ref: 63e140eaa714f2427eb99296aa551a79d078eac7
parent: c2ff1882ffe8cbb8499b55bea4e59e0facda0ba7
parent: 635ba269be143743f3513a3dc16db39d71b1945a
author: Johann <[email protected]>
date: Thu Aug 15 10:55:08 EDT 2013
Merge "vp9: neon: add vp9_convolve_copy_neon"
--- /dev/null
+++ b/vp9/common/arm/neon/vp9_copy_neon.asm
@@ -1,0 +1,84 @@
+;
+; Copyright (c) 2013 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 |vp9_convolve_copy_neon|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+|vp9_convolve_copy_neon| PROC
+ push {r4-r5, lr}+ ldrd r4, r5, [sp, #28]
+
+ cmp r4, #32
+ bgt copy64
+ beq copy32
+ cmp r4, #8
+ bgt copy16
+ beq copy8
+ b copy4
+
+copy64
+ sub lr, r1, #32
+ sub r3, r3, #32
+copy64_h
+ pld [r0, r1, lsl #1]
+ vld1.8 {q0-q1}, [r0]!+ vld1.8 {q2-q3}, [r0], lr+ vst1.8 {q0-q1}, [r2@128]!+ vst1.8 {q2-q3}, [r2@128], r3+ subs r5, r5, #1
+ bgt copy64_h
+ pop {r4-r5, pc}+
+copy32
+ pld [r0, r1, lsl #1]
+ vld1.8 {q0-q1}, [r0], r1+ pld [r0, r1, lsl #1]
+ vld1.8 {q2-q3}, [r0], r1+ vst1.8 {q0-q1}, [r2@128], r3+ vst1.8 {q2-q3}, [r2@128], r3+ subs r5, r5, #2
+ bgt copy32
+ pop {r4-r5, pc}+
+copy16
+ pld [r0, r1, lsl #1]
+ vld1.8 {q0}, [r0], r1+ pld [r0, r1, lsl #1]
+ vld1.8 {q1}, [r0], r1+ vst1.8 {q0}, [r2@128], r3+ vst1.8 {q1}, [r2@128], r3+ subs r5, r5, #2
+ bgt copy16
+ pop {r4-r5, pc}+
+copy8
+ pld [r0, r1, lsl #1]
+ vld1.8 {d0}, [r0], r1+ pld [r0, r1, lsl #1]
+ vld1.8 {d2}, [r0], r1+ vst1.8 {d0}, [r2@64], r3+ vst1.8 {d2}, [r2@64], r3+ subs r5, r5, #2
+ bgt copy8
+ pop {r4-r5, pc}+
+copy4
+ ldr r12, [r0], r1
+ str r12, [r2], r3
+ subs r5, r5, #1
+ bgt copy4
+ pop {r4-r5, pc}+ ENDP
+
+ END
--- a/vp9/common/vp9_rtcd_defs.sh
+++ b/vp9/common/vp9_rtcd_defs.sh
@@ -268,7 +268,7 @@
# Sub Pixel Filters
#
prototype void vp9_convolve_copy "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h"
-specialize vp9_convolve_copy $sse2_x86inc
+specialize vp9_convolve_copy $sse2_x86inc neon
prototype void vp9_convolve_avg "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h"
specialize vp9_convolve_avg $sse2_x86inc
--- a/vp9/vp9_common.mk
+++ b/vp9/vp9_common.mk
@@ -100,5 +100,6 @@
VP9_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/vp9_short_idct8x8_add_neon$(ASM)
VP9_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/vp9_short_idct16x16_add_neon$(ASM)
VP9_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/vp9_mb_lpf_neon$(ASM)
+VP9_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/vp9_copy_neon$(ASM)
$(eval $(call rtcd_h_template,vp9_rtcd,vp9/common/vp9_rtcd_defs.sh))