michael@0: ;
michael@0: ;  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
michael@0: ;
michael@0: ;  Use of this source code is governed by a BSD-style license
michael@0: ;  that can be found in the LICENSE file in the root of the source
michael@0: ;  tree. An additional intellectual property rights grant can be found
michael@0: ;  in the file PATENTS.  All contributing project authors may
michael@0: ;  be found in the AUTHORS file in the root of the source tree.
michael@0: ;
michael@0: 
michael@0: 
michael@0:     EXPORT |vp8_memcpy_partial_neon|
michael@0: 
michael@0:     ARM
michael@0:     REQUIRE8
michael@0:     PRESERVE8
michael@0: 
michael@0:     AREA ||.text||, CODE, READONLY, ALIGN=2
michael@0: ;=========================================
michael@0: ;this is not a full memcpy function!!!
michael@0: ;void vp8_memcpy_partial_neon(unsigned char *dst_ptr, unsigned char *src_ptr,
michael@0: ;                             int sz);
michael@0: |vp8_memcpy_partial_neon| PROC
michael@0:     ;pld                [r1]                        ;preload pred data
michael@0:     ;pld                [r1, #128]
michael@0:     ;pld                [r1, #256]
michael@0:     ;pld                [r1, #384]
michael@0: 
michael@0:     mov             r12, r2, lsr #8                 ;copy 256 bytes data at one time
michael@0: 
michael@0: memcpy_neon_loop
michael@0:     vld1.8          {q0, q1}, [r1]!                 ;load src data
michael@0:     subs            r12, r12, #1
michael@0:     vld1.8          {q2, q3}, [r1]!
michael@0:     vst1.8          {q0, q1}, [r0]!                 ;copy to dst_ptr
michael@0:     vld1.8          {q4, q5}, [r1]!
michael@0:     vst1.8          {q2, q3}, [r0]!
michael@0:     vld1.8          {q6, q7}, [r1]!
michael@0:     vst1.8          {q4, q5}, [r0]!
michael@0:     vld1.8          {q8, q9}, [r1]!
michael@0:     vst1.8          {q6, q7}, [r0]!
michael@0:     vld1.8          {q10, q11}, [r1]!
michael@0:     vst1.8          {q8, q9}, [r0]!
michael@0:     vld1.8          {q12, q13}, [r1]!
michael@0:     vst1.8          {q10, q11}, [r0]!
michael@0:     vld1.8          {q14, q15}, [r1]!
michael@0:     vst1.8          {q12, q13}, [r0]!
michael@0:     vst1.8          {q14, q15}, [r0]!
michael@0: 
michael@0:     ;pld                [r1]                        ;preload pred data -- need to adjust for real device
michael@0:     ;pld                [r1, #128]
michael@0:     ;pld                [r1, #256]
michael@0:     ;pld                [r1, #384]
michael@0: 
michael@0:     bne             memcpy_neon_loop
michael@0: 
michael@0:     ands            r3, r2, #0xff                   ;extra copy
michael@0:     beq             done_copy_neon_loop
michael@0: 
michael@0: extra_copy_neon_loop
michael@0:     vld1.8          {q0}, [r1]!                 ;load src data
michael@0:     subs            r3, r3, #16
michael@0:     vst1.8          {q0}, [r0]!
michael@0:     bne             extra_copy_neon_loop
michael@0: 
michael@0: done_copy_neon_loop
michael@0:     bx              lr
michael@0:     ENDP
michael@0: 
michael@0:     END