whole_8, g_whole; \
1501 vshrn.u32 u_whole_low, u_block, #16; \
1503 vst2.u8 { u_whole_8, v_whole_8 }, [block_ptr_a, :128], c_32; \
1504 vshrn.u32 v_whole_low, v_block, #16; \
1506 vst1.u32 { r_whole_8, g_whole_8 }, [block_ptr_b, :128], c_32; \
1507 vshrn.u32 r_whole_low, r_block, #16; \
1509 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1510 vshrn.u32 g_whole_low, g_block, #16; \
1512 vdup.u32 dx4, uv_dx4[0]; \
1513 vshrn.u32 b_whole_low, b_block, #16; \
1515 vaddhn.u32 u_whole_high, u_block, dx4; \
1516 vdup.u32 dx4, uv_dx4[1]; \
1518 vaddhn.u32 v_whole_high, v_block, dx4; \
1519 vdup.u32 dx4, rg_dx4[0]; \
1521 vaddhn.u32 r_whole_high, r_block, dx4; \
1522 vdup.u32 dx4, rg_dx4[1]; \
1524 vaddhn.u32 g_whole_high, g_block, dx4; \
1525 vdup.u32 dx4, b_dx4; \
1527 vaddhn.u32 b_whole_high, b_block, dx4; \
1528 vdup.u32 dx8, uv_dx8[0]; \
1530 vadd.u32 u_block, u_block, dx8; \
1531 vdup.u32 dx8, uv_dx8[1]; \
1533 vadd.u32 v_block, v_block, dx8; \
1534 vdup.u32 dx8, rg_dx8[0]; \
1536 vadd.u32 r_block, r_block, dx8; \
1537 vdup.u32 dx8, rg_dx8[1]; \
1539 vadd.u32 g_block, g_block, dx8; \
1540 vdup.u32 dx8, b_dx8; \
1542 vadd.u32 b_block, b_block, dx8; \
1543 vmovn.u16 u_whole_8, u_whole; \
1545 add fb_ptr, fb_ptr, #16; \
1546 vmovn.u16 v_whole_8, v_whole; \
1548 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1549 vmovn.u16 b_whole_8, b_whole; \
1553 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1554 subs span_num_blocks, span_num_blocks, #1; \
1556 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1557 setup_blocks_texture_##swizzling(); \
1559 vmovn.u16 r_whole_8, r_whole; \
1563 vmovn.u16 g_whole_8, g_whole; \
1564 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
1566 vld1.u32 { test_mask }, [psx_gpu, :128]; \
1567 vdup.u8 draw_mask, right_mask; \
1569 vmov.u32 fb_mask_ptrs[0], right_mask; \
1570 vtst.u16 draw_mask, draw_mask, test_mask; \
1571 vzip.u8 u_whole_8, v_whole_8; \
1573 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1574 vst1.u32 { r_whole_8, g_whole_8 }, [block_ptr_b, :128], c_32; \
1575 vst1.u32 { uv_whole_8 }, [block_ptr_a, :128], c_32; \
1576 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1577 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1580 add span_uvrg_offset, span_uvrg_offset, #16; \
1581 add span_b_offset, span_b_offset, #4; \
1583 add span_edge_data, span_edge_data, #8; \
1584 subs num_spans, num_spans, #1; \
1586 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1589 ldmia sp!, { r4 - r11, pc }; \
1592 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1593 vpush { texture_mask }; \
1594 vpush { uvrg_dx4 }; \
1596 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
1597 bl flush_render_block_buffer; \
1598 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
1600 vpop { uvrg_dx4 }; \
1601 vpop { texture_mask }; \
1603 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1604 vmov.u8 fb_mask_ptrs, #0; \
1606 mov num_blocks, span_num_blocks; \
1607 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1611 setup_blocks_shaded_textured_builder(swizzled)
1612 setup_blocks_shaded_textured_builder(unswizzled)
1615 #define setup_blocks_unshaded_textured_builder(swizzling) \
1618 function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \
1619 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
1620 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1622 vld1.u32 { uvrg_dx }, [uvrg_dx_ptr, :128]; \
1623 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1625 cmp num_spans, #0; \
1628 stmdb sp!, { r4 - r11, r14 }; \
1629 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1631 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1633 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [texture_mask_ptr, :16]; \
1634 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1636 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1637 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1639 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1641 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1644 vmov.u8 fb_mask_ptrs, #0; \
1646 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
1647 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1649 ldrh y, [span_edge_data, #edge_data_y_offset]; \
1650 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
1652 cmp span_num_blocks, #0; \
1655 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
1656 add num_blocks, span_num_blocks, num_blocks; \
1658 cmp num_blocks, #MAX_BLOCKS; \
1662 add fb_ptr, fb_ptr, y, lsl #11; \
1664 vdup.u32 v_left_x, left_x; \
1667 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
1668 add fb_ptr, fb_ptr, left_x, lsl #1; \
1670 and dither_shift, left_x, #0x03; \
1672 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
1673 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1675 mov dither_shift, dither_shift, lsl #3; \
1676 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1679 subs span_num_blocks, span_num_blocks, #1; \
1681 mov dither_row, dither_row, ror dither_shift; \
1683 vdup.u32 dither_offsets_short, dither_row; \
1684 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1686 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1688 vdup.u32 u_block, uv[0]; \
1690 vdup.u32 v_block, uv[1]; \
1691 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1693 vadd.u32 u_block, u_block, block_span; \
1694 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1696 vadd.u32 v_block, v_block, block_span; \
1697 add block_ptr_b, block_ptr_a, #16; \
1699 vshrn.u32 u_whole_low, u_block, #16; \
1700 vshrn.u32 v_whole_low, v_block, #16; \
1702 vdup.u32 dx4, uv_dx4[0]; \
1704 vaddhn.u32 u_whole_high, u_block, dx4; \
1705 vdup.u32 dx4, uv_dx4[1]; \
1707 vaddhn.u32 v_whole_high, v_block, dx4; \
1708 vdup.u32 dx8, uv_dx8[0]; \
1710 vadd.u32 u_block, u_block, dx8; \
1711 vdup.u32 dx8, uv_dx8[1]; \
1713 vadd.u32 v_block, v_block, dx8; \
1714 vmovn.u16 u_whole_8, u_whole; \
1716 vmovn.u16 v_whole_8, v_whole; \
1719 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1721 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1722 setup_blocks_texture_##swizzling(); \
1727 vshrn.u32 u_whole_low, u_block, #16; \
1729 vst2.u8 { u_whole_8, v_whole_8 }, [block_ptr_a, :128], c_32; \
1730 vshrn.u32 v_whole_low, v_block, #16; \
1732 add block_ptr_b, block_ptr_b, #32; \
1733 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1735 vdup.u32 dx4, uv_dx4[0]; \
1736 vaddhn.u32 u_whole_high, u_block, dx4; \
1737 vdup.u32 dx4, uv_dx4[1]; \
1739 vaddhn.u32 v_whole_high, v_block, dx4; \
1740 vdup.u32 dx8, uv_dx8[0]; \
1742 vadd.u32 u_block, u_block, dx8; \
1743 vdup.u32 dx8, uv_dx8[1]; \
1745 vadd.u32 v_block, v_block, dx8; \
1746 vmovn.u16 u_whole_8, u_whole; \
1748 add fb_ptr, fb_ptr, #16; \
1749 vmovn.u16 v_whole_8, v_whole; \
1751 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1754 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1755 subs span_num_blocks, span_num_blocks, #1; \
1757 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1758 setup_blocks_texture_##swizzling(); \
1763 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
1765 vld1.u32 { test_mask }, [psx_gpu, :128]; \
1766 vdup.u8 draw_mask, right_mask; \
1768 vmov.u32 fb_mask_ptrs[0], right_mask; \
1769 vtst.u16 draw_mask, draw_mask, test_mask; \
1770 vzip.u8 u_whole_8, v_whole_8; \
1772 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1773 add block_ptr_b, block_ptr_b, #32; \
1774 vst1.u32 { uv_whole_8 }, [block_ptr_a, :128], c_32; \
1775 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1776 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1779 add span_uvrg_offset, span_uvrg_offset, #16; \
1780 add span_edge_data, span_edge_data, #8; \
1781 subs num_spans, num_spans, #1; \
1783 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1786 ldmia sp!, { r4 - r11, pc }; \
1789 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1790 vpush { texture_mask }; \
1791 vpush { uvrg_dx4 }; \
1793 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
1794 bl flush_render_block_buffer; \
1795 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
1797 vpop { uvrg_dx4 }; \
1798 vpop { texture_mask }; \
1800 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1801 vmov.u8 fb_mask_ptrs, #0; \
1803 mov num_blocks, span_num_blocks; \
1804 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1808 setup_blocks_unshaded_textured_builder(swizzled)
1809 setup_blocks_unshaded_textured_builder(unswizzled)
1814 function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect)
1815 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]
1816 veor.u32 draw_mask, draw_mask, draw_mask
1821 stmdb sp!, { r4 - r11, r14 }
1822 vld1.u32 { test_mask }, [psx_gpu, :128]
1824 ldr color, [psx_gpu, #psx_gpu_triangle_color_offset]
1826 ubfx color_r, color, #3, #5
1827 ubfx color_g, color, #11, #5
1828 ubfx color_b, color, #19, #5
1830 orr color, color_r, color_b, lsl #10
1831 orr color, color, color_g, lsl #5
1833 vdup.u16 colors, color
1835 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
1836 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1838 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1839 add block_ptr_a, block_ptr_a, num_blocks, lsl #6
1842 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]
1843 ldrh y, [span_edge_data, #edge_data_y_offset]
1845 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
1847 cmp span_num_blocks, #0
1850 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]
1851 add num_blocks, span_num_blocks, num_blocks
1853 cmp num_blocks, #MAX_BLOCKS
1857 add fb_ptr, fb_ptr, y, lsl #11
1860 add fb_ptr, fb_ptr, left_x, lsl #1
1863 subs span_num_blocks, span_num_blocks, #1
1865 add block_ptr_b, block_ptr_a, #16
1868 vmov.u32 fb_mask_ptrs[1], fb_ptr
1872 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_32
1873 vst1.u32 { colors }, [block_ptr_b, :128], c_32
1874 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32
1876 add fb_ptr, fb_ptr, #16
1877 add block_ptr_b, block_ptr_b, #32
1881 vmov.u32 fb_mask_ptrs[1], fb_ptr
1882 subs span_num_blocks, span_num_blocks, #1
1887 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]
1889 vdup.u8 draw_mask_edge, right_mask
1890 vtst.u16 draw_mask_edge, draw_mask_edge, test_mask
1892 vst1.u32 { colors }, [block_ptr_b, :128], c_32
1893 vst1.u32 { draw_mask_edge }, [block_ptr_a, :128], c_32
1894 add block_ptr_b, block_ptr_b, #32
1895 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32
1898 add span_edge_data, span_edge_data, #8
1899 subs num_spans, num_spans, #1
1901 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
1904 ldmia sp!, { r4 - r11, pc }
1909 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
1910 bl flush_render_block_buffer
1911 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
1915 vld1.u32 { test_mask }, [psx_gpu, :128]
1916 veor.u32 draw_mask, draw_mask, draw_mask
1918 mov num_blocks, span_num_blocks
1919 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1923 #define mask_msb_scalar r14
1925 #define msb_mask q15
1927 #define pixels_low d16
1929 #define msb_mask_low d30
1930 #define msb_mask_high d31
1935 function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct)
1936 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]
1941 stmdb sp!, { r4 - r11, r14 }
1943 ldr color, [psx_gpu, #psx_gpu_triangle_color_offset]
1945 ubfx color_r, color, #3, #5
1946 ubfx color_g, color, #11, #5
1948 ldrh mask_msb_scalar, [psx_gpu, #psx_gpu_mask_msb_offset]
1949 ubfx color_b, color, #19, #5
1951 orr color, color_r, color_b, lsl #10
1952 orr color, color, color_g, lsl #5
1953 orr color, color, mask_msb_scalar
1955 vdup.u16 colors, color
1957 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1958 orr color, color, color, lsl #16
1962 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]
1963 ldrh y, [span_edge_data, #edge_data_y_offset]
1965 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
1967 cmp span_num_blocks, #0
1970 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]
1972 add fb_ptr, fb_ptr, y, lsl #11
1973 subs span_num_blocks, span_num_blocks, #1
1975 add fb_ptr, fb_ptr, left_x, lsl #1
1979 vst1.u32 { colors }, [fb_ptr]!
1980 subs span_num_blocks, span_num_blocks, #1
1985 ldrb right_mask, [span_edge_data, #edge_data_right_mask_offset]
1987 cmp right_mask, #0x0
1990 tst right_mask, #0xF
1991 streq color, [fb_ptr], #4
1992 moveq right_mask, right_mask, lsr #4
1993 streq color, [fb_ptr], #4
1995 tst right_mask, #0x3
1996 streq color, [fb_ptr], #4
1997 moveq right_mask, right_mask, lsr #2
1999 tst right_mask, #0x1
2000 strheq color, [fb_ptr]
2003 add span_edge_data, span_edge_data, #8
2004 subs num_spans, num_spans, #1
2007 ldmia sp!, { r4 - r11, pc }
2010 vst1.u32 { colors }, [fb_ptr]
2017 #define rg_dx_ptr r2
2035 #undef dither_offsets
2055 #define r_whole_low d6
2056 #define r_whole_high d7
2057 #define g_whole_low d8
2058 #define g_whole_high d9
2059 #define b_whole_low d10
2060 #define b_whole_high d11
2062 #define gb_whole_8 q6
2064 #define g_whole_8 d12
2065 #define b_whole_8 d13
2067 #define r_whole_8 d14
2079 #define block_span q5
2087 #define d128_0x7 q13
2091 #define dither_offsets q14
2092 #define draw_mask q15
2094 #define dither_offsets_low d28
2097 #define test_mask q10
2100 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2101 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2102 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2104 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2105 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2106 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2108 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2110 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2113 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2116 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2117 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
2118 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2120 vld1.u32 { rg_dx }, [rg_dx_ptr, :64]; \
2122 cmp num_spans, #0; \
2125 stmdb sp!, { r4 - r11, r14 }; \
2126 vshl.u32 rg_dx4, rg_dx, #2; \
2128 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
2129 vshl.u32 rg_dx8, rg_dx, #3; \
2131 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2133 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
2134 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2136 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2137 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2139 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2140 vmov.u8 d64_1, #1; \
2142 vmov.u8 d128_4, #4; \
2143 vmov.u8 d64_128, #128; \
2145 vmov.u8 d128_0x7, #0x7; \
2148 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
2149 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2151 ldrh y, [span_edge_data, #edge_data_y_offset]; \
2152 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
2154 cmp span_num_blocks, #0; \
2157 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
2158 add num_blocks, span_num_blocks, num_blocks; \
2160 cmp num_blocks, #MAX_BLOCKS; \
2164 ldr b, [span_b_offset]; \
2165 add fb_ptr, fb_ptr, y, lsl #11; \
2167 vdup.u32 v_left_x, left_x; \
2170 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
2171 add fb_ptr, fb_ptr, left_x, lsl #1; \
2173 mla b, b_dx, left_x, b; \
2174 and dither_shift, left_x, #0x03; \
2176 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
2177 vshr.u32 rg_dx, rg_dx4, #2; \
2179 mov dither_shift, dither_shift, lsl #3; \
2180 vmla.u32 rg, rg_dx, v_left_x; \
2183 subs span_num_blocks, span_num_blocks, #1; \
2185 mov dither_row, dither_row, ror dither_shift; \
2186 mov b_dx4, b_dx, lsl #2; \
2188 vdup.u32 dither_offsets, dither_row; \
2189 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2191 vdup.u32 b_block, b; \
2192 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2194 mov b_dx8, b_dx, lsl #3; \
2195 vdup.u32 r_block, rg[0]; \
2196 vdup.u32 g_block, rg[1]; \
2198 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2200 vadd.u32 r_block, r_block, block_span; \
2201 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2203 vadd.u32 g_block, g_block, block_span; \
2204 vld1.u32 { block_span }, [block_span_ptr, :128]; \
2206 vadd.u32 b_block, b_block, block_span; \
2207 add block_ptr_b, block_ptr_a, #16; \
2209 vshrn.u32 r_whole_low, r_block, #16; \
2210 vshrn.u32 g_whole_low, g_block, #16; \
2211 vshrn.u32 b_whole_low, b_block, #16; \
2212 vdup.u32 dx4, rg_dx4[0]; \
2214 vaddhn.u32 r_whole_high, r_block, dx4; \
2215 vdup.u32 dx4, rg_dx4[1]; \
2217 vaddhn.u32 g_whole_high, g_block, dx4; \
2218 vdup.u32 dx4, b_dx4; \
2220 vaddhn.u32 b_whole_high, b_block, dx4; \
2221 vdup.u32 dx8, rg_dx8[0]; \
2223 vadd.u32 r_block, r_block, dx8; \
2224 vdup.u32 dx8, rg_dx8[1]; \
2226 vadd.u32 g_block, g_block, dx8; \
2227 vdup.u32 dx8, b_dx8; \
2229 vadd.u32 b_block, b_block, dx8; \
2231 vmovn.u16 r_whole_8, r_whole; \
2232 vmovn.u16 g_whole_8, g_whole; \
2233 vmovn.u16 b_whole_8, b_whole; \
2236 veor.u32 draw_mask, draw_mask, draw_mask; \
2239 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2240 vshrn.u32 r_whole_low, r_block, #16; \
2242 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2243 vshrn.u32 g_whole_low, g_block, #16; \
2245 vshrn.u32 b_whole_low, b_block, #16; \
2246 str fb_ptr, [block_ptr_a, #44]; \
2248 vdup.u32 dx4, rg_dx4[0]; \
2249 vshr.u8 r_whole_8, r_whole_8, #3; \
2250 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2252 vaddhn.u32 r_whole_high, r_block, dx4; \
2253 vdup.u32 dx4, rg_dx4[1]; \
2255 vaddhn.u32 g_whole_high, g_block, dx4; \
2256 vdup.u32 dx4, b_dx4; \
2258 vaddhn.u32 b_whole_high, b_block, dx4; \
2259 vdup.u32 dx8, rg_dx8[0]; \
2261 vmull.u8 pixels, r_whole_8, d64_1; \
2262 vmlal.u8 pixels, g_whole_8, d64_4; \
2263 vmlal.u8 pixels, b_whole_8, d64_128; \
2265 vadd.u32 r_block, r_block, dx8; \
2266 vdup.u32 dx8, rg_dx8[1]; \
2268 vadd.u32 g_block, g_block, dx8; \
2269 vdup.u32 dx8, b_dx8; \
2271 vadd.u32 b_block, b_block, dx8; \
2272 add fb_ptr, fb_ptr, #16; \
2274 vmovn.u16 r_whole_8, r_whole; \
2275 vmovn.u16 g_whole_8, g_whole; \
2276 vmovn.u16 b_whole_8, b_whole; \
2278 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_64; \
2279 vst1.u32 { pixels }, [block_ptr_b, :128], c_64; \
2283 subs span_num_blocks, span_num_blocks, #1; \
2287 str fb_ptr, [block_ptr_a, #44]; \
2288 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2290 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
2291 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2293 vshr.u8 r_whole_8, r_whole_8, #3; \
2294 vdup.u8 draw_mask, right_mask; \
2296 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2297 vld1.u32 { test_mask }, [psx_gpu, :128]; \
2299 vtst.u16 draw_mask, draw_mask, test_mask; \
2301 vmull.u8 pixels, r_whole_8, d64_1; \
2302 vmlal.u8 pixels, g_whole_8, d64_4; \
2303 vmlal.u8 pixels, b_whole_8, d64_128; \
2305 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_64; \
2306 vst1.u32 { pixels }, [block_ptr_b, :128], c_64; \
2309 add span_uvrg_offset, span_uvrg_offset, #16; \
2310 add span_b_offset, span_b_offset, #4; \
2312 add span_edge_data, span_edge_data, #8; \
2313 subs num_spans, num_spans, #1; \
2315 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
2318 ldmia sp!, { r4 - r11, pc }; \
2321 /* TODO: Load from psx_gpu instead of saving/restoring these */\
2324 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
2325 bl flush_render_block_buffer; \
2326 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
2330 vmov.u8 d64_1, #1; \
2331 vmov.u8 d128_4, #4; \
2332 vmov.u8 d64_128, #128; \
2333 vmov.u8 d128_0x7, #0x7; \
2335 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2337 mov num_blocks, span_num_blocks; \
2338 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2342 setup_blocks_shaded_untextured_indirect_builder(undithered)
2343 setup_blocks_shaded_untextured_indirect_builder(dithered)
2348 #define mask_msb_ptr r14
2350 #define draw_mask q0
2351 #define pixels_low d16
2352 #define pixels_high d17
2356 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2359 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2360 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
2361 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2363 vld1.u32 { rg_dx }, [rg_dx_ptr, :64]; \
2365 cmp num_spans, #0; \
2368 stmdb sp!, { r4 - r11, r14 }; \
2369 vshl.u32 rg_dx4, rg_dx, #2; \
2371 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
2372 vshl.u32 rg_dx8, rg_dx, #3; \
2374 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2375 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2377 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2378 vmov.u8 d64_1, #1; \
2380 vmov.u8 d128_4, #4; \
2381 vmov.u8 d64_128, #128; \
2383 vmov.u8 d128_0x7, #0x7; \
2384 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2385 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
2388 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
2389 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2391 ldrh y, [span_edge_data, #edge_data_y_offset]; \
2392 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
2394 cmp span_num_blocks, #0; \
2397 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
2398 add fb_ptr, fb_ptr, y, lsl #11; \
2400 ldr b, [span_b_offset]; \
2401 vdup.u32 v_left_x, left_x; \
2404 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
2405 add fb_ptr, fb_ptr, left_x, lsl #1; \
2407 mla b, b_dx, left_x, b; \
2408 and dither_shift, left_x, #0x03; \
2410 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
2411 vshr.u32 rg_dx, rg_dx4, #2; \
2413 mov dither_shift, dither_shift, lsl #3; \
2414 vmla.u32 rg, rg_dx, v_left_x; \
2416 subs span_num_blocks, span_num_blocks, #1; \
2418 mov dither_row, dither_row, ror dither_shift; \
2419 mov b_dx4, b_dx, lsl #2; \
2421 vdup.u32 dither_offsets, dither_row; \
2422 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2424 vdup.u32 b_block, b; \
2425 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2427 mov b_dx8, b_dx, lsl #3; \
2428 vdup.u32 r_block, rg[0]; \
2429 vdup.u32 g_block, rg[1]; \
2431 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2433 vadd.u32 r_block, r_block, block_span; \
2434 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2436 vadd.u32 g_block, g_block, block_span; \
2437 vld1.u32 { block_span }, [block_span_ptr, :128]; \
2439 vadd.u32 b_block, b_block, block_span; \
2440 add block_ptr_b, block_ptr_a, #16; \
2442 vshrn.u32 r_whole_low, r_block, #16; \
2443 vshrn.u32 g_whole_low, g_block, #16; \
2444 vshrn.u32 b_whole_low, b_block, #16; \
2445 vdup.u32 dx4, rg_dx4[0]; \
2447 vaddhn.u32 r_whole_high, r_block, dx4; \
2448 vdup.u32 dx4, rg_dx4[1]; \
2450 vaddhn.u32 g_whole_high, g_block, dx4; \
2451 vdup.u32 dx4, b_dx4; \
2453 vaddhn.u32 b_whole_high, b_block, dx4; \
2454 vdup.u32 dx8, rg_dx8[0]; \
2456 vadd.u32 r_block, r_block, dx8; \
2457 vdup.u32 dx8, rg_dx8[1]; \
2459 vadd.u32 g_block, g_block, dx8; \
2460 vdup.u32 dx8, b_dx8; \
2462 vadd.u32 b_block, b_block, dx8; \
2464 vmovn.u16 r_whole_8, r_whole; \
2465 vmovn.u16 g_whole_8, g_whole; \
2466 vmovn.u16 b_whole_8, b_whole; \
2471 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2472 vshrn.u32 r_whole_low, r_block, #16; \
2474 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2475 vshrn.u32 g_whole_low, g_block, #16; \
2477 vshrn.u32 b_whole_low, b_block, #16; \
2479 vdup.u32 dx4, rg_dx4[0]; \
2480 vshr.u8 r_whole_8, r_whole_8, #3; \
2481 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2483 vaddhn.u32 r_whole_high, r_block, dx4; \
2484 vdup.u32 dx4, rg_dx4[1]; \
2486 vmov pixels, msb_mask; \
2487 vaddhn.u32 g_whole_high, g_block, dx4; \
2488 vdup.u32 dx4, b_dx4; \
2490 vaddhn.u32 b_whole_high, b_block, dx4; \
2491 vdup.u32 dx8, rg_dx8[0]; \
2493 vmlal.u8 pixels, r_whole_8, d64_1; \
2494 vmlal.u8 pixels, g_whole_8, d64_4; \
2495 vmlal.u8 pixels, b_whole_8, d64_128; \
2497 vadd.u32 r_block, r_block, dx8; \
2498 vdup.u32 dx8, rg_dx8[1]; \
2500 vadd.u32 g_block, g_block, dx8; \
2501 vdup.u32 dx8, b_dx8; \
2503 vadd.u32 b_block, b_block, dx8; \
2505 vmovn.u16 r_whole_8, r_whole; \
2506 vmovn.u16 g_whole_8, g_whole; \
2507 vmovn.u16 b_whole_8, b_whole; \
2509 vst1.u32 { pixels }, [fb_ptr]!; \
2510 subs span_num_blocks, span_num_blocks, #1; \
2514 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2516 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
2517 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2519 vshr.u8 r_whole_8, r_whole_8, #3; \
2520 rbit right_mask, right_mask; \
2521 vmov pixels, msb_mask; \
2522 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2523 clz right_mask, right_mask; \
2525 vmlal.u8 pixels, r_whole_8, d64_1; \
2526 vmlal.u8 pixels, g_whole_8, d64_4; \
2527 vmlal.u8 pixels, b_whole_8, d64_128; \
2529 JT_OP_REL(100f, right_mask, temp); \
2530 JT_OP(ldr pc, [pc, right_mask, lsl #2]); \
2534 .word JTE(100b, 4f); \
2535 .word JTE(100b, 5f); \
2536 .word JTE(100b, 6f); \
2537 .word JTE(100b, 7f); \
2538 .word JTE(100b, 8f); \
2539 .word JTE(100b, 9f); \
2540 .word JTE(100b, 10f); \
2541 .word JTE(100b, 11f); \
2544 vst1.u16 { pixels_low[0] }, [fb_ptr]; \
2548 vst1.u32 { pixels_low[0] }, [fb_ptr]; \
2552 vst1.u32 { pixels_low[0] }, [fb_ptr]!; \
2553 vst1.u16 { pixels_low[2] }, [fb_ptr]; \
2557 vst1.u32 { pixels_low }, [fb_ptr]; \
2561 vst1.u32 { pixels_low }, [fb_ptr]!; \
2562 vst1.u16 { pixels_high[0] }, [fb_ptr]; \
2566 vst1.u32 { pixels_low }, [fb_ptr]!; \
2567 vst1.u32 { pixels_high[0] }, [fb_ptr]!; \
2571 vst1.u32 { pixels_low }, [fb_ptr]!; \
2572 vst1.u32 { pixels_high[0] }, [fb_ptr]!; \
2573 vst1.u16 { pixels_high[2] }, [fb_ptr]; \
2577 vst1.u32 { pixels }, [fb_ptr]; \
2581 add span_uvrg_offset, span_uvrg_offset, #16; \
2582 add span_b_offset, span_b_offset, #4; \
2584 add span_edge_data, span_edge_data, #8; \
2585 subs num_spans, num_spans, #1; \
2589 ldmia sp!, { r4 - r11, pc } \
2591 setup_blocks_shaded_untextured_direct_builder(undithered)
2592 setup_blocks_shaded_untextured_direct_builder(dithered)
2601 #define block_ptr r1
2602 #define num_blocks r2
2615 #define texture_ptr r11
2629 #define pixels_d r10
2633 #define clut_ptr r12
2634 #define current_texture_mask r5
2635 #define dirty_textures_mask r6
2639 #define clut_low_a d2
2640 #define clut_low_b d3
2641 #define clut_high_a d4
2642 #define clut_high_b d5
2647 #define texels_low d6
2648 #define texels_high d7
2652 function(texture_blocks_untextured)
2658 function(texture_blocks_4bpp)
2659 stmdb sp!, { r3 - r11, r14 }
2660 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2662 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2663 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2665 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
2666 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]
2668 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
2669 vuzp.u8 clut_a, clut_b
2671 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
2672 tst dirty_textures_mask, current_texture_mask
2678 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2680 uxtah uv_0, texture_ptr, uv_01
2681 uxtah uv_1, texture_ptr, uv_01, ror #16
2683 uxtah uv_2, texture_ptr, uv_23
2684 uxtah uv_3, texture_ptr, uv_23, ror #16
2686 uxtah uv_4, texture_ptr, uv_45
2687 ldrb pixel_0, [uv_0]
2689 uxtah uv_5, texture_ptr, uv_45, ror #16
2690 ldrb pixel_1, [uv_1]
2692 uxtah uv_6, texture_ptr, uv_67
2693 ldrb pixel_2, [uv_2]
2695 uxtah uv_7, texture_ptr, uv_67, ror #16
2696 ldrb pixel_3, [uv_3]
2698 ldrb pixel_4, [uv_4]
2699 subs num_blocks, num_blocks, #1
2701 ldrb pixel_5, [uv_5]
2702 orr pixels_a, pixel_0, pixel_1, lsl #8
2704 ldrb pixel_6, [uv_6]
2705 orr pixels_b, pixel_4, pixel_5, lsl #8
2707 ldrb pixel_7, [uv_7]
2708 orr pixels_a, pixels_a, pixel_2, lsl #16
2710 orr pixels_b, pixels_b, pixel_6, lsl #16
2711 orr pixels_a, pixels_a, pixel_3, lsl #24
2713 orr pixels_b, pixels_b, pixel_7, lsl #24
2714 vmov texels, pixels_a, pixels_b
2716 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2717 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2719 vst2.u8 { texels_low, texels_high }, [block_ptr, :128], c_64
2722 ldmia sp!, { r3 - r11, pc }
2725 stmdb sp!, { r1 - r2 }
2726 bl update_texture_4bpp_cache
2729 ldmia sp!, { r1 - r2 }
2735 function(texture_blocks_8bpp)
2736 stmdb sp!, { r3 - r11, r14 }
2737 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2739 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2740 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2742 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
2743 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
2745 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset]
2746 tst dirty_textures_mask, current_texture_mask
2752 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2754 uxtah uv_0, texture_ptr, uv_01
2755 uxtah uv_1, texture_ptr, uv_01, ror #16
2757 uxtah uv_2, texture_ptr, uv_23
2758 uxtah uv_3, texture_ptr, uv_23, ror #16
2760 uxtah uv_4, texture_ptr, uv_45
2761 ldrb pixel_0, [uv_0]
2763 uxtah uv_5, texture_ptr, uv_45, ror #16
2764 ldrb pixel_1, [uv_1]
2766 uxtah uv_6, texture_ptr, uv_67
2767 ldrb pixel_2, [uv_2]
2769 uxtah uv_7, texture_ptr, uv_67, ror #16
2770 ldrb pixel_3, [uv_3]
2772 ldrb pixel_4, [uv_4]
2773 add pixel_0, pixel_0, pixel_0
2775 ldrb pixel_5, [uv_5]
2776 add pixel_1, pixel_1, pixel_1
2778 ldrb pixel_6, [uv_6]
2779 add pixel_2, pixel_2, pixel_2
2781 ldrb pixel_7, [uv_7]
2782 add pixel_3, pixel_3, pixel_3
2784 ldrh pixel_0, [clut_ptr, pixel_0]
2785 add pixel_4, pixel_4, pixel_4
2787 ldrh pixel_1, [clut_ptr, pixel_1]
2788 add pixel_5, pixel_5, pixel_5
2790 ldrh pixel_2, [clut_ptr, pixel_2]
2791 add pixel_6, pixel_6, pixel_6
2793 ldrh pixel_3, [clut_ptr, pixel_3]
2794 add pixel_7, pixel_7, pixel_7
2796 ldrh pixel_4, [clut_ptr, pixel_4]
2797 orr pixels_a, pixel_0, pixel_1, lsl #16
2799 ldrh pixel_5, [clut_ptr, pixel_5]
2800 orr pixels_c, pixel_2, pixel_3, lsl #16
2802 ldrh pixel_6, [clut_ptr, pixel_6]
2803 subs num_blocks, num_blocks, #1
2805 ldrh pixel_7, [clut_ptr, pixel_7]
2806 orr pixels_b, pixel_4, pixel_5, lsl #16
2808 orr pixels_d, pixel_6, pixel_7, lsl #16
2809 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2811 add block_ptr, block_ptr, #64
2814 ldmia sp!, { r3 - r11, pc }
2817 stmdb sp!, { r1 - r2, EXTRA_UNSAVED_REGS r12 }
2819 bl update_texture_8bpp_cache
2821 ldmia sp!, { r1 - r2, EXTRA_UNSAVED_REGS r12 }
2851 #define block_ptr r1
2852 #define num_blocks r2
2896 #define texture_ptr r12
2901 function(texture_blocks_16bpp)
2902 stmdb sp!, { r3 - r11, r14 }
2903 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2905 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2906 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2909 ldrh uv_0, [block_ptr]
2910 subs num_blocks, num_blocks, #1
2912 ldrh uv_1, [block_ptr, #2]
2914 and v_0, uv_0, #0xFF00
2915 and v_1, uv_1, #0xFF00
2917 and u_0, uv_0, #0xFF
2918 and u_1, uv_1, #0xFF
2920 add uv_0, u_0, v_0, lsl #2
2921 ldrh uv_2, [block_ptr, #4]
2923 add uv_1, u_1, v_1, lsl #2
2924 ldrh uv_3, [block_ptr, #6]
2926 add uv_0, uv_0, uv_0
2927 add uv_1, uv_1, uv_1
2929 and v_2, uv_2, #0xFF00
2930 and v_3, uv_3, #0xFF00
2932 and u_2, uv_2, #0xFF
2933 and u_3, uv_3, #0xFF
2935 add uv_2, u_2, v_2, lsl #2
2936 ldrh uv_4, [block_ptr, #8]
2938 add uv_3, u_3, v_3, lsl #2
2939 ldrh uv_5, [block_ptr, #10]
2941 add uv_2, uv_2, uv_2
2942 add uv_3, uv_3, uv_3
2944 and v_4, uv_4, #0xFF00
2945 and v_5, uv_5, #0xFF00
2947 and u_4, uv_4, #0xFF
2948 and u_5, uv_5, #0xFF
2950 add uv_4, u_4, v_4, lsl #2
2951 ldrh uv_6, [block_ptr, #12]
2953 add uv_5, u_5, v_5, lsl #2
2954 ldrh uv_7, [block_ptr, #14]
2956 add uv_4, uv_4, uv_4
2957 ldrh pixel_0, [texture_ptr, uv_0]
2959 add uv_5, uv_5, uv_5
2960 ldrh pixel_1, [texture_ptr, uv_1]
2962 and v_6, uv_6, #0xFF00
2963 ldrh pixel_2, [texture_ptr, uv_2]
2965 and v_7, uv_7, #0xFF00
2966 ldrh pixel_3, [texture_ptr, uv_3]
2968 and u_6, uv_6, #0xFF
2969 ldrh pixel_4, [texture_ptr, uv_4]
2971 and u_7, uv_7, #0xFF
2972 ldrh pixel_5, [texture_ptr, uv_5]
2974 add uv_6, u_6, v_6, lsl #2
2975 add uv_7, u_7, v_7, lsl #2
2977 add uv_6, uv_6, uv_6
2978 add uv_7, uv_7, uv_7
2980 orr pixels_a, pixel_0, pixel_1, lsl #16
2981 orr pixels_b, pixel_2, pixel_3, lsl #16
2983 ldrh pixel_6, [texture_ptr, uv_6]
2984 orr pixels_c, pixel_4, pixel_5, lsl #16
2986 ldrh pixel_7, [texture_ptr, uv_7]
2987 orr pixels_d, pixel_6, pixel_7, lsl #16
2989 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
2990 add block_ptr, block_ptr, #64
2994 ldmia sp!, { r3 - r11, pc }
3009 #undef msb_mask_high
3017 #define num_blocks r1
3018 #define color_ptr r2
3019 #define colors_scalar r2
3020 #define colors_scalar_compare r3
3021 #define mask_msb_ptr r2
3023 #define block_ptr_load_a r0
3024 #define block_ptr_store r3
3025 #define block_ptr_load_b r12
3030 #define draw_mask_bits_scalar r5
3032 #define d128_0x07 q0
3033 #define d128_0x1F q1
3034 #define d128_0x8000 q2
3035 #define test_mask q3
3037 #define colors_rg q5
3038 #define colors_b_dm_bits q6
3039 #define texels_rg q7
3042 #define pixels_b q10
3044 #define zero_mask q4
3045 #define draw_mask q12
3046 #define msb_mask q13
3048 #define fb_pixels q8
3050 #define pixels_gb_low q9
3052 #define colors_r d10
3053 #define colors_g d11
3054 #define colors_b d12
3055 #define draw_mask_bits d13
3056 #define texels_r d14
3057 #define texels_g d15
3058 #define pixels_r_low d16
3059 #define pixels_g_low d18
3060 #define pixels_b_low d19
3061 #define msb_mask_low d26
3062 #define msb_mask_high d27
3067 #define texels_b d31
3069 #define shade_blocks_textured_modulated_prologue_indirect() \
3071 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3073 #define shade_blocks_textured_modulated_prologue_direct() \
3074 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3075 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16] \
3078 #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \
3080 #define shade_blocks_textured_false_modulation_check_undithered(target) \
3081 ldr colors_scalar, [psx_gpu, #psx_gpu_triangle_color_offset]; \
3082 movw colors_scalar_compare, #0x8080; \
3084 movt colors_scalar_compare, #0x80; \
3085 cmp colors_scalar, colors_scalar_compare; \
3086 beq shade_blocks_textured_unmodulated_##target \
3088 #define shade_blocks_textured_false_modulation_check_dithered(target) \
3090 #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \
3091 shade_blocks_textured_false_modulation_check_##dithering(target); \
3092 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3093 vld1.u32 { colors_r[] }, [color_ptr, :32]; \
3094 vdup.u8 colors_g, colors_r[1]; \
3095 vdup.u8 colors_b, colors_r[2]; \
3096 vdup.u8 colors_r, colors_r[0] \
3099 #define shade_blocks_textured_modulated_load_dithered(target) \
3100 vld1.u32 { target }, [block_ptr_load_b, :128] \
3102 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3103 vld1.u32 { target }, [block_ptr_load_b, :128], c_32 \
3105 #define shade_blocks_textured_modulated_load_undithered(target) \
3107 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3108 add block_ptr_load_b, block_ptr_load_b, #32 \
3110 #define shade_blocks_textured_modulate_dithered(channel) \
3111 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3113 #define shade_blocks_textured_modulate_undithered(channel) \
3114 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3117 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3118 vst1.u32 { draw_mask }, [block_ptr_store, :128]! \
3120 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3121 ldr fb_ptr, [block_ptr_load_b, #(offset - 64)]; \
3122 vld1.u32 { fb_pixels }, [fb_ptr]; \
3123 vbit.u16 pixels, fb_pixels, draw_mask \
3125 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3126 vst1.u32 { pixels }, [block_ptr_store, :128], c_48 \
3128 #define shade_blocks_textured_modulated_store_pixels_direct() \
3129 vst1.u32 { pixels }, [fb_ptr] \
3132 #define shade_blocks_textured_modulated_load_rg_shaded() \
3133 vld1.u32 { colors_r, colors_g }, [block_ptr_load_b, :128], c_32 \
3135 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3136 add block_ptr_load_b, block_ptr_load_b, #32 \
3138 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3139 vld1.u32 { colors_b, draw_mask_bits }, [block_ptr_load_a, :128], c_32 \
3141 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3142 ldr draw_mask_bits_scalar, [block_ptr_load_a, #8]; \
3143 add block_ptr_load_a, block_ptr_load_a, #32 \
3145 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3146 vdup.u16 draw_mask, draw_mask_bits[0] \
3148 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3149 vdup.u16 draw_mask, draw_mask_bits_scalar \
3152 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3154 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3155 vorr.u16 pixels, pixels, msb_mask \
3158 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3161 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3162 shade_blocks_textured_modulated_prologue_##shading(dithering, target); \
3163 stmdb sp!, { r4 - r5, lr }; \
3164 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3166 vld1.u32 { test_mask }, [psx_gpu, :128]; \
3168 shade_blocks_textured_modulated_prologue_##target(); \
3170 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3173 add block_ptr_load_b, block_ptr_load_a, #16; \
3174 vmov.u8 d64_1, #1; \
3175 vmov.u8 d64_4, #4; \
3176 vmov.u8 d64_128, #128; \
3178 vld1.u32 { texels }, [block_ptr_load_a, :128], c_32; \
3179 vmov.u8 d128_0x07, #0x07; \
3181 shade_blocks_textured_modulated_load_rg_##shading(); \
3182 vmov.u8 d128_0x1F, #0x1F; \
3184 shade_blocks_textured_modulated_load_bdm_##shading(); \
3185 vmov.u16 d128_0x8000, #0x8000; \
3187 vmovn.u16 texels_r, texels; \
3188 vshrn.u16 texels_g, texels, #5; \
3190 vshrn.u16 texels_b, texels, #7; \
3191 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3193 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3194 vtst.u16 draw_mask, draw_mask, test_mask; \
3196 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3197 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3199 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3200 vshr.u8 texels_b, texels_b, #3; \
3202 shade_blocks_textured_modulate_##dithering(r); \
3203 shade_blocks_textured_modulate_##dithering(g); \
3204 shade_blocks_textured_modulate_##dithering(b); \
3206 vand.u16 pixels, texels, d128_0x8000; \
3207 vceq.u16 zero_mask, texels, #0; \
3209 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3210 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3211 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3213 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3214 vorr.u16 draw_mask, draw_mask, zero_mask; \
3215 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3216 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3218 subs num_blocks, num_blocks, #1; \
3224 vld1.u32 { texels }, [block_ptr_load_a, :128], c_32; \
3225 shade_blocks_textured_modulated_load_rg_##shading(); \
3226 vshrn.u16 texels_g, texels, #5; \
3228 shade_blocks_textured_modulated_load_bdm_##shading(); \
3229 vshrn.u16 texels_b, texels, #7; \
3231 pld [block_ptr_load_a]; \
3232 vmovn.u16 texels_r, texels; \
3233 vmlal.u8 pixels, pixels_r_low, d64_1; \
3235 vmlal.u8 pixels, pixels_g_low, d64_4; \
3236 vmlal.u8 pixels, pixels_b_low, d64_128; \
3237 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3239 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3240 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3242 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3243 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3245 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3246 vtst.u16 draw_mask, draw_mask, test_mask; \
3248 shade_blocks_textured_modulated_store_pixels_##target(); \
3249 vshr.u8 texels_b, texels_b, #3; \
3251 shade_blocks_textured_modulate_##dithering(r); \
3252 shade_blocks_textured_modulate_##dithering(g); \
3253 shade_blocks_textured_modulate_##dithering(b); \
3255 vand.u16 pixels, texels, d128_0x8000; \
3256 vceq.u16 zero_mask, texels, #0; \
3258 subs num_blocks, num_blocks, #1; \
3260 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3261 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3262 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3264 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3265 vorr.u16 draw_mask, draw_mask, zero_mask; \
3266 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3267 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3272 vmlal.u8 pixels, pixels_r_low, d64_1; \
3273 vmlal.u8 pixels, pixels_g_low, d64_4; \
3274 vmlal.u8 pixels, pixels_b_low, d64_128; \
3276 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3277 shade_blocks_textured_modulated_store_pixels_##target(); \
3279 ldmia sp!, { r4 - r5, pc } \
3282 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3283 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3284 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3285 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3287 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3288 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3289 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3290 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3308 #undef msb_mask_high
3311 #define num_blocks r1
3312 #define mask_msb_ptr r2
3313 #define color_ptr r3
3315 #define block_ptr_load r0
3316 #define draw_mask_store_ptr r3
3317 #define draw_mask_bits_ptr r12
3318 #define draw_mask_ptr r12
3319 #define pixel_store_ptr r14
3321 #define fb_ptr_cmp r4
3324 #define fb_ptr_next r14
3328 #define test_mask q0
3330 #define draw_mask q2
3331 #define zero_mask q3
3332 #define draw_mask_combined q4
3333 #define fb_pixels q5
3334 #define fb_pixels_next q6
3337 #define draw_mask_low d4
3338 #define draw_mask_high d5
3339 #define msb_mask_low d14
3340 #define msb_mask_high d15
3343 function(shade_blocks_textured_unmodulated_indirect)
3345 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3347 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3348 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3350 vld1.u32 { test_mask }, [psx_gpu, :128]
3351 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3354 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3356 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3357 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3358 [draw_mask_bits_ptr, :16], c_64
3359 vceq.u16 zero_mask, pixels, #0
3361 vtst.u16 draw_mask, draw_mask, test_mask
3362 vst1.u32 { pixels }, [pixel_store_ptr, :128], c_64
3364 subs num_blocks, num_blocks, #1
3368 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3369 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3371 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3372 [draw_mask_bits_ptr, :16], c_64
3373 vceq.u16 zero_mask, pixels, #0
3375 vtst.u16 draw_mask, draw_mask, test_mask
3376 vst1.u32 { pixels }, [pixel_store_ptr, :128], c_64
3378 vst1.u32 { draw_mask_combined }, [draw_mask_store_ptr, :128], c_64
3379 subs num_blocks, num_blocks, #1
3384 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3385 vst1.u32 { draw_mask_combined }, [draw_mask_store_ptr, :128], c_64
3392 function(shade_blocks_textured_unmodulated_direct)
3393 stmdb sp!, { r4, r14 }
3394 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3396 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3397 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3399 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
3402 vld1.u32 { test_mask }, [psx_gpu, :128]
3403 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3405 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3406 [draw_mask_bits_ptr, :16], c_64
3407 ldr fb_ptr_next, [block_ptr_load, #44]
3409 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3410 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3411 vceq.u16 zero_mask, pixels, #0
3412 vtst.u16 draw_mask, draw_mask, test_mask
3414 subs num_blocks, num_blocks, #1
3418 mov fb_ptr, fb_ptr_next
3419 ldr fb_ptr_next, [block_ptr_load, #44]
3421 vorr.u16 pixels, pixels, msb_mask
3423 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3424 vmov fb_pixels, fb_pixels_next
3426 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3427 [draw_mask_bits_ptr, :16], c_64
3428 vbif.u16 fb_pixels, pixels, draw_mask_combined
3430 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3431 pld [fb_ptr_next, #64]
3433 add fb_ptr_cmp, fb_ptr_cmp, #14
3434 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3439 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3440 vceq.u16 zero_mask, pixels, #0
3442 vst1.u16 { fb_pixels }, [fb_ptr]
3443 vtst.u16 draw_mask, draw_mask, test_mask
3446 subs num_blocks, num_blocks, #1
3450 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3451 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3453 vst1.u16 { fb_pixels_next }, [fb_ptr_next]
3455 ldmia sp!, { r4, pc }
3458 vst1.u16 { fb_pixels }, [fb_ptr]
3459 vceq.u16 zero_mask, pixels, #0
3461 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3462 vtst.u16 draw_mask, draw_mask, test_mask
3467 function(shade_blocks_unshaded_untextured_indirect)
3472 function(shade_blocks_unshaded_untextured_direct)
3473 stmdb sp!, { r4, r14 }
3474 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3476 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3477 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3479 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
3480 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3482 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3483 vld1.u16 { pixels }, [color_ptr, :128]
3486 vld1.u16 { draw_mask }, [draw_mask_ptr, :128], c_64
3488 vorr.u16 pixels, pixels, msb_mask
3489 subs num_blocks, num_blocks, #1
3491 ldr fb_ptr_next, [block_ptr_load], #64
3493 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3497 vmov fb_pixels, fb_pixels_next
3498 mov fb_ptr, fb_ptr_next
3499 ldr fb_ptr_next, [block_ptr_load], #64
3501 vbif.u16 fb_pixels, pixels, draw_mask
3502 vld1.u16 { draw_mask }, [draw_mask_ptr, :128], c_64
3504 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3505 add fb_ptr_cmp, fb_ptr_cmp, #14
3509 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3510 vst1.u16 { fb_pixels }, [fb_ptr]
3513 subs num_blocks, num_blocks, #1
3517 vbif.u16 fb_pixels_next, pixels, draw_mask
3518 vst1.u16 { fb_pixels_next }, [fb_ptr_next]
3520 ldmia sp!, { r4, pc }
3523 vst1.u16 { fb_pixels }, [fb_ptr]
3524 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3528 #undef draw_mask_ptr
3535 #define num_blocks r1
3536 #define msb_mask_ptr r2
3537 #define pixel_ptr r3
3538 #define draw_mask_ptr r0
3541 #define fb_ptr_next r14
3542 #define fb_ptr_cmp r4
3550 #undef msb_mask_high
3551 #undef draw_mask_next
3554 #undef fb_pixels_next
3557 #define draw_mask q1
3559 #define fb_pixels q3
3560 #define blend_pixels q4
3561 #define pixels_no_msb q5
3562 #define blend_mask q6
3563 #define fb_pixels_no_msb q7
3564 #define d128_0x8000 q8
3565 #define d128_0x0421 q9
3566 #define fb_pixels_next q10
3567 #define blend_pixels_next q11
3568 #define pixels_next q12
3569 #define draw_mask_next q13
3570 #define write_mask q14
3572 #define pixels_rb q5
3573 #define pixels_mg q7
3575 #define d128_0x7C1F q8
3576 #define d128_0x03E0 q9
3577 #define fb_pixels_rb q10
3578 #define fb_pixels_g q11
3579 #define fb_pixels_masked q11
3580 #define d128_0x83E0 q15
3581 #define pixels_fourth q7
3582 #define d128_0x1C07 q12
3583 #define d128_0x00E0 q13
3584 #define d128_0x80E0 q13
3586 #define msb_mask_low d0
3587 #define msb_mask_high d1
3589 #define blend_blocks_average_set_blend_mask_textured(source) \
3590 vclt.s16 blend_mask, source, #0 \
3592 #define blend_blocks_average_set_stp_bit_textured() \
3593 vorr.u16 blend_pixels, #0x8000 \
3595 #define blend_blocks_average_combine_textured(source) \
3596 vbif.u16 blend_pixels, source, blend_mask \
3598 #define blend_blocks_average_set_blend_mask_untextured(source) \
3600 #define blend_blocks_average_set_stp_bit_untextured() \
3602 #define blend_blocks_average_combine_untextured(source) \
3604 #define blend_blocks_average_mask_set_on() \
3605 vclt.s16 write_mask, fb_pixels_next, #0 \
3607 #define blend_blocks_average_mask_copy_on() \
3608 vorr.u16 draw_mask, draw_mask_next, write_mask \
3610 #define blend_blocks_average_mask_copy_b_on() \
3611 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3613 #define blend_blocks_average_mask_set_off() \
3615 #define blend_blocks_average_mask_copy_off() \
3616 vmov draw_mask, draw_mask_next \
3618 #define blend_blocks_average_mask_copy_b_off() \
3620 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3623 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3624 stmdb sp!, { r4, r14 }; \
3625 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3626 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3628 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3629 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3631 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3634 vmov.u16 d128_0x8000, #0x8000; \
3635 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
3636 ldr fb_ptr_next, [pixel_ptr, #28]; \
3638 vmov.u16 d128_0x0421, #0x0400; \
3639 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
3641 vorr.u16 d128_0x0421, #0x0021; \
3642 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3644 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3645 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3646 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3647 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3648 blend_blocks_average_mask_set_##mask_evaluate(); \
3649 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3651 subs num_blocks, num_blocks, #1; \
3655 mov fb_ptr, fb_ptr_next; \
3656 ldr fb_ptr_next, [pixel_ptr, #28]; \
3658 vmov pixels, pixels_next; \
3659 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
3661 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3663 blend_blocks_average_mask_copy_##mask_evaluate(); \
3664 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
3666 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3667 blend_blocks_average_set_stp_bit_##texturing(); \
3668 vmov fb_pixels, fb_pixels_next; \
3669 blend_blocks_average_combine_##texturing(pixels); \
3671 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3672 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3673 cmp fb_ptr_cmp, #28; \
3676 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3677 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3679 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3680 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3682 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3683 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3685 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3686 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3687 blend_blocks_average_mask_set_##mask_evaluate(); \
3688 vst1.u16 { fb_pixels }, [fb_ptr]; \
3691 subs num_blocks, num_blocks, #1; \
3695 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3696 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3698 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3699 blend_blocks_average_set_stp_bit_##texturing(); \
3700 blend_blocks_average_combine_##texturing(pixels_next); \
3702 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3703 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3704 vst1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3706 ldmia sp!, { r4, pc }; \
3709 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3710 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3711 vst1.u16 { fb_pixels }, [fb_ptr]; \
3713 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3714 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3715 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3716 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3717 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3718 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3722 blend_blocks_average_builder(textured, off)
3723 blend_blocks_average_builder(untextured, off)
3724 blend_blocks_average_builder(textured, on)
3725 blend_blocks_average_builder(untextured, on)
3728 #define blend_blocks_add_mask_set_on() \
3729 vclt.s16 write_mask, fb_pixels, #0 \
3731 #define blend_blocks_add_mask_copy_on() \
3732 vorr.u16 draw_mask, draw_mask, write_mask \
3734 #define blend_blocks_add_mask_set_off() \
3736 #define blend_blocks_add_mask_copy_off() \
3739 #define blend_blocks_add_textured_builder(mask_evaluate) \
3742 function(blend_blocks_textured_add_##mask_evaluate) \
3743 stmdb sp!, { r4, r14 }; \
3744 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3745 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3747 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3748 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3750 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3753 vmov.u16 d128_0x7C1F, #0x7C00; \
3754 vmov.u16 d128_0x03E0, #0x0300; \
3755 vmov.u16 d128_0x83E0, #0x8000; \
3756 vorr.u16 d128_0x03E0, #0x00E0; \
3757 vorr.u16 d128_0x7C1F, #0x001F; \
3758 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3760 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3761 ldr fb_ptr_next, [pixel_ptr, #28]; \
3762 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3763 vclt.s16 blend_mask, pixels, #0; \
3764 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3765 blend_blocks_add_mask_set_##mask_evaluate(); \
3766 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3768 blend_blocks_add_mask_copy_##mask_evaluate(); \
3769 vorr.u16 pixels, pixels, msb_mask; \
3770 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3771 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3772 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3773 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3774 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3775 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3776 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3777 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3779 subs num_blocks, num_blocks, #1; \
3783 mov fb_ptr, fb_ptr_next; \
3785 ldr fb_ptr_next, [pixel_ptr, #28]; \
3787 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3788 vclt.s16 blend_mask, pixels, #0; \
3790 vorr.u16 pixels, pixels, msb_mask; \
3791 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3792 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3794 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3795 pld [fb_ptr_next, #64]; \
3797 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3798 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3800 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3801 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3803 cmp fb_ptr_cmp, #28; \
3806 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3807 blend_blocks_add_mask_set_##mask_evaluate(); \
3808 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3809 blend_blocks_add_mask_copy_##mask_evaluate(); \
3810 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3811 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3812 vst1.u16 { blend_pixels }, [fb_ptr]; \
3815 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3816 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3817 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3818 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3819 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3821 subs num_blocks, num_blocks, #1; \
3825 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3826 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3827 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
3829 ldmia sp!, { r4, pc }; \
3832 vst1.u16 { blend_pixels }, [fb_ptr]; \
3833 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3835 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3836 blend_blocks_add_mask_set_##mask_evaluate(); \
3837 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3838 blend_blocks_add_mask_copy_##mask_evaluate(); \
3839 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3843 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3846 function(blend_blocks_untextured_add_##mask_evaluate) \
3847 stmdb sp!, { r4, r14 }; \
3848 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3849 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3851 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3852 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3854 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3857 vmov.u16 d128_0x7C1F, #0x7C00; \
3858 vmov.u16 d128_0x03E0, #0x0300; \
3859 vorr.u16 d128_0x7C1F, #0x001F; \
3860 vorr.u16 d128_0x03E0, #0x00E0; \
3862 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3863 ldr fb_ptr_next, [pixel_ptr, #28]; \
3864 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3865 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3866 blend_blocks_add_mask_set_##mask_evaluate(); \
3867 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3869 blend_blocks_add_mask_copy_##mask_evaluate(); \
3870 vand.u16 pixels_g, pixels, d128_0x03E0; \
3871 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3872 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3873 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3874 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3875 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3876 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3878 subs num_blocks, num_blocks, #1; \
3882 mov fb_ptr, fb_ptr_next; \
3884 ldr fb_ptr_next, [pixel_ptr, #28]; \
3886 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3888 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3889 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3890 vand.u16 pixels_g, pixels, d128_0x03E0; \
3892 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3893 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3895 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3896 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3897 cmp fb_ptr_cmp, #28; \
3900 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3901 blend_blocks_add_mask_set_##mask_evaluate(); \
3902 blend_blocks_add_mask_copy_##mask_evaluate(); \
3903 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3904 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3905 vst1.u16 { blend_pixels }, [fb_ptr]; \
3908 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3909 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3910 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3911 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3912 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3914 subs num_blocks, num_blocks, #1; \
3918 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3919 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3920 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3921 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
3923 ldmia sp!, { r4, pc }; \
3926 vst1.u16 { blend_pixels }, [fb_ptr]; \
3927 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3929 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3930 blend_blocks_add_mask_set_##mask_evaluate(); \
3931 blend_blocks_add_mask_copy_##mask_evaluate(); \
3932 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3936 blend_blocks_add_textured_builder(off)
3937 blend_blocks_add_textured_builder(on)
3938 blend_blocks_add_untextured_builder(off)
3939 blend_blocks_add_untextured_builder(on)
3941 #define blend_blocks_subtract_set_blend_mask_textured() \
3942 vclt.s16 blend_mask, pixels_next, #0 \
3944 #define blend_blocks_subtract_combine_textured() \
3945 vbif.u16 blend_pixels, pixels, blend_mask \
3947 #define blend_blocks_subtract_set_stb_textured() \
3948 vorr.u16 blend_pixels, #0x8000 \
3950 #define blend_blocks_subtract_msb_mask_textured() \
3951 vorr.u16 pixels, pixels_next, msb_mask \
3953 #define blend_blocks_subtract_set_blend_mask_untextured() \
3955 #define blend_blocks_subtract_combine_untextured() \
3957 #define blend_blocks_subtract_set_stb_untextured() \
3958 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3960 #define blend_blocks_subtract_msb_mask_untextured() \
3963 #define blend_blocks_subtract_mask_set_on() \
3964 vclt.s16 write_mask, fb_pixels, #0 \
3966 #define blend_blocks_subtract_mask_copy_on() \
3967 vorr.u16 draw_mask, draw_mask_next, write_mask \
3969 #define blend_blocks_subtract_mask_set_off() \
3971 #define blend_blocks_subtract_mask_copy_off() \
3972 vmov draw_mask, draw_mask_next \
3975 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
3978 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
3979 stmdb sp!, { r4, r14 }; \
3980 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3981 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3983 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3984 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3986 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3989 vmov.u16 d128_0x7C1F, #0x7C00; \
3990 vmov.u16 d128_0x03E0, #0x0300; \
3991 vorr.u16 d128_0x7C1F, #0x001F; \
3992 vorr.u16 d128_0x03E0, #0x00E0; \
3994 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
3995 ldr fb_ptr_next, [pixel_ptr, #28]; \
3996 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
3997 blend_blocks_subtract_set_blend_mask_##texturing(); \
3998 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3999 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4000 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4002 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4003 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4004 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4005 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4006 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4008 subs num_blocks, num_blocks, #1; \
4012 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4013 mov fb_ptr, fb_ptr_next; \
4014 ldr fb_ptr_next, [pixel_ptr, #28]; \
4016 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
4017 blend_blocks_subtract_msb_mask_##texturing(); \
4019 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
4020 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4021 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4022 blend_blocks_subtract_set_stb_##texturing(); \
4023 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4024 blend_blocks_subtract_combine_##texturing(); \
4025 blend_blocks_subtract_set_blend_mask_##texturing(); \
4026 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4028 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4029 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4030 cmp fb_ptr_cmp, #28; \
4033 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4034 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4035 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4036 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4037 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4038 vst1.u16 { blend_pixels }, [fb_ptr]; \
4039 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4042 subs num_blocks, num_blocks, #1; \
4046 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4048 blend_blocks_subtract_msb_mask_##texturing(); \
4049 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4050 blend_blocks_subtract_set_stb_##texturing(); \
4051 blend_blocks_subtract_combine_##texturing(); \
4052 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4053 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4055 ldmia sp!, { r4, pc }; \
4058 vst1.u16 { blend_pixels }, [fb_ptr]; \
4059 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4060 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4061 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4062 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4063 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4064 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4068 blend_blocks_subtract_builder(textured, off)
4069 blend_blocks_subtract_builder(textured, on)
4070 blend_blocks_subtract_builder(untextured, off)
4071 blend_blocks_subtract_builder(untextured, on)
4074 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4077 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4078 stmdb sp!, { r4, r14 }; \
4079 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4080 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
4082 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4083 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
4085 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4088 vmov.u16 d128_0x7C1F, #0x7C00; \
4089 vmov.u16 d128_0x03E0, #0x0300; \
4090 vmov.u16 d128_0x1C07, #0x1C00; \
4091 vmov.u16 d128_0x00E0, #0x00E0; \
4092 vorr.u16 d128_0x7C1F, #0x001F; \
4093 vorr.u16 d128_0x03E0, #0x00E0; \
4094 vorr.u16 d128_0x1C07, #0x0007; \
4096 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4097 ldr fb_ptr_next, [pixel_ptr, #28]; \
4098 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4099 vclt.s16 blend_mask, pixels, #0; \
4100 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4101 blend_blocks_add_mask_set_##mask_evaluate(); \
4102 vshr.s16 pixels_fourth, pixels, #2; \
4103 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4105 blend_blocks_add_mask_copy_##mask_evaluate(); \
4106 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4107 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4108 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4109 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4110 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4111 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4112 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4114 subs num_blocks, num_blocks, #1; \
4118 mov fb_ptr, fb_ptr_next; \
4119 ldr fb_ptr_next, [pixel_ptr, #28]; \
4121 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4122 vbif.u16 blend_pixels, pixels, blend_mask; \
4124 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4125 vclt.s16 blend_mask, pixels, #0; \
4126 vshr.s16 pixels_fourth, pixels, #2; \
4127 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4128 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4130 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4131 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4133 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4134 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4135 cmp fb_ptr_cmp, #28; \
4138 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4139 blend_blocks_add_mask_set_##mask_evaluate(); \
4140 blend_blocks_add_mask_copy_##mask_evaluate(); \
4141 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4142 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4143 vst1.u16 { blend_pixels }, [fb_ptr]; \
4146 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4147 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4148 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4149 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4150 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4152 subs num_blocks, num_blocks, #1; \
4156 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4157 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4158 vbif.u16 blend_pixels, pixels, blend_mask; \
4159 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4160 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4162 ldmia sp!, { r4, pc }; \
4165 vst1.u16 { blend_pixels }, [fb_ptr]; \
4166 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4168 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4169 blend_blocks_add_mask_set_##mask_evaluate(); \
4170 blend_blocks_add_mask_copy_##mask_evaluate(); \
4171 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4176 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4179 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4180 stmdb sp!, { r4, r14 }; \
4181 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4182 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
4184 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4185 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
4187 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4190 vmov.u16 d128_0x7C1F, #0x7C00; \
4191 vmov.u16 d128_0x03E0, #0x0300; \
4192 vmov.u16 d128_0x1C07, #0x1C00; \
4193 vmov.u16 d128_0x00E0, #0x00E0; \
4194 vorr.u16 d128_0x7C1F, #0x001F; \
4195 vorr.u16 d128_0x03E0, #0x00E0; \
4196 vorr.u16 d128_0x1C07, #0x0007; \
4198 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4199 ldr fb_ptr_next, [pixel_ptr, #28]; \
4200 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4201 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4202 blend_blocks_add_mask_set_##mask_evaluate(); \
4203 vshr.s16 pixels_fourth, pixels, #2; \
4204 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4206 blend_blocks_add_mask_copy_##mask_evaluate(); \
4207 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4208 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4209 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4210 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4211 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4212 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4213 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4215 subs num_blocks, num_blocks, #1; \
4219 mov fb_ptr, fb_ptr_next; \
4220 ldr fb_ptr_next, [pixel_ptr, #28]; \
4222 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4224 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4225 vshr.s16 pixels_fourth, pixels, #2; \
4226 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4227 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4229 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4230 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4232 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4233 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4234 cmp fb_ptr_cmp, #28; \
4237 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4238 blend_blocks_add_mask_set_##mask_evaluate(); \
4239 blend_blocks_add_mask_copy_##mask_evaluate(); \
4240 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4241 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4242 vst1.u16 { blend_pixels }, [fb_ptr]; \
4245 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4246 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4247 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4248 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4249 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4251 subs num_blocks, num_blocks, #1; \
4255 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4256 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4257 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4258 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4260 ldmia sp!, { r4, pc }; \
4263 vst1.u16 { blend_pixels }, [fb_ptr]; \
4264 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4266 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4267 blend_blocks_add_mask_set_##mask_evaluate(); \
4268 blend_blocks_add_mask_copy_##mask_evaluate(); \
4269 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4273 blend_blocks_add_fourth_textured_builder(off)
4274 blend_blocks_add_fourth_textured_builder(on)
4275 blend_blocks_add_fourth_untextured_builder(off)
4276 blend_blocks_add_fourth_untextured_builder(on)
4278 // TODO: Optimize this more. Need a scene that actually uses it for
4283 function(blend_blocks_textured_unblended_on)
4284 stmdb sp!, { r4, r14 }
4285 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4286 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
4288 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4289 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
4291 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4294 ldr fb_ptr, [pixel_ptr, #28]
4295 vld1.u16 { fb_pixels }, [fb_ptr]
4296 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64
4297 vclt.s16 write_mask, fb_pixels, #0
4298 vld1.u32 { pixels }, [pixel_ptr, :128], c_64
4300 subs num_blocks, num_blocks, #1
4304 vorr.u16 pixels, pixels, msb_mask
4305 vorr.u16 draw_mask, draw_mask, write_mask
4306 vbif.u16 fb_pixels, pixels, draw_mask
4307 vst1.u16 { fb_pixels }, [fb_ptr]
4309 ldr fb_ptr, [pixel_ptr, #28]
4310 vld1.u16 { fb_pixels }, [fb_ptr]
4311 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64
4312 vclt.s16 write_mask, fb_pixels, #0
4313 vld1.u32 { pixels }, [pixel_ptr, :128], c_64
4315 subs num_blocks, num_blocks, #1
4319 vorr.u16 pixels, pixels, msb_mask
4320 vorr.u16 draw_mask, draw_mask, write_mask
4321 vbif.u16 fb_pixels, pixels, draw_mask
4322 vst1.u16 { fb_pixels }, [fb_ptr]
4324 ldmia sp!, { r4, pc }
4327 function(blend_blocks_textured_unblended_off)
4337 vld1.u32 { u_whole_8, v_whole_8 }, [r1, :128], r3
4357 #define num_width r12
4367 function(render_block_fill_body)
4368 vdup.u16 colors_a, color
4371 vmov colors_b, colors_a
4372 sub pitch, pitch, width, lsl #1
4374 mov num_width, width
4377 vst1.u32 { colors_a, colors_b }, [vram_ptr, :256]!
4379 subs num_width, num_width, #16
4382 add vram_ptr, vram_ptr, pitch
4383 mov num_width, width
4385 subs height, height, #1
4399 #undef dirty_textures_mask
4401 #undef current_texture_mask
4412 #define offset_u_right r10
4413 #define width_rounded r11
4414 #define height_rounded r12
4416 #define texture_offset_base r1
4417 #define tile_width r2
4418 #define tile_height r3
4419 #define num_blocks r4
4421 #define sub_tile_height r6
4423 #define texture_mask r8
4424 #define column_data r9
4425 #define texture_offset r10
4426 #define tiles_remaining r11
4427 #define fb_ptr_advance_column r12
4428 #define texture_block_ptr r14
4432 #define texture_page_ptr r3
4433 #define left_block_mask r4
4434 #define right_block_mask r5
4435 #define texture_mask_rev r10
4436 #define control_mask r11
4438 #define dirty_textures_mask r4
4440 #define current_texture_mask r6
4454 #define draw_masks_fb_ptrs q1
4456 #define draw_mask_fb_ptr_left d2
4457 #define draw_mask_fb_ptr_right d3
4459 #define draw_mask_fb_ptr_left_a d2
4460 #define draw_mask_fb_ptr_left_b d3
4461 #define draw_mask_fb_ptr_right_a d10
4462 #define draw_mask_fb_ptr_right_b d11
4463 #define draw_masks_fb_ptrs2 q5
4465 #define clut_low_a d4
4466 #define clut_low_b d5
4467 #define clut_high_a d6
4468 #define clut_high_b d7
4470 #define block_masks d8
4471 #define block_masks_shifted d9
4476 #define texels_low d12
4477 #define texels_high d13
4479 #define texels_wide_low d14
4480 #define texels_wide_high d15
4481 #define texels_wide q7
4484 setup_sprite_flush_blocks:
4487 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
4488 bl flush_render_block_buffer
4489 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
4493 add block, psx_gpu, #psx_gpu_blocks_offset
4497 setup_sprite_update_texture_4bpp_cache:
4498 stmdb sp!, { r0 - r3, r14 }
4499 bl update_texture_4bpp_cache
4500 ldmia sp!, { r0 - r3, pc }
4503 setup_sprite_update_texture_8bpp_cache:
4504 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r14 }
4505 bl update_texture_8bpp_cache
4506 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS pc }
4509 #define setup_sprite_tiled_initialize_4bpp() \
4510 ldr dirty_textures_mask, \
4511 [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]; \
4512 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]; \
4514 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]; \
4515 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]; \
4517 tst current_texture_mask, dirty_textures_mask; \
4518 vuzp.u8 clut_a, clut_b; \
4520 blne setup_sprite_update_texture_4bpp_cache \
4522 #define setup_sprite_tiled_initialize_8bpp() \
4523 ldr dirty_textures_mask, \
4524 [psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset]; \
4525 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]; \
4527 tst current_texture_mask, dirty_textures_mask; \
4528 blne setup_sprite_update_texture_8bpp_cache \
4531 #define setup_sprite_block_count_single() \
4534 #define setup_sprite_block_count_double() \
4535 sub_tile_height, lsl #1 \
4537 #define setup_sprite_tile_add_blocks(type) \
4538 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4539 cmp num_blocks, #MAX_BLOCKS; \
4541 movgt num_blocks, setup_sprite_block_count_##type(); \
4542 blgt setup_sprite_flush_blocks \
4545 #define setup_sprite_tile_full_4bpp(edge) \
4546 setup_sprite_tile_add_blocks(double); \
4549 and texture_block_ptr, texture_offset, texture_mask; \
4550 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4553 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4554 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4556 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4557 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4559 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4560 add texture_block_ptr, texture_offset, #8; \
4562 and texture_block_ptr, texture_block_ptr, texture_mask; \
4563 add block, block, #40; \
4565 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4566 add fb_ptr, fb_ptr, #16; \
4568 vst1.u32 { draw_mask_fb_ptr_left }, [block, :64]; \
4569 add block, block, #24; \
4571 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4572 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4575 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4576 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4578 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4579 add block, block, #40; \
4581 add texture_offset, texture_offset, #0x10; \
4582 add fb_ptr, fb_ptr, #(2048 - 16); \
4584 vst1.u32 { draw_mask_fb_ptr_right }, [block, :64]; \
4585 add block, block, #24; \
4587 subs sub_tile_height, sub_tile_height, #1; \
4590 add texture_offset, texture_offset, #0xF00; \
4591 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4594 #define setup_sprite_tile_half_4bpp(edge) \
4595 setup_sprite_tile_add_blocks(single); \
4598 and texture_block_ptr, texture_offset, texture_mask; \
4599 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4602 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4603 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4605 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4606 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4608 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4609 add block, block, #40; \
4611 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4612 vst1.u32 { draw_mask_fb_ptr_##edge }, [block, :64]; \
4614 add block, block, #24; \
4615 add texture_offset, texture_offset, #0x10; \
4617 add fb_ptr, fb_ptr, #2048; \
4618 subs sub_tile_height, sub_tile_height, #1; \
4622 add texture_offset, texture_offset, #0xF00; \
4623 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4626 #define setup_sprite_tile_full_8bpp(edge) \
4627 setup_sprite_tile_add_blocks(double); \
4628 add block, block, #16; \
4631 and texture_block_ptr, texture_offset, texture_mask; \
4632 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4635 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4636 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4638 add texture_block_ptr, texture_offset, #8; \
4639 vst1.u32 { texels }, [block, :64]; \
4641 and texture_block_ptr, texture_block_ptr, texture_mask; \
4642 add block, block, #24; \
4644 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4646 add fb_ptr, fb_ptr, #16; \
4647 vst1.u32 { draw_mask_fb_ptr_left }, [block, :64]; \
4649 add block, block, #40; \
4650 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4653 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4654 vst1.u32 { texels }, [block, :64]; \
4655 add block, block, #24; \
4657 add texture_offset, texture_offset, #0x10; \
4658 add fb_ptr, fb_ptr, #(2048 - 16); \
4660 vst1.u32 { draw_mask_fb_ptr_right }, [block, :64]; \
4661 add block, block, #40; \
4663 subs sub_tile_height, sub_tile_height, #1; \
4666 sub block, block, #16; \
4667 add texture_offset, texture_offset, #0xF00; \
4668 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4671 #define setup_sprite_tile_half_8bpp(edge) \
4672 setup_sprite_tile_add_blocks(single); \
4673 add block, block, #16; \
4676 and texture_block_ptr, texture_offset, texture_mask; \
4677 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4680 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4681 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4683 vst1.u32 { texels }, [block, :64]; \
4684 add block, block, #24; \
4686 vst1.u32 { draw_mask_fb_ptr_##edge }, [block, :64]; \
4687 add block, block, #40; \
4689 add texture_offset, texture_offset, #0x10; \
4690 add fb_ptr, fb_ptr, #2048; \
4692 subs sub_tile_height, sub_tile_height, #1; \
4695 sub block, block, #16; \
4696 add texture_offset, texture_offset, #0xF00; \
4697 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4700 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4701 add texture_offset, texture_offset_base, #8; \
4702 add fb_ptr, fb_ptr, #16 \
4704 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4705 mov texture_offset, texture_offset_base \
4707 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4708 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4710 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4711 mov texture_offset, texture_offset_base \
4713 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4714 sub fb_ptr, fb_ptr, #16 \
4716 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4718 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4719 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4721 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4724 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \
4726 mov sub_tile_height, column_data; \
4727 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4728 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4729 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4731 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \
4733 and sub_tile_height, column_data, #0xFF; \
4734 mov tiles_remaining, column_data, lsr #16; \
4735 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4736 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4738 subs tiles_remaining, tiles_remaining, #1; \
4742 mov sub_tile_height, #16; \
4743 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4744 subs tiles_remaining, tiles_remaining, #1; \
4748 uxtb sub_tile_height, column_data, ror #8; \
4749 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4750 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4753 #define setup_sprite_column_data_single() \
4754 mov column_data, height; \
4755 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset] \
4757 #define setup_sprite_column_data_multi() \
4758 and height_rounded, height_rounded, #0xF; \
4759 rsb column_data, offset_v, #16; \
4761 add height_rounded, height_rounded, #1; \
4762 sub tile_height, tile_height, #1; \
4764 orr column_data, column_data, tile_height, lsl #16; \
4765 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]; \
4767 orr column_data, column_data, height_rounded, lsl #8 \
4769 #define setup_sprite_setup_left_draw_mask_fb_ptr() \
4770 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4771 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4773 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column() \
4774 mov fb_ptr_advance_column, #32; \
4775 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4777 sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11; \
4778 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4780 #define setup_sprite_setup_right_draw_mask_fb_ptr() \
4781 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4782 vdup.u8 draw_mask_fb_ptr_right, block_masks[5] \
4784 #define setup_sprite_tile_column_width_single(tm, multi_height, edge_mode, \
4786 setup_sprite_##tm##_single_##multi_height##_##edge_mode##_##edge##x4mode: \
4787 setup_sprite_column_data_##multi_height(); \
4788 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4789 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4790 setup_sprite_setup_left_draw_mask_fb_ptr##x4mode(); \
4792 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, tm, x4mode); \
4793 ldmia sp!, { r4 - r11, pc } \
4795 #define setup_sprite_tiled_advance_column() \
4796 add texture_offset_base, texture_offset_base, #0x100; \
4797 tst texture_offset_base, #0xF00; \
4798 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4800 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4801 right_mode, x4mode) \
4802 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode##x4mode:\
4803 setup_sprite_column_data_##multi_height(); \
4805 setup_sprite_setup_left_draw_mask_fb_ptr_advance_column##x4mode(); \
4807 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm, x4mode);\
4809 subs tile_width, tile_width, #2; \
4810 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4814 vmov.u8 draw_masks_fb_ptrs, #0; \
4815 vmov.u8 draw_masks_fb_ptrs2, #0; \
4818 setup_sprite_tiled_advance_column(); \
4819 setup_sprite_tile_column_height_##multi_height(full, none, tm, x4mode); \
4820 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4821 subs tile_width, tile_width, #1; \
4825 setup_sprite_setup_right_draw_mask_fb_ptr##x4mode(); \
4827 setup_sprite_tiled_advance_column(); \
4828 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm, x4mode);\
4829 ldmia sp!, { r4 - r11, pc } \
4832 #define setup_sprite_offset_u_adjust() \
4834 #define setup_sprite_get_left_block_mask() \
4835 and left_block_mask, left_block_mask, #0xFF \
4837 #define setup_sprite_compare_left_block_mask() \
4838 cmp left_block_mask, #0xFF \
4840 #define setup_sprite_get_right_block_mask() \
4841 uxtb right_block_mask, right_block_mask, ror #8 \
4843 #define setup_sprite_compare_right_block_mask() \
4844 cmp right_block_mask, #0xFF \
4849 #define fb_ptr2 column_data
4851 #define setup_sprite_offset_u_adjust_4x() \
4852 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4853 lsl offset_u_right, #1; \
4855 add offset_u_right, #1 \
4857 #define setup_sprite_get_left_block_mask_4x() \
4858 sxth left_block_mask, left_block_mask \
4860 #define setup_sprite_compare_left_block_mask_4x() \
4861 cmp left_block_mask, #0xFFFFFFFF \
4863 #define setup_sprite_get_right_block_mask_4x() \
4864 sxth right_block_mask, right_block_mask, ror #16 \
4866 #define setup_sprite_compare_right_block_mask_4x() \
4867 cmp right_block_mask, #0xFFFFFFFF \
4870 #define widen_texels_16bpp(texels_) \
4871 vmov texels_wide_low, texels_; \
4872 vmov texels_wide_high, texels_; \
4873 vzip.16 texels_wide_low, texels_wide_high \
4875 #define widen_texels_8bpp(texels_) \
4876 vmov texels_wide_low, texels_; \
4877 vmov texels_wide_high, texels_; \
4878 vzip.8 texels_wide_low, texels_wide_high \
4880 #define write_block_16bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4881 vst1.u32 { texels_ }, [block_, :128]; \
4882 add block_, block_, #40; \
4884 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4885 vst1.u32 { draw_mask_fb_ptr_ }, [block_, :64]; \
4886 add block_, block_, #24 \
4888 /* assumes 16-byte offset already added to block_ */
4889 #define write_block_8bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4890 vst1.u32 { texels_ }, [block_, :64]; \
4891 add block_, block_, #24; \
4893 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4894 vst1.u32 { draw_mask_fb_ptr_ }, [block_, :64]; \
4895 add block_, block_, #40 \
4897 #define do_texture_block_16bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4898 draw_mask_fb_ptr_b_) \
4899 widen_texels_16bpp(texels_low); \
4900 add fb_ptr_tmp, fb_ptr, #1024*2; \
4902 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr); \
4904 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4905 widen_texels_16bpp(texels_high); \
4907 add fb_ptr_tmp, fb_ptr, #8*2; \
4908 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4910 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4911 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4913 #define do_texture_block_8bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4914 draw_mask_fb_ptr_b_) \
4915 widen_texels_8bpp(texels); \
4916 add fb_ptr_tmp, fb_ptr, #1024*2; \
4918 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr); \
4919 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4921 add fb_ptr_tmp, fb_ptr, #8*2; \
4922 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4924 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4925 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4928 #define setup_sprite_tiled_initialize_4bpp_4x() \
4929 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]; \
4930 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]; \
4932 vuzp.u8 clut_a, clut_b \
4934 #define setup_sprite_tiled_initialize_8bpp_4x() \
4937 #define setup_sprite_block_count_single_4x() \
4938 sub_tile_height, lsl #2 \
4940 #define setup_sprite_block_count_double_4x() \
4941 sub_tile_height, lsl #(1+2) \
4943 #define setup_sprite_tile_full_4bpp_4x(edge) \
4944 setup_sprite_tile_add_blocks(double_4x); \
4945 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4948 and texture_block_ptr, texture_offset, texture_mask; \
4951 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4952 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4954 add texture_block_ptr, texture_offset, #8; \
4955 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4957 and texture_block_ptr, texture_block_ptr, texture_mask; \
4958 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4960 vzip.8 texels_low, texels_high; \
4961 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
4962 draw_mask_fb_ptr_left_b); \
4964 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4965 pld [fb_ptr, #2048]; \
4967 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4968 add fb_ptr, fb_ptr, #16*2; \
4970 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4971 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4973 vzip.8 texels_low, texels_high; \
4974 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
4975 draw_mask_fb_ptr_right_b); \
4977 add texture_offset, texture_offset, #0x10; \
4978 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
4980 subs sub_tile_height, sub_tile_height, #1; \
4983 ldr column_data, [sp], #8; /* fb_ptr2 */ \
4984 add texture_offset, texture_offset, #0xF00; \
4985 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4988 #define setup_sprite_tile_half_4bpp_4x(edge) \
4989 setup_sprite_tile_add_blocks(single_4x); \
4990 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4993 and texture_block_ptr, texture_offset, texture_mask; \
4996 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4997 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4999 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5000 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
5002 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
5003 add texture_offset, texture_offset, #0x10; \
5005 vzip.8 texels_low, texels_high; \
5006 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5007 draw_mask_fb_ptr_##edge##_b); \
5009 pld [fb_ptr, #2048]; \
5010 add fb_ptr, fb_ptr, #2048 * 2; \
5012 subs sub_tile_height, sub_tile_height, #1; \
5015 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5016 add texture_offset, texture_offset, #0xF00; \
5017 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5020 #define setup_sprite_tile_full_8bpp_4x(edge) \
5021 setup_sprite_tile_add_blocks(double_4x); \
5022 add block, block, #16; \
5023 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5026 and texture_block_ptr, texture_offset, texture_mask; \
5029 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5030 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5032 add texture_block_ptr, texture_offset, #8; \
5033 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
5034 draw_mask_fb_ptr_left_b); \
5036 pld [fb_ptr, #2048]; \
5037 and texture_block_ptr, texture_block_ptr, texture_mask; \
5039 add fb_ptr, fb_ptr, #16*2; \
5040 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5042 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5044 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
5045 draw_mask_fb_ptr_right_b); \
5047 add texture_offset, texture_offset, #0x10; \
5048 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
5050 subs sub_tile_height, sub_tile_height, #1; \
5053 sub block, block, #16; \
5054 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5055 add texture_offset, texture_offset, #0xF00; \
5056 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5059 #define setup_sprite_tile_half_8bpp_4x(edge) \
5060 setup_sprite_tile_add_blocks(single_4x); \
5061 add block, block, #16; \
5062 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5065 and texture_block_ptr, texture_offset, texture_mask; \
5068 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5069 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5071 pld [fb_ptr, #2048]; \
5072 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5073 draw_mask_fb_ptr_##edge##_b); \
5075 add texture_offset, texture_offset, #0x10; \
5076 add fb_ptr, fb_ptr, #2048 * 2; \
5078 subs sub_tile_height, sub_tile_height, #1; \
5081 sub block, block, #16; \
5082 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5083 add texture_offset, texture_offset, #0xF00; \
5084 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5087 #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \
5088 add texture_offset, texture_offset_base, #8; \
5089 add fb_ptr, fb_ptr, #16 * 2 \
5091 #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \
5092 mov texture_offset, texture_offset_base \
5094 #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \
5095 setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \
5097 #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \
5098 mov texture_offset, texture_offset_base \
5100 #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \
5101 sub fb_ptr, fb_ptr, #16 * 2 \
5103 #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \
5105 #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \
5106 setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \
5108 #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \
5111 #define setup_sprite_setup_left_draw_mask_fb_ptr_4x() \
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 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5115 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5117 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column_4x() \
5118 mov fb_ptr_advance_column, #32 * 2; \
5119 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5120 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5121 sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11 + 1; \
5122 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5123 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5125 #define setup_sprite_setup_right_draw_mask_fb_ptr_4x() \
5126 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[4]; \
5127 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[5]; \
5128 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[6]; \
5129 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[7] \
5139 // [sp + 12]: color (unused)
5141 #define setup_sprite_tiled_builder(texture_mode, x4mode) \
5143 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \
5145 setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \
5147 setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \
5149 setup_sprite_tile_column_width_single(texture_mode, single, full, none, \
5151 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \
5153 setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \
5155 setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \
5157 setup_sprite_tile_column_width_single(texture_mode, single, half, right, \
5159 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \
5161 setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \
5163 setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \
5165 setup_sprite_tile_column_width_single(texture_mode, single, half, left, \
5167 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \
5169 setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \
5174 function(setup_sprite_##texture_mode##x4mode) \
5175 stmdb sp!, { r4 - r11, r14 }; \
5176 setup_sprite_tiled_initialize_##texture_mode##x4mode(); \
5179 and offset_u, u, #0xF; \
5181 ldr width, [sp, #40]; \
5182 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
5184 ldr height, [sp, #44]; \
5185 add fb_ptr, fb_ptr, y, lsl #11; \
5187 add fb_ptr, fb_ptr, x, lsl #1; \
5188 and offset_v, v, #0xF; \
5190 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
5191 add width_rounded, offset_u, width; \
5193 add height_rounded, offset_v, height; \
5194 add width_rounded, width_rounded, #15; \
5196 add height_rounded, height_rounded, #15; \
5197 mov tile_width, width_rounded, lsr #4; \
5199 /* texture_offset_base = VH-VL-00-00 */\
5200 mov texture_offset_base, v, lsl #8; \
5201 and offset_u_right, width_rounded, #0xF; \
5203 /* texture_offset_base = VH-UH-UL-00 */\
5204 bfi texture_offset_base, u, #4, #8; \
5205 mov right_block_mask, #0xFFFFFFFE; \
5207 setup_sprite_offset_u_adjust##x4mode(); \
5209 /* texture_offset_base = VH-UH-VL-00 */\
5210 bfi texture_offset_base, v, #4, #4; \
5211 mov left_block_mask, #0xFFFFFFFF; \
5213 mov tile_height, height_rounded, lsr #4; \
5214 mvn left_block_mask, left_block_mask, lsl offset_u; \
5216 /* texture_mask = HH-HL-WH-WL */\
5217 ldrh texture_mask, [psx_gpu, #psx_gpu_texture_mask_width_offset]; \
5218 mov right_block_mask, right_block_mask, lsl offset_u_right; \
5220 /* texture_mask_rev = WH-WL-HH-HL */\
5221 rev16 texture_mask_rev, texture_mask; \
5222 vmov block_masks, left_block_mask, right_block_mask; \
5224 /* texture_mask = HH-HL-HL-WL */\
5225 bfi texture_mask, texture_mask_rev, #4, #4; \
5226 /* texture_mask_rev = 00-00-00-WH */\
5227 mov texture_mask_rev, texture_mask_rev, lsr #12; \
5229 /* texture_mask = HH-WH-HL-WL */\
5230 bfi texture_mask, texture_mask_rev, #8, #4; \
5231 setup_sprite_get_left_block_mask##x4mode(); \
5233 mov control_mask, #0; \
5234 setup_sprite_compare_left_block_mask##x4mode(); \
5236 setup_sprite_get_right_block_mask##x4mode(); \
5237 orreq control_mask, control_mask, #0x4; \
5239 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
5240 setup_sprite_compare_right_block_mask##x4mode(); \
5242 orreq control_mask, control_mask, #0x8; \
5243 cmp tile_width, #1; \
5245 add block, psx_gpu, #psx_gpu_blocks_offset; \
5246 orreq control_mask, control_mask, #0x1; \
5248 cmp tile_height, #1; \
5249 add block, block, num_blocks, lsl #6; \
5251 orreq control_mask, control_mask, #0x2; \
5252 JT_OP_REL(9f, control_mask, temp); \
5253 JT_OP(ldr pc, [pc, control_mask, lsl #2]); \
5257 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_full_full##x4mode); \
5258 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_full_none##x4mode); \
5259 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_full_full##x4mode); \
5260 .word JTE(9b, setup_sprite_##texture_mode##_single_single_full_none##x4mode); \
5261 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_half_full##x4mode); \
5262 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_half_right##x4mode); \
5263 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_half_full##x4mode); \
5264 .word JTE(9b, setup_sprite_##texture_mode##_single_single_half_right##x4mode);\
5265 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_full_half##x4mode); \
5266 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_half_left##x4mode); \
5267 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_full_half##x4mode); \
5268 .word JTE(9b, setup_sprite_##texture_mode##_single_single_half_left##x4mode); \
5269 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_half_half##x4mode); \
5271 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_half_half##x4mode); \
5274 setup_sprite_tiled_builder(4bpp,);
5275 setup_sprite_tiled_builder(8bpp,);
5277 #undef draw_mask_fb_ptr_left
5278 #undef draw_mask_fb_ptr_right
5280 setup_sprite_tiled_builder(4bpp, _4x);
5281 setup_sprite_tiled_builder(8bpp, _4x);
5289 #define block_ptr r0
5290 #define num_blocks r1
5292 #define texel_shift_mask r3
5293 #define block_pixels_a r4
5294 #define block_pixels_b r5
5303 #define texels_01 r6
5304 #define texels_23 r7
5305 #define texels_45 r8
5306 #define texels_67 r9
5308 function(texture_sprite_blocks_8bpp)
5309 stmdb sp!, { r4 - r11, r14 }
5310 movw texel_shift_mask, #(0xFF << 1)
5312 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5313 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
5315 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
5316 ldr block_pixels_a, [block_ptr, #16]
5319 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
5320 ldr block_pixels_b, [block_ptr, #20]
5322 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
5323 ldrh texel_0, [clut_ptr, texel_0]
5325 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
5326 ldrh texel_1, [clut_ptr, texel_1]
5328 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
5329 ldr block_pixels_a, [block_ptr, #(64 + 16)]
5331 ldrh texel_2, [clut_ptr, texel_2]
5332 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
5334 ldrh texel_3, [clut_ptr, texel_3]
5335 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
5337 ldrh texel_4, [clut_ptr, texel_4]
5338 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
5340 ldrh texel_5, [clut_ptr, texel_5]
5341 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
5343 ldrh texel_6, [clut_ptr, texel_6]
5344 orr texels_01, texel_0, texel_1, lsl #16
5346 ldrh texel_7, [clut_ptr, texel_7]
5347 orr texels_23, texel_2, texel_3, lsl #16
5349 orr texels_45, texel_4, texel_5, lsl #16
5350 str texels_01, [block_ptr, #0]
5352 orr texels_67, texel_6, texel_7, lsl #16
5353 str texels_23, [block_ptr, #4]
5355 subs num_blocks, num_blocks, #1
5356 str texels_45, [block_ptr, #8]
5358 str texels_67, [block_ptr, #12]
5359 add block_ptr, block_ptr, #64
5363 ldmia sp!, { r4 - r11, pc }
5366 #undef width_rounded
5369 #undef texture_offset
5372 #undef texels_wide_low
5373 #undef texels_wide_high
5385 #define left_offset r8
5386 #define width_rounded r9
5387 #define right_width r10
5389 #define block_width r11
5391 #define texture_offset_base r1
5392 #define texture_mask r2
5393 #define texture_page_ptr r3
5394 #define num_blocks r4
5397 #define texture_offset r8
5398 #define blocks_remaining r9
5400 #define fb_ptr_pitch r12
5401 #define texture_block_ptr r14
5403 #define texture_mask_width r2
5404 #define texture_mask_height r3
5405 #define left_mask_bits r4
5406 #define right_mask_bits r5
5410 #undef block_masks_shifted
5413 #define block_masks d0
5414 #define block_masks_shifted d1
5415 #define draw_mask_fb_ptr d2
5418 #define draw_mask_fb_ptr_a d2
5419 #define draw_mask_fb_ptr_b d3
5420 #define texels_low d4
5421 #define texels_high d5
5422 #define texels_wide_low d6
5423 #define texels_wide_high d7
5424 #define texels_wide q3
5427 setup_sprites_16bpp_flush:
5430 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
5431 bl flush_render_block_buffer
5432 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
5436 add block, psx_gpu, #psx_gpu_blocks_offset
5437 mov num_blocks, block_width
5441 function(setup_sprite_16bpp)
5442 stmdb sp!, { r4 - r11, r14 }
5443 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5446 add fb_ptr, fb_ptr, y, lsl #11
5448 ldr width, [sp, #40]
5449 add fb_ptr, fb_ptr, x, lsl #1
5451 ldr height, [sp, #44]
5452 and left_offset, u, #0x7
5454 add texture_offset_base, u, u
5455 add width_rounded, width, #7
5457 add texture_offset_base, texture_offset_base, v, lsl #11
5458 mov left_mask_bits, #0xFF
5460 ldrb texture_mask_width, [psx_gpu, #psx_gpu_texture_mask_width_offset]
5461 add width_rounded, width_rounded, left_offset
5463 ldrb texture_mask_height, [psx_gpu, #psx_gpu_texture_mask_height_offset]
5464 sub fb_ptr, fb_ptr, left_offset, lsl #1
5466 add texture_mask, texture_mask_width, texture_mask_width
5467 mov right_mask_bits, #0xFE
5469 and right_width, width_rounded, #0x7
5470 mvn left_mask_bits, left_mask_bits, lsl left_offset
5472 add texture_mask, texture_mask, texture_mask_height, lsl #11
5473 mov block_width, width_rounded, lsr #3
5475 mov right_mask_bits, right_mask_bits, lsl right_width
5476 movw fb_ptr_pitch, #(2048 + 16)
5478 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5479 vmov block_masks, left_mask_bits, right_mask_bits
5481 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5482 add block, psx_gpu, #psx_gpu_blocks_offset
5484 bic texture_offset_base, texture_offset_base, #0xF
5487 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
5488 add block, block, num_blocks, lsl #6
5492 vext.32 block_masks_shifted, block_masks, block_masks, #1
5493 vorr.u32 block_masks, block_masks, block_masks_shifted
5494 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5497 add num_blocks, num_blocks, #1
5498 cmp num_blocks, #MAX_BLOCKS
5499 blgt setup_sprites_16bpp_flush
5501 and texture_block_ptr, texture_offset_base, texture_mask
5502 subs height, height, #1
5504 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5505 vld1.u32 { texels }, [texture_block_ptr, :128]
5507 vst1.u32 { texels }, [block, :128]
5508 add block, block, #40
5510 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5513 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5515 add block, block, #24
5516 add texture_offset_base, texture_offset_base, #2048
5517 add fb_ptr, fb_ptr, #2048
5518 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5521 ldmia sp!, { r4 - r11, pc }
5524 add num_blocks, num_blocks, block_width
5525 mov texture_offset, texture_offset_base
5527 cmp num_blocks, #MAX_BLOCKS
5528 blgt setup_sprites_16bpp_flush
5530 add texture_offset_base, texture_offset_base, #2048
5531 and texture_block_ptr, texture_offset, texture_mask
5533 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5534 vld1.u32 { texels }, [texture_block_ptr, :128]
5536 vst1.u32 { texels }, [block, :128]
5537 add block, block, #40
5539 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5540 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5543 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5544 subs blocks_remaining, block_width, #2
5546 add texture_offset, texture_offset, #16
5547 add fb_ptr, fb_ptr, #16
5549 vmov.u8 draw_mask_fb_ptr, #0
5551 add block, block, #24
5555 and texture_block_ptr, texture_offset, texture_mask
5556 subs blocks_remaining, blocks_remaining, #1
5558 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5559 vld1.u32 { texels }, [texture_block_ptr, :128]
5561 vst1.u32 { texels }, [block, :128]
5562 add block, block, #40
5564 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5567 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5569 add texture_offset, texture_offset, #16
5570 add fb_ptr, fb_ptr, #16
5572 add block, block, #24
5576 and texture_block_ptr, texture_offset, texture_mask
5577 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5579 vld1.u32 { texels }, [texture_block_ptr, :128]
5580 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5582 vst1.u32 { texels }, [block, :128]
5583 add block, block, #40
5585 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5586 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5588 add block, block, #24
5589 subs height, height, #1
5591 add fb_ptr, fb_ptr, fb_ptr_pitch
5592 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5596 ldmia sp!, { r4 - r11, pc }
5600 // FIXME: duplicate code with normal version :(
5601 #undef draw_mask_fb_ptr
5603 function(setup_sprite_16bpp_4x)
5604 stmdb sp!, { r4 - r11, r14 }
5605 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5608 add fb_ptr, fb_ptr, y, lsl #11
5610 ldr width, [sp, #40]
5611 add fb_ptr, fb_ptr, x, lsl #1
5613 ldr height, [sp, #44]
5614 and left_offset, u, #0x7
5616 add texture_offset_base, u, u
5617 add width_rounded, width, #7
5619 add texture_offset_base, texture_offset_base, v, lsl #11
5620 movw left_mask_bits, #0xFFFF
5622 ldrb texture_mask_width, [psx_gpu, #psx_gpu_texture_mask_width_offset]
5623 add width_rounded, width_rounded, left_offset
5627 ldrb texture_mask_height, [psx_gpu, #psx_gpu_texture_mask_height_offset]
5628 sub fb_ptr, fb_ptr, left_offset, lsl #1
5630 add texture_mask, texture_mask_width, texture_mask_width
5631 movw right_mask_bits, #0xFFFC
5633 and right_width, width_rounded, #0x7
5634 mvn left_mask_bits, left_mask_bits, lsl left_offset
5638 add texture_mask, texture_mask, texture_mask_height, lsl #11
5639 mov block_width, width_rounded, lsr #3
5641 mov right_mask_bits, right_mask_bits, lsl right_width
5642 movw fb_ptr_pitch, #(2048 + 16) * 2
5644 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4+1
5645 vmov block_masks, left_mask_bits, right_mask_bits
5647 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5648 add block, psx_gpu, #psx_gpu_blocks_offset
5650 bic texture_offset_base, texture_offset_base, #0xF
5653 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
5654 add block, block, num_blocks, lsl #6
5659 vext.32 block_masks_shifted, block_masks, block_masks, #1
5660 vorr.u32 block_masks, block_masks, block_masks_shifted
5661 vdup.u8 draw_mask_fb_ptr_a, block_masks[0]
5662 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5665 add num_blocks, num_blocks, block_width
5666 cmp num_blocks, #MAX_BLOCKS
5667 blgt setup_sprites_16bpp_flush
5669 and texture_block_ptr, texture_offset_base, texture_mask
5670 subs height, height, #1
5672 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5673 vld1.u32 { texels }, [texture_block_ptr, :128]
5675 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5677 add texture_offset_base, texture_offset_base, #2048
5678 add fb_ptr, fb_ptr, #2048*2
5679 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5682 ldmia sp!, { r4 - r11, pc }
5685 add num_blocks, num_blocks, block_width
5686 mov texture_offset, texture_offset_base
5688 vdup.u8 draw_mask_fb_ptr_a, block_masks[0] // left_mask_bits
5689 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5691 cmp num_blocks, #MAX_BLOCKS
5692 blgt setup_sprites_16bpp_flush
5694 add texture_offset_base, texture_offset_base, #2048
5695 and texture_block_ptr, texture_offset, texture_mask
5697 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5698 vld1.u32 { texels }, [texture_block_ptr, :128]
5700 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5702 subs blocks_remaining, block_width, #2*4
5703 add texture_offset, texture_offset, #16
5705 vmov.u8 draw_mask_fb_ptr_a, #0
5706 vmov.u8 draw_mask_fb_ptr_b, #0
5708 add fb_ptr, fb_ptr, #16*2
5712 and texture_block_ptr, texture_offset, texture_mask
5713 subs blocks_remaining, blocks_remaining, #4
5715 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5716 vld1.u32 { texels }, [texture_block_ptr, :128]
5718 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5719 add texture_offset, texture_offset, #16
5721 add fb_ptr, fb_ptr, #16*2
5725 vdup.u8 draw_mask_fb_ptr_a, block_masks[4] // right_mask_bits
5726 vdup.u8 draw_mask_fb_ptr_b, block_masks[5]
5728 and texture_block_ptr, texture_offset, texture_mask
5729 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5731 vld1.u32 { texels }, [texture_block_ptr, :128]
5733 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5734 subs height, height, #1
5736 add fb_ptr, fb_ptr, fb_ptr_pitch
5737 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5741 ldmia sp!, { r4 - r11, pc }
5746 #undef right_mask_bits
5749 #undef blocks_remaining
5759 #define right_width r5
5760 #define right_mask_bits r6
5764 #define fb_ptr_pitch r12
5766 // referenced by setup_sprites_16bpp_flush
5767 #define num_blocks r4
5769 #define block_width r11
5774 #define blocks_remaining r6
5777 #define right_mask q1
5778 #define test_mask q2
5779 #define draw_mask q2
5780 #define draw_mask_bits_fb_ptr d6
5785 function(setup_sprite_untextured)
5786 ldrh r12, [psx_gpu, #psx_gpu_render_state_offset]
5787 tst r12, #(RENDER_STATE_MASK_EVALUATE | RENDER_FLAGS_MODULATE_TEXELS \
5788 | RENDER_FLAGS_BLEND)
5789 ldrbeq r12, [psx_gpu, #psx_gpu_render_mode_offset]
5790 tsteq r12, #RENDER_INTERLACE_ENABLED
5791 beq setup_sprite_untextured_simple
5793 stmdb sp!, { r4 - r11, r14 }
5795 ldr width, [sp, #40]
5796 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5798 ldr height, [sp, #44]
5799 add fb_ptr, fb_ptr, y, lsl #11
5801 add fb_ptr, fb_ptr, x, lsl #1
5802 sub right_width, width, #1
5804 ldr color, [sp, #48]
5807 add block_width, width, #7
5811 mov right_mask_bits, #0xff
5813 sub fb_ptr_pitch, block_width, #1
5814 lsl right_mask_bits, right_width
5816 lsl fb_ptr_pitch, #3+1
5817 ubfx color_r, color, #3, #5
5819 rsb fb_ptr_pitch, #1024*2
5820 ubfx color_g, color, #11, #5
5822 vld1.u32 { test_mask }, [psx_gpu, :128]
5823 ubfx color_b, color, #19, #5
5825 vdup.u16 right_mask, right_mask_bits
5826 orr color, color_r, color_b, lsl #10
5828 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5829 orr color, color, color_g, lsl #5
5831 vtst.u16 right_mask, right_mask, test_mask
5832 add block, psx_gpu, #psx_gpu_blocks_offset
5834 vdup.u16 colors, color
5835 add block, block, num_blocks, lsl #6
5838 setup_sprite_untextured_height_loop:
5839 add num_blocks, block_width
5840 sub blocks_remaining, block_width, #1
5842 cmp num_blocks, #MAX_BLOCKS
5843 blgt setup_sprites_16bpp_flush
5845 cmp blocks_remaining, #0
5848 vmov.u8 draw_mask, #0 /* zero_mask */
5849 vmov.u8 draw_mask_bits_fb_ptr, #0
5852 vst1.u32 { draw_mask }, [block, :128]!
5853 subs blocks_remaining, #1
5855 vst1.u32 { colors }, [block, :128]
5856 add block, block, #24
5858 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5859 vst1.u32 { draw_mask_bits_fb_ptr }, [block, :64]
5861 add block, block, #24
5866 vst1.u32 { right_mask }, [block, :128]!
5869 vst1.u32 { colors }, [block, :128]
5870 add block, block, #24
5872 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5873 vst1.u32 { draw_mask_bits_fb_ptr }, [block, :64]
5875 add block, block, #24
5876 add fb_ptr, fb_ptr_pitch
5878 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5879 bgt setup_sprite_untextured_height_loop
5881 ldmia sp!, { r4 - r11, pc }
5885 #undef texture_page_ptr
5887 #undef dirty_textures_mask
5888 #undef current_texture_mask
5891 #define current_texture_page r1
5892 #define texture_page_ptr r2
5893 #define vram_ptr_a r3
5894 #define current_texture_page_x r12
5895 #define current_texture_page_y r4
5896 #define dirty_textures_mask r5
5900 #define current_texture_mask r9
5902 #define vram_ptr_b r11
5904 #define texel_block_a d0
5905 #define texel_block_b d1
5906 #define texel_block_expanded_a q1
5907 #define texel_block_expanded_b q2
5908 #define texel_block_expanded_ab q2
5909 #define texel_block_expanded_c q3
5910 #define texel_block_expanded_d q4
5911 #define texel_block_expanded_cd q3
5913 function(update_texture_4bpp_cache)
5914 stmdb sp!, { r4 - r11, r14 }
5917 ldrb current_texture_page, [psx_gpu, #psx_gpu_current_texture_page_offset]
5919 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_base_offset]
5920 ldr vram_ptr_a, [psx_gpu, #psx_gpu_vram_ptr_offset]
5922 and current_texture_page_x, current_texture_page, #0xF
5923 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
5925 mov current_texture_page_y, current_texture_page, lsr #4
5926 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
5928 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5931 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5932 bic dirty_textures_mask, current_texture_mask
5935 str dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
5940 add vram_ptr_b, vram_ptr_a, #2048
5943 vld1.u32 { texel_block_a }, [vram_ptr_a, :64], c_4096
5944 vld1.u32 { texel_block_b }, [vram_ptr_b, :64], c_4096
5946 vmovl.u8 texel_block_expanded_a, texel_block_a
5947 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5948 vmovl.u8 texel_block_expanded_c, texel_block_b
5949 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5951 vbic.u16 texel_block_expanded_a, #0x00F0
5952 vbic.u16 texel_block_expanded_b, #0x00F0
5953 vbic.u16 texel_block_expanded_c, #0x00F0
5954 vbic.u16 texel_block_expanded_d, #0x00F0
5956 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
5957 texel_block_expanded_b
5958 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
5959 texel_block_expanded_d
5961 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
5962 [texture_page_ptr, :256]!
5964 subs sub_y, sub_y, #1
5968 add vram_ptr_a, vram_ptr_a, #8
5969 add vram_ptr_b, vram_ptr_b, #8
5971 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
5972 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
5974 subs tile_x, tile_x, #1
5978 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
5979 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
5981 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
5982 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
5984 subs tile_y, tile_y, #1
5988 ldmia sp!, { r4 - r11, pc }
5991 #undef current_texture_page
5994 #define texture_page r1
5995 #define texture_page_ptr r2
5996 #define vram_ptr_a r3
5997 #define texture_page_x r12
5998 #define texture_page_y r4
5999 #define current_texture_page r5
6004 #define vram_ptr_b r11
6016 function(update_texture_8bpp_cache_slice)
6017 stmdb sp!, { r4 - r11, r14 }
6020 ldrb current_texture_page, [psx_gpu, #psx_gpu_current_texture_page_offset]
6021 ldr vram_ptr_a, [psx_gpu, #psx_gpu_vram_ptr_offset]
6023 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_base_offset]
6026 and texture_page_x, texture_page, #0xF
6027 mov texture_page_y, texture_page, lsr #4
6029 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
6032 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
6033 eor current_texture_page, current_texture_page, texture_page
6035 ands current_texture_page, current_texture_page, #0x1
6038 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6041 add vram_ptr_b, vram_ptr_a, #2048
6044 vld1.u32 { texels_a }, [vram_ptr_a, :128], c_4096
6045 vld1.u32 { texels_b }, [vram_ptr_b, :128], c_4096
6046 vld1.u32 { texels_c }, [vram_ptr_a, :128], c_4096
6047 vld1.u32 { texels_d }, [vram_ptr_b, :128], c_4096
6049 vst1.u32 { texels_a, texels_b }, [texture_page_ptr, :256]!
6050 vst1.u32 { texels_c, texels_d }, [texture_page_ptr, :256]!
6052 subs sub_y, sub_y, #1
6057 add vram_ptr_a, vram_ptr_a, #16
6058 add vram_ptr_b, vram_ptr_b, #16
6060 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
6061 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
6063 subs tile_x, tile_x, #1
6068 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
6069 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
6071 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
6072 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
6074 subs tile_y, tile_y, #1
6075 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6080 ldmia sp!, { r4 - r11, pc }
6083 /* void scale2x_tiles8(void *dst, const void *src, int w8, int h) */
6084 function(scale2x_tiles8)
6088 add r12, r0, #1024*2
6092 vld1.u16 { q0 }, [r1, :128]!
6093 vld1.u16 { q2 }, [r1, :128]!
6099 vst1.u16 { q0, q1 }, [r0, :128]!
6100 vst1.u16 { q0, q1 }, [r12, :128]!
6102 vst1.u16 { q2, q3 }, [r0, :128]!
6103 vst1.u16 { q2, q3 }, [r12, :128]!
6110 sub r0, r0, r2, lsl #4+1
6112 add r12, r0, #1024*2
6118 // vim:filetype=armasm