generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
e50c842755
vpx/vp8/encoder/arm/neon/subtract_neon.asm

200 lines
6.2 KiB
NASM
Raw Normal View History

Initial WebM release
2010-05-18 17:58:33 +02:00
;
Use WebM in copyright notice for consistency Changes 'The VP8 project' to 'The WebM project', for consistency with other webmproject.org repositories. Fixes issue #97. Change-Id: I37c13ed5fbdb9d334ceef71c6350e9febed9bbba
2010-09-09 14:16:39 +02:00
; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
Initial WebM release
2010-05-18 17:58:33 +02:00
;
cosmetics: trim trailing whitespace When the license headers were updated, they accidentally contained trailing whitespace, so unfortunately we have to touch all the files again. Change-Id: I236c05fade06589e417179c0444cb39b09e4200d
2010-06-18 18:39:21 +02:00
; Use of this source code is governed by a BSD-style license
LICENSE: update with latest text Change-Id: Ieebea089095d9073b3a94932791099f614ce120c
2010-06-04 22:19:40 +02:00
; that can be found in the LICENSE file in the root of the source
; tree. An additional intellectual property rights grant can be found
cosmetics: trim trailing whitespace When the license headers were updated, they accidentally contained trailing whitespace, so unfortunately we have to touch all the files again. Change-Id: I236c05fade06589e417179c0444cb39b09e4200d
2010-06-18 18:39:21 +02:00
; in the file PATENTS. All contributing project authors may
LICENSE: update with latest text Change-Id: Ieebea089095d9073b3a94932791099f614ce120c
2010-06-04 22:19:40 +02:00
; be found in the AUTHORS file in the root of the source tree.
Initial WebM release
2010-05-18 17:58:33 +02:00
;
Wrapper function removed from vp8_subtract_b_neon function call Address calculations moved from encodemb_arm.c file to neon optimized assembly function to save cycles in function calls. - vp8_subtract_b_neon_func replaced with vp8_subtract_b_neon that contains all needed address calculations - unnecessary file encodemb_arm.c removed - consistent with ARMv6 optimized version Change-Id: I6cbc1a2670b56c2077f59995fcf8f70786b4990b
2011-03-30 12:45:59 +02:00
EXPORT |vp8_subtract_b_neon|
Initial WebM release
2010-05-18 17:58:33 +02:00
EXPORT |vp8_subtract_mby_neon|
EXPORT |vp8_subtract_mbuv_neon|
Wrapper function removed from vp8_subtract_b_neon function call Address calculations moved from encodemb_arm.c file to neon optimized assembly function to save cycles in function calls. - vp8_subtract_b_neon_func replaced with vp8_subtract_b_neon that contains all needed address calculations - unnecessary file encodemb_arm.c removed - consistent with ARMv6 optimized version Change-Id: I6cbc1a2670b56c2077f59995fcf8f70786b4990b
2011-03-30 12:45:59 +02:00
INCLUDE asm_enc_offsets.asm
Initial WebM release
2010-05-18 17:58:33 +02:00
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
Wrapper function removed from vp8_subtract_b_neon function call Address calculations moved from encodemb_arm.c file to neon optimized assembly function to save cycles in function calls. - vp8_subtract_b_neon_func replaced with vp8_subtract_b_neon that contains all needed address calculations - unnecessary file encodemb_arm.c removed - consistent with ARMv6 optimized version Change-Id: I6cbc1a2670b56c2077f59995fcf8f70786b4990b
2011-03-30 12:45:59 +02:00
;void vp8_subtract_b_neon(BLOCK *be, BLOCKD *bd, int pitch)
|vp8_subtract_b_neon| PROC
stmfd sp!, {r4-r7}
ldr r3, [r0, #vp8_block_base_src]
ldr r4, [r0, #vp8_block_src]
ldr r5, [r0, #vp8_block_src_diff]
ldr r3, [r3]
ldr r6, [r0, #vp8_block_src_stride]
add r3, r3, r4 ; src = *base_src + src
ldr r7, [r1, #vp8_blockd_predictor]
vld1.8 {d0}, [r3], r6 ;load src
vld1.8 {d1}, [r7], r2 ;load pred
vld1.8 {d2}, [r3], r6
vld1.8 {d3}, [r7], r2
vld1.8 {d4}, [r3], r6
vld1.8 {d5}, [r7], r2
vld1.8 {d6}, [r3], r6
vld1.8 {d7}, [r7], r2
Initial WebM release
2010-05-18 17:58:33 +02:00
vsubl.u8 q10, d0, d1
vsubl.u8 q11, d2, d3
vsubl.u8 q12, d4, d5
vsubl.u8 q13, d6, d7
Wrapper function removed from vp8_subtract_b_neon function call Address calculations moved from encodemb_arm.c file to neon optimized assembly function to save cycles in function calls. - vp8_subtract_b_neon_func replaced with vp8_subtract_b_neon that contains all needed address calculations - unnecessary file encodemb_arm.c removed - consistent with ARMv6 optimized version Change-Id: I6cbc1a2670b56c2077f59995fcf8f70786b4990b
2011-03-30 12:45:59 +02:00
mov r2, r2, lsl #1
Initial WebM release
2010-05-18 17:58:33 +02:00
Wrapper function removed from vp8_subtract_b_neon function call Address calculations moved from encodemb_arm.c file to neon optimized assembly function to save cycles in function calls. - vp8_subtract_b_neon_func replaced with vp8_subtract_b_neon that contains all needed address calculations - unnecessary file encodemb_arm.c removed - consistent with ARMv6 optimized version Change-Id: I6cbc1a2670b56c2077f59995fcf8f70786b4990b
2011-03-30 12:45:59 +02:00
vst1.16 {d20}, [r5], r2 ;store diff
vst1.16 {d22}, [r5], r2
vst1.16 {d24}, [r5], r2
vst1.16 {d26}, [r5], r2
Initial WebM release
2010-05-18 17:58:33 +02:00
Wrapper function removed from vp8_subtract_b_neon function call Address calculations moved from encodemb_arm.c file to neon optimized assembly function to save cycles in function calls. - vp8_subtract_b_neon_func replaced with vp8_subtract_b_neon that contains all needed address calculations - unnecessary file encodemb_arm.c removed - consistent with ARMv6 optimized version Change-Id: I6cbc1a2670b56c2077f59995fcf8f70786b4990b
2011-03-30 12:45:59 +02:00
ldmfd sp!, {r4-r7}
Initial WebM release
2010-05-18 17:58:33 +02:00
bx lr
Wrapper function removed from vp8_subtract_b_neon function call Address calculations moved from encodemb_arm.c file to neon optimized assembly function to save cycles in function calls. - vp8_subtract_b_neon_func replaced with vp8_subtract_b_neon that contains all needed address calculations - unnecessary file encodemb_arm.c removed - consistent with ARMv6 optimized version Change-Id: I6cbc1a2670b56c2077f59995fcf8f70786b4990b
2011-03-30 12:45:59 +02:00
Initial WebM release
2010-05-18 17:58:33 +02:00
ENDP
Wrapper function removed from vp8_subtract_b_neon function call Address calculations moved from encodemb_arm.c file to neon optimized assembly function to save cycles in function calls. - vp8_subtract_b_neon_func replaced with vp8_subtract_b_neon that contains all needed address calculations - unnecessary file encodemb_arm.c removed - consistent with ARMv6 optimized version Change-Id: I6cbc1a2670b56c2077f59995fcf8f70786b4990b
2011-03-30 12:45:59 +02:00
Initial WebM release
2010-05-18 17:58:33 +02:00
;==========================================
Added predictor stride argument(s) to subtract functions Patch set 2: 64 bit build fix Patch set 3: 64 bit crash fix [Tero] Patch set 4: Updated ARMv6 and NEON assembly. Added also minor NEON optimizations to subtract functions. Patch set 5: x86 stride bug fix Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15
2011-11-15 18:53:01 +01:00
;void vp8_subtract_mby_neon(short *diff, unsigned char *src, int src_stride
; unsigned char *pred, int pred_stride)
Initial WebM release
2010-05-18 17:58:33 +02:00
|vp8_subtract_mby_neon| PROC
Added predictor stride argument(s) to subtract functions Patch set 2: 64 bit build fix Patch set 3: 64 bit crash fix [Tero] Patch set 4: Updated ARMv6 and NEON assembly. Added also minor NEON optimizations to subtract functions. Patch set 5: x86 stride bug fix Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15
2011-11-15 18:53:01 +01:00
push {r4-r7}
Initial WebM release
2010-05-18 17:58:33 +02:00
mov r12, #4
Added predictor stride argument(s) to subtract functions Patch set 2: 64 bit build fix Patch set 3: 64 bit crash fix [Tero] Patch set 4: Updated ARMv6 and NEON assembly. Added also minor NEON optimizations to subtract functions. Patch set 5: x86 stride bug fix Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15
2011-11-15 18:53:01 +01:00
ldr r4, [sp, #16] ; pred_stride
mov r6, #32 ; "diff" stride x2
add r5, r0, #16 ; second diff pointer
Initial WebM release
2010-05-18 17:58:33 +02:00
subtract_mby_loop
Added predictor stride argument(s) to subtract functions Patch set 2: 64 bit build fix Patch set 3: 64 bit crash fix [Tero] Patch set 4: Updated ARMv6 and NEON assembly. Added also minor NEON optimizations to subtract functions. Patch set 5: x86 stride bug fix Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15
2011-11-15 18:53:01 +01:00
vld1.8 {q0}, [r1], r2 ;load src
vld1.8 {q1}, [r3], r4 ;load pred
vld1.8 {q2}, [r1], r2
vld1.8 {q3}, [r3], r4
vld1.8 {q4}, [r1], r2
vld1.8 {q5}, [r3], r4
vld1.8 {q6}, [r1], r2
vld1.8 {q7}, [r3], r4
Initial WebM release
2010-05-18 17:58:33 +02:00
vsubl.u8 q8, d0, d2
vsubl.u8 q9, d1, d3
vsubl.u8 q10, d4, d6
vsubl.u8 q11, d5, d7
vsubl.u8 q12, d8, d10
vsubl.u8 q13, d9, d11
vsubl.u8 q14, d12, d14
vsubl.u8 q15, d13, d15
Added predictor stride argument(s) to subtract functions Patch set 2: 64 bit build fix Patch set 3: 64 bit crash fix [Tero] Patch set 4: Updated ARMv6 and NEON assembly. Added also minor NEON optimizations to subtract functions. Patch set 5: x86 stride bug fix Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15
2011-11-15 18:53:01 +01:00
vst1.16 {q8}, [r0], r6 ;store diff
vst1.16 {q9}, [r5], r6
vst1.16 {q10}, [r0], r6
vst1.16 {q11}, [r5], r6
vst1.16 {q12}, [r0], r6
vst1.16 {q13}, [r5], r6
vst1.16 {q14}, [r0], r6
vst1.16 {q15}, [r5], r6
Initial WebM release
2010-05-18 17:58:33 +02:00
subs r12, r12, #1
bne subtract_mby_loop
Added predictor stride argument(s) to subtract functions Patch set 2: 64 bit build fix Patch set 3: 64 bit crash fix [Tero] Patch set 4: Updated ARMv6 and NEON assembly. Added also minor NEON optimizations to subtract functions. Patch set 5: x86 stride bug fix Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15
2011-11-15 18:53:01 +01:00
pop {r4-r7}
Initial WebM release
2010-05-18 17:58:33 +02:00
bx lr
ENDP
;=================================
Added predictor stride argument(s) to subtract functions Patch set 2: 64 bit build fix Patch set 3: 64 bit crash fix [Tero] Patch set 4: Updated ARMv6 and NEON assembly. Added also minor NEON optimizations to subtract functions. Patch set 5: x86 stride bug fix Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15
2011-11-15 18:53:01 +01:00
;void vp8_subtract_mbuv_c(short *diff, unsigned char *usrc, unsigned char *vsrc,
; int src_stride, unsigned char *upred,
; unsigned char *vpred, int pred_stride)
Initial WebM release
2010-05-18 17:58:33 +02:00
|vp8_subtract_mbuv_neon| PROC
Added predictor stride argument(s) to subtract functions Patch set 2: 64 bit build fix Patch set 3: 64 bit crash fix [Tero] Patch set 4: Updated ARMv6 and NEON assembly. Added also minor NEON optimizations to subtract functions. Patch set 5: x86 stride bug fix Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15
2011-11-15 18:53:01 +01:00
push {r4-r7}
ldr r4, [sp, #16] ; upred
ldr r5, [sp, #20] ; vpred
ldr r6, [sp, #24] ; pred_stride
add r0, r0, #512 ; short *udiff = diff + 256;
mov r12, #32 ; "diff" stride x2
add r7, r0, #16 ; second diff pointer
Initial WebM release
2010-05-18 17:58:33 +02:00
;u
Added predictor stride argument(s) to subtract functions Patch set 2: 64 bit build fix Patch set 3: 64 bit crash fix [Tero] Patch set 4: Updated ARMv6 and NEON assembly. Added also minor NEON optimizations to subtract functions. Patch set 5: x86 stride bug fix Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15
2011-11-15 18:53:01 +01:00
vld1.8 {d0}, [r1], r3 ;load usrc
vld1.8 {d1}, [r4], r6 ;load upred
vld1.8 {d2}, [r1], r3
vld1.8 {d3}, [r4], r6
vld1.8 {d4}, [r1], r3
vld1.8 {d5}, [r4], r6
vld1.8 {d6}, [r1], r3
vld1.8 {d7}, [r4], r6
vld1.8 {d8}, [r1], r3
vld1.8 {d9}, [r4], r6
vld1.8 {d10}, [r1], r3
vld1.8 {d11}, [r4], r6
vld1.8 {d12}, [r1], r3
vld1.8 {d13}, [r4], r6
vld1.8 {d14}, [r1], r3
vld1.8 {d15}, [r4], r6
Initial WebM release
2010-05-18 17:58:33 +02:00
vsubl.u8 q8, d0, d1
vsubl.u8 q9, d2, d3
vsubl.u8 q10, d4, d5
vsubl.u8 q11, d6, d7
vsubl.u8 q12, d8, d9
vsubl.u8 q13, d10, d11
vsubl.u8 q14, d12, d13
vsubl.u8 q15, d14, d15
Added predictor stride argument(s) to subtract functions Patch set 2: 64 bit build fix Patch set 3: 64 bit crash fix [Tero] Patch set 4: Updated ARMv6 and NEON assembly. Added also minor NEON optimizations to subtract functions. Patch set 5: x86 stride bug fix Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15
2011-11-15 18:53:01 +01:00
vst1.16 {q8}, [r0], r12 ;store diff
vst1.16 {q9}, [r7], r12
vst1.16 {q10}, [r0], r12
vst1.16 {q11}, [r7], r12
vst1.16 {q12}, [r0], r12
vst1.16 {q13}, [r7], r12
vst1.16 {q14}, [r0], r12
vst1.16 {q15}, [r7], r12
Initial WebM release
2010-05-18 17:58:33 +02:00
;v
Added predictor stride argument(s) to subtract functions Patch set 2: 64 bit build fix Patch set 3: 64 bit crash fix [Tero] Patch set 4: Updated ARMv6 and NEON assembly. Added also minor NEON optimizations to subtract functions. Patch set 5: x86 stride bug fix Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15
2011-11-15 18:53:01 +01:00
vld1.8 {d0}, [r2], r3 ;load vsrc
vld1.8 {d1}, [r5], r6 ;load vpred
vld1.8 {d2}, [r2], r3
vld1.8 {d3}, [r5], r6
vld1.8 {d4}, [r2], r3
vld1.8 {d5}, [r5], r6
vld1.8 {d6}, [r2], r3
vld1.8 {d7}, [r5], r6
vld1.8 {d8}, [r2], r3
vld1.8 {d9}, [r5], r6
vld1.8 {d10}, [r2], r3
vld1.8 {d11}, [r5], r6
vld1.8 {d12}, [r2], r3
vld1.8 {d13}, [r5], r6
vld1.8 {d14}, [r2], r3
vld1.8 {d15}, [r5], r6
Initial WebM release
2010-05-18 17:58:33 +02:00
vsubl.u8 q8, d0, d1
vsubl.u8 q9, d2, d3
vsubl.u8 q10, d4, d5
vsubl.u8 q11, d6, d7
vsubl.u8 q12, d8, d9
vsubl.u8 q13, d10, d11
vsubl.u8 q14, d12, d13
vsubl.u8 q15, d14, d15
Added predictor stride argument(s) to subtract functions Patch set 2: 64 bit build fix Patch set 3: 64 bit crash fix [Tero] Patch set 4: Updated ARMv6 and NEON assembly. Added also minor NEON optimizations to subtract functions. Patch set 5: x86 stride bug fix Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15
2011-11-15 18:53:01 +01:00
vst1.16 {q8}, [r0], r12 ;store diff
vst1.16 {q9}, [r7], r12
vst1.16 {q10}, [r0], r12
vst1.16 {q11}, [r7], r12
vst1.16 {q12}, [r0], r12
vst1.16 {q13}, [r7], r12
vst1.16 {q14}, [r0], r12
vst1.16 {q15}, [r7], r12
Initial WebM release
2010-05-18 17:58:33 +02:00
Added predictor stride argument(s) to subtract functions Patch set 2: 64 bit build fix Patch set 3: 64 bit crash fix [Tero] Patch set 4: Updated ARMv6 and NEON assembly. Added also minor NEON optimizations to subtract functions. Patch set 5: x86 stride bug fix Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15
2011-11-15 18:53:01 +01:00
pop {r4-r7}
Initial WebM release
2010-05-18 17:58:33 +02:00
bx lr
Added predictor stride argument(s) to subtract functions Patch set 2: 64 bit build fix Patch set 3: 64 bit crash fix [Tero] Patch set 4: Updated ARMv6 and NEON assembly. Added also minor NEON optimizations to subtract functions. Patch set 5: x86 stride bug fix Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15
2011-11-15 18:53:01 +01:00
Initial WebM release
2010-05-18 17:58:33 +02:00
ENDP
END
Reference in New Issue Copy Permalink