[pcsx_rearmed.git] / frontend / cspace_neon.S

/*
 * (C) Gražvydas "notaz" Ignotas, 2010
 *
 * This work is licensed under the terms of any of these licenses
 * (at your option):
 *  - GNU GPL, version 2 or later.
 *  - GNU LGPL, version 2.1 or later.
 * See the COPYING file in the top-level directory.
 */

#include "arm_features.h"

.text
.align 2

FUNCTION(bgr555_to_rgb565):
    pld         [r1]
    mov         r3, #0x07c0
    vdup.16     q15, r3
    subs        r2, r2, #64
    blt         btr16_end64
0:
    pld         [r1, #64*2]
    vldmia      r1!, {q0-q3}
    vshl.u16    q4, q0, #11
    vshl.u16    q5, q1, #11
    vshl.u16    q6, q2, #11
    vshl.u16    q7, q3, #11
    vsri.u16    q4, q0, #10
    vsri.u16    q5, q1, #10
    vsri.u16    q6, q2, #10
    vsri.u16    q7, q3, #10
    vshl.u16    q0, q0, #1
    vshl.u16    q1, q1, #1
    vshl.u16    q2, q2, #1
    vshl.u16    q3, q3, #1
    vbit        q4, q0, q15
    vbit        q5, q1, q15
    vbit        q6, q2, q15
    vbit        q7, q3, q15
    vstmia      r0!, {q4-q7}
    subs        r2, r2, #64
    bge         0b

btr16_end64:
    adds        r2, r2, #64
    bxeq        lr
    subs        r2, r2, #16
    blt         btr16_end16

    @ handle the remainder (reasonably rare)
0:
    vld1.16     {q0}, [r1]!
    vshl.u16    q1, q0, #11
    vshl.u16    q2, q0, #1
    vsri.u16    q1, q0, #10
    vbit        q1, q2, q15
    subs        r2, r2, #16
    vst1.16     {q1}, [r0]!
    bge         0b

btr16_end16:
    adds        r2, r2, #16
    bxeq        lr
    subs        r2, r2, #8
    bxlt        lr

    @ very rare
    vld1.16     d0, [r1]!
    vshl.u16    d1, d0, #11
    vshl.u16    d2, d0, #1
    vsri.u16    d1, d0, #10
    vbit        d1, d2, d30
    vst1.16     d1, [r0]!
    bx          lr


FUNCTION(bgr888_to_rgb888):
    pld         [r1]
    @ r2 /= 48
    mov         r2, r2, lsr #4
    movw        r3, #0x5556
    movt        r3, #0x5555
    umull       r12,r2, r3, r2
0:
    pld         [r1, #48*3]
    vld3.8      {d0-d2}, [r1, :64]!
    vld3.8      {d3-d5}, [r1, :64]!
    vswp        d0, d2
    vswp        d3, d5
    vst3.8      {d0-d2}, [r0, :64]!
    vst3.8      {d3-d5}, [r0, :64]!
    subs        r2, r2, #1
    bne         0b

    bx          lr


FUNCTION(bgr888_to_rgb565):
    pld         [r1]
    @ r2 /= 48
    mov         r2, r2, lsr #4
    movw        r3, #0x5556
    movt        r3, #0x5555
    umull       r12,r2, r3, r2

    mov         r3, #0x07e0
    vdup.16     q15, r3
0:
    pld         [r1, #48*3]
    vld3.8      {d1-d3}, [r1, :64]!
    vld3.8      {d5-d7}, [r1, :64]!

    vshll.u8    q8, d2, #3      @ g
    vshll.u8    q9, d6, #3
    vshr.u8     d0, d3, #3      @ b
    vshr.u8     d4, d7, #3
    vzip.8      d0, d1          @ rb
    vzip.8      d4, d5
    vbit        q0, q8, q15
    vbit        q2, q9, q15

    vstmia      r0!, {d0,d1}
    vstmia      r0!, {d4,d5}
    subs        r2, r2, #1
    bne         0b

    bx          lr


FUNCTION(rgb888_to_rgb565):
    pld         [r1]
    @ r2 /= 48
    mov         r2, r2, lsr #4
    movw        r3, #0x5556
    movt        r3, #0x5555
    umull       r12,r2, r3, r2

    mov         r3, #0x07e0
    vdup.16     q15, r3
0:
    pld         [r1, #48*3]
    vld3.8      {d1-d3}, [r1, :64]!
    vld3.8      {d5-d7}, [r1, :64]!

    vshll.u8    q8, d2, #3      @ g
    vshll.u8    q9, d6, #3
    vshr.u8     d2, d1, #3      @ b
    vshr.u8     d6, d5, #3
    vzip.8      d2, d3          @ rb
    vzip.8      d6, d7
    vbit        q1, q8, q15
    vbit        q3, q9, q15

    vstmia      r0!, {d2,d3}
    vstmia      r0!, {d6,d7}
    subs        r2, r2, #1
    bne         0b

    bx          lr


@ vim:filetype=armasm
Commit	Line	Data
	1	/*
	2	* (C) Gražvydas "notaz" Ignotas, 2010
	3	*
	4	* This work is licensed under the terms of any of these licenses
	5	* (at your option):
	6	* - GNU GPL, version 2 or later.
	7	* - GNU LGPL, version 2.1 or later.
	8	* See the COPYING file in the top-level directory.
	9	*/
	10
	11	#include "arm_features.h"
	12
	13	.text
	14	.align 2
	15
	16	FUNCTION(bgr555_to_rgb565):
	17	pld [r1]
	18	mov r3, #0x07c0
	19	vdup.16 q15, r3
	20	subs r2, r2, #64
	21	blt btr16_end64
	22	0:
	23	pld [r1, #64*2]
	24	vldmia r1!, {q0-q3}
	25	vshl.u16 q4, q0, #11
	26	vshl.u16 q5, q1, #11
	27	vshl.u16 q6, q2, #11
	28	vshl.u16 q7, q3, #11
	29	vsri.u16 q4, q0, #10
	30	vsri.u16 q5, q1, #10
	31	vsri.u16 q6, q2, #10
	32	vsri.u16 q7, q3, #10
	33	vshl.u16 q0, q0, #1
	34	vshl.u16 q1, q1, #1
	35	vshl.u16 q2, q2, #1
	36	vshl.u16 q3, q3, #1
	37	vbit q4, q0, q15
	38	vbit q5, q1, q15
	39	vbit q6, q2, q15
	40	vbit q7, q3, q15
	41	vstmia r0!, {q4-q7}
	42	subs r2, r2, #64
	43	bge 0b
	44
	45	btr16_end64:
	46	adds r2, r2, #64
	47	bxeq lr
	48	subs r2, r2, #16
	49	blt btr16_end16
	50
	51	@ handle the remainder (reasonably rare)
	52	0:
	53	vld1.16 {q0}, [r1]!
	54	vshl.u16 q1, q0, #11
	55	vshl.u16 q2, q0, #1
	56	vsri.u16 q1, q0, #10
	57	vbit q1, q2, q15
	58	subs r2, r2, #16
	59	vst1.16 {q1}, [r0]!
	60	bge 0b
	61
	62	btr16_end16:
	63	adds r2, r2, #16
	64	bxeq lr
	65	subs r2, r2, #8
	66	bxlt lr
	67
	68	@ very rare
	69	vld1.16 d0, [r1]!
	70	vshl.u16 d1, d0, #11
	71	vshl.u16 d2, d0, #1
	72	vsri.u16 d1, d0, #10
	73	vbit d1, d2, d30
	74	vst1.16 d1, [r0]!
	75	bx lr
	76
	77
	78	FUNCTION(bgr888_to_rgb888):
	79	pld [r1]
	80	@ r2 /= 48
	81	mov r2, r2, lsr #4
	82	movw r3, #0x5556
	83	movt r3, #0x5555
	84	umull r12,r2, r3, r2
	85	0:
	86	pld [r1, #48*3]
	87	vld3.8 {d0-d2}, [r1, :64]!
	88	vld3.8 {d3-d5}, [r1, :64]!
	89	vswp d0, d2
	90	vswp d3, d5
	91	vst3.8 {d0-d2}, [r0, :64]!
	92	vst3.8 {d3-d5}, [r0, :64]!
	93	subs r2, r2, #1
	94	bne 0b
	95
	96	bx lr
	97
	98
	99	FUNCTION(bgr888_to_rgb565):
	100	pld [r1]
	101	@ r2 /= 48
	102	mov r2, r2, lsr #4
	103	movw r3, #0x5556
	104	movt r3, #0x5555
	105	umull r12,r2, r3, r2
	106
	107	mov r3, #0x07e0
	108	vdup.16 q15, r3
	109	0:
	110	pld [r1, #48*3]
	111	vld3.8 {d1-d3}, [r1, :64]!
	112	vld3.8 {d5-d7}, [r1, :64]!
	113
	114	vshll.u8 q8, d2, #3 @ g
	115	vshll.u8 q9, d6, #3
	116	vshr.u8 d0, d3, #3 @ b
	117	vshr.u8 d4, d7, #3
	118	vzip.8 d0, d1 @ rb
	119	vzip.8 d4, d5
	120	vbit q0, q8, q15
	121	vbit q2, q9, q15
	122
	123	vstmia r0!, {d0,d1}
	124	vstmia r0!, {d4,d5}
	125	subs r2, r2, #1
	126	bne 0b
	127
	128	bx lr
	129
	130
	131	FUNCTION(rgb888_to_rgb565):
	132	pld [r1]
	133	@ r2 /= 48
	134	mov r2, r2, lsr #4
	135	movw r3, #0x5556
	136	movt r3, #0x5555
	137	umull r12,r2, r3, r2
	138
	139	mov r3, #0x07e0
	140	vdup.16 q15, r3
	141	0:
	142	pld [r1, #48*3]
	143	vld3.8 {d1-d3}, [r1, :64]!
	144	vld3.8 {d5-d7}, [r1, :64]!
	145
	146	vshll.u8 q8, d2, #3 @ g
	147	vshll.u8 q9, d6, #3
	148	vshr.u8 d2, d1, #3 @ b
	149	vshr.u8 d6, d5, #3
	150	vzip.8 d2, d3 @ rb
	151	vzip.8 d6, d7
	152	vbit q1, q8, q15
	153	vbit q3, q9, q15
	154
	155	vstmia r0!, {d2,d3}
	156	vstmia r0!, {d6,d7}
	157	subs r2, r2, #1
	158	bne 0b
	159
	160	bx lr
	161
	162
	163	@ vim:filetype=armasm