Commit c1ba6860 authored by Johann's avatar Johann Committed by Gerrit Code Review
Browse files

Merge "VP8 for ARMv8 by using NEON intrinsics 09"

parents 1b91fa8a 55c95f2d
;
; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
;
; Use of this source code is governed by a BSD-style license
; that can be found in the LICENSE file in the root of the source
; tree. An additional intellectual property rights grant can be found
; in the file PATENTS. All contributing project authors may
; be found in the AUTHORS file in the root of the source tree.
;
EXPORT |vp8_loop_filter_bhs_neon|
EXPORT |vp8_loop_filter_mbhs_neon|
ARM
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
; r0 unsigned char *s, PRESERVE
; r1 int p, PRESERVE
; q1 limit, PRESERVE
|vp8_loop_filter_simple_horizontal_edge_neon| PROC
vpush {d8-d15}
sub r3, r0, r1, lsl #1 ; move src pointer down by 2 lines
vld1.u8 {q7}, [r0@128], r1 ; q0
vld1.u8 {q5}, [r3@128], r1 ; p0
vld1.u8 {q8}, [r0@128] ; q1
vld1.u8 {q6}, [r3@128] ; p1
vabd.u8 q15, q6, q7 ; abs(p0 - q0)
vabd.u8 q14, q5, q8 ; abs(p1 - q1)
vqadd.u8 q15, q15, q15 ; abs(p0 - q0) * 2
vshr.u8 q14, q14, #1 ; abs(p1 - q1) / 2
vmov.u8 q0, #0x80 ; 0x80
vmov.s16 q13, #3
vqadd.u8 q15, q15, q14 ; abs(p0 - q0) * 2 + abs(p1 - q1) / 2
veor q7, q7, q0 ; qs0: q0 offset to convert to a signed value
veor q6, q6, q0 ; ps0: p0 offset to convert to a signed value
veor q5, q5, q0 ; ps1: p1 offset to convert to a signed value
veor q8, q8, q0 ; qs1: q1 offset to convert to a signed value
vcge.u8 q15, q1, q15 ; (abs(p0 - q0)*2 + abs(p1-q1)/2 > limit)*-1
vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
vsubl.s8 q3, d15, d13
vqsub.s8 q4, q5, q8 ; q4: vp8_filter = vp8_signed_char_clamp(ps1-qs1)
vmul.s16 q2, q2, q13 ; 3 * ( qs0 - ps0)
vmul.s16 q3, q3, q13
vmov.u8 q10, #0x03 ; 0x03
vmov.u8 q9, #0x04 ; 0x04
vaddw.s8 q2, q2, d8 ; vp8_filter + 3 * ( qs0 - ps0)
vaddw.s8 q3, q3, d9
vqmovn.s16 d8, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
vqmovn.s16 d9, q3
vand q14, q4, q15 ; vp8_filter &= mask
vqadd.s8 q2, q14, q10 ; Filter2 = vp8_signed_char_clamp(vp8_filter+3)
vqadd.s8 q3, q14, q9 ; Filter1 = vp8_signed_char_clamp(vp8_filter+4)
vshr.s8 q2, q2, #3 ; Filter2 >>= 3
vshr.s8 q4, q3, #3 ; Filter1 >>= 3
sub r0, r0, r1
;calculate output
vqadd.s8 q11, q6, q2 ; u = vp8_signed_char_clamp(ps0 + Filter2)
vqsub.s8 q10, q7, q4 ; u = vp8_signed_char_clamp(qs0 - Filter1)
veor q6, q11, q0 ; *op0 = u^0x80
veor q7, q10, q0 ; *oq0 = u^0x80
vst1.u8 {q6}, [r3@128] ; store op0
vst1.u8 {q7}, [r0@128] ; store oq0
vpop {d8-d15}
bx lr
ENDP ; |vp8_loop_filter_simple_horizontal_edge_neon|
; r0 unsigned char *y
; r1 int ystride
; r2 const unsigned char *blimit
|vp8_loop_filter_bhs_neon| PROC
push {r4, lr}
ldrb r3, [r2] ; load blim from mem
vdup.s8 q1, r3 ; duplicate blim
add r0, r0, r1, lsl #2 ; src = y_ptr + 4 * y_stride
bl vp8_loop_filter_simple_horizontal_edge_neon
; vp8_loop_filter_simple_horizontal_edge_neon preserves r0, r1 and q1
add r0, r0, r1, lsl #2 ; src = y_ptr + 8* y_stride
bl vp8_loop_filter_simple_horizontal_edge_neon
add r0, r0, r1, lsl #2 ; src = y_ptr + 12 * y_stride
pop {r4, lr}
b vp8_loop_filter_simple_horizontal_edge_neon
ENDP ;|vp8_loop_filter_bhs_neon|
; r0 unsigned char *y
; r1 int ystride
; r2 const unsigned char *blimit
|vp8_loop_filter_mbhs_neon| PROC
ldrb r3, [r2] ; load blim from mem
vdup.s8 q1, r3 ; duplicate mblim
b vp8_loop_filter_simple_horizontal_edge_neon
ENDP ;|vp8_loop_filter_bhs_neon|
END
/*
* Copyright (c) 2014 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include <arm_neon.h>
static inline void vp8_loop_filter_simple_horizontal_edge_neon(
unsigned char *s,
int p,
const unsigned char *blimit) {
uint8_t *sp;
uint8x16_t qblimit, q0u8;
uint8x16_t q5u8, q6u8, q7u8, q8u8, q9u8, q10u8, q14u8, q15u8;
int16x8_t q2s16, q3s16, q13s16;
int8x8_t d8s8, d9s8;
int8x16_t q2s8, q3s8, q4s8, q10s8, q11s8, q14s8;
qblimit = vdupq_n_u8(*blimit);
sp = s - (p << 1);
q5u8 = vld1q_u8(sp);
sp += p;
q6u8 = vld1q_u8(sp);
sp += p;
q7u8 = vld1q_u8(sp);
sp += p;
q8u8 = vld1q_u8(sp);
q15u8 = vabdq_u8(q6u8, q7u8);
q14u8 = vabdq_u8(q5u8, q8u8);
q15u8 = vqaddq_u8(q15u8, q15u8);
q14u8 = vshrq_n_u8(q14u8, 1);
q0u8 = vdupq_n_u8(0x80);
q13s16 = vdupq_n_s16(3);
q15u8 = vqaddq_u8(q15u8, q14u8);
q5u8 = veorq_u8(q5u8, q0u8);
q6u8 = veorq_u8(q6u8, q0u8);
q7u8 = veorq_u8(q7u8, q0u8);
q8u8 = veorq_u8(q8u8, q0u8);
q15u8 = vcgeq_u8(qblimit, q15u8);
q2s16 = vsubl_s8(vget_low_s8(vreinterpretq_s8_u8(q7u8)),
vget_low_s8(vreinterpretq_s8_u8(q6u8)));
q3s16 = vsubl_s8(vget_high_s8(vreinterpretq_s8_u8(q7u8)),
vget_high_s8(vreinterpretq_s8_u8(q6u8)));
q4s8 = vqsubq_s8(vreinterpretq_s8_u8(q5u8),
vreinterpretq_s8_u8(q8u8));
q2s16 = vmulq_s16(q2s16, q13s16);
q3s16 = vmulq_s16(q3s16, q13s16);
q10u8 = vdupq_n_u8(3);
q9u8 = vdupq_n_u8(4);
q2s16 = vaddw_s8(q2s16, vget_low_s8(q4s8));
q3s16 = vaddw_s8(q3s16, vget_high_s8(q4s8));
d8s8 = vqmovn_s16(q2s16);
d9s8 = vqmovn_s16(q3s16);
q4s8 = vcombine_s8(d8s8, d9s8);
q14s8 = vandq_s8(q4s8, vreinterpretq_s8_u8(q15u8));
q2s8 = vqaddq_s8(q14s8, vreinterpretq_s8_u8(q10u8));
q3s8 = vqaddq_s8(q14s8, vreinterpretq_s8_u8(q9u8));
q2s8 = vshrq_n_s8(q2s8, 3);
q3s8 = vshrq_n_s8(q3s8, 3);
q11s8 = vqaddq_s8(vreinterpretq_s8_u8(q6u8), q2s8);
q10s8 = vqsubq_s8(vreinterpretq_s8_u8(q7u8), q3s8);
q6u8 = veorq_u8(vreinterpretq_u8_s8(q11s8), q0u8);
q7u8 = veorq_u8(vreinterpretq_u8_s8(q10s8), q0u8);
vst1q_u8(s, q7u8);
s -= p;
vst1q_u8(s, q6u8);
return;
}
void vp8_loop_filter_bhs_neon(
unsigned char *y_ptr,
int y_stride,
const unsigned char *blimit) {
y_ptr += y_stride * 4;
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, blimit);
y_ptr += y_stride * 4;
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, blimit);
y_ptr += y_stride * 4;
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, blimit);
return;
}
void vp8_loop_filter_mbhs_neon(
unsigned char *y_ptr,
int y_stride,
const unsigned char *blimit) {
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, blimit);
return;
}
......@@ -159,7 +159,6 @@ VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/vp8_variance_halfpixvar16x16_
VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/vp8_variance_halfpixvar16x16_hv_armv6$(ASM)
# common (neon)
VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/loopfiltersimplehorizontaledge_neon$(ASM)
VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/loopfiltersimpleverticaledge_neon$(ASM)
VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/mbloopfilter_neon$(ASM)
VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/shortidct4x4llm_neon$(ASM)
......@@ -186,6 +185,7 @@ VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/dequantizeb_neon.c
VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/idct_dequant_full_2x_neon.c
VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/iwalsh_neon.c
VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/loopfilter_neon.c
VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/loopfiltersimplehorizontaledge_neon.c
$(eval $(call rtcd_h_template,vp8_rtcd,vp8/common/rtcd_defs.pl))
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment