generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
4decd27947
vpx/vp8/encoder/x86/quantize_mmx.asm

287 lines
7.9 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
;
%include "vpx_ports/x86_abi_support.asm"
;int vp8_fast_quantize_b_impl_mmx(short *coeff_ptr, short *zbin_ptr,
; short *qcoeff_ptr,short *dequant_ptr,
; short *scan_mask, short *round_ptr,
; short *quant_ptr, short *dqcoeff_ptr);
global sym(vp8_fast_quantize_b_impl_mmx)
sym(vp8_fast_quantize_b_impl_mmx):
push rbp
mov rbp, rsp
SHADOW_ARGS_TO_STACK 8
push rsi
push rdi
; end prolog
mov rsi, arg(0) ;coeff_ptr
movq mm0, [rsi]
mov rax, arg(1) ;zbin_ptr
movq mm1, [rax]
movq mm3, mm0
psraw mm0, 15
pxor mm3, mm0
psubw mm3, mm0 ; abs
movq mm2, mm3
pcmpgtw mm1, mm2
pandn mm1, mm2
movq mm3, mm1
mov rdx, arg(6) ;quant_ptr
movq mm1, [rdx]
mov rcx, arg(5) ;round_ptr
movq mm2, [rcx]
paddw mm3, mm2
pmulhuw mm3, mm1
pxor mm3, mm0
psubw mm3, mm0 ;gain the sign back
mov rdi, arg(2) ;qcoeff_ptr
movq mm0, mm3
movq [rdi], mm3
mov rax, arg(3) ;dequant_ptr
movq mm2, [rax]
pmullw mm3, mm2
mov rax, arg(7) ;dqcoeff_ptr
movq [rax], mm3
; next 8
movq mm4, [rsi+8]
mov rax, arg(1) ;zbin_ptr
movq mm5, [rax+8]
movq mm7, mm4
psraw mm4, 15
pxor mm7, mm4
psubw mm7, mm4 ; abs
movq mm6, mm7
pcmpgtw mm5, mm6
pandn mm5, mm6
movq mm7, mm5
movq mm5, [rdx+8]
movq mm6, [rcx+8]
paddw mm7, mm6
pmulhuw mm7, mm5
pxor mm7, mm4
psubw mm7, mm4;gain the sign back
mov rdi, arg(2) ;qcoeff_ptr
movq mm1, mm7
movq [rdi+8], mm7
mov rax, arg(3) ;dequant_ptr
movq mm6, [rax+8]
pmullw mm7, mm6
mov rax, arg(7) ;dqcoeff_ptr
movq [rax+8], mm7
; next 8
movq mm4, [rsi+16]
mov rax, arg(1) ;zbin_ptr
movq mm5, [rax+16]
movq mm7, mm4
psraw mm4, 15
pxor mm7, mm4
psubw mm7, mm4 ; abs
movq mm6, mm7
pcmpgtw mm5, mm6
pandn mm5, mm6
movq mm7, mm5
movq mm5, [rdx+16]
movq mm6, [rcx+16]
paddw mm7, mm6
pmulhuw mm7, mm5
pxor mm7, mm4
psubw mm7, mm4;gain the sign back
mov rdi, arg(2) ;qcoeff_ptr
movq mm1, mm7
movq [rdi+16], mm7
mov rax, arg(3) ;dequant_ptr
movq mm6, [rax+16]
pmullw mm7, mm6
mov rax, arg(7) ;dqcoeff_ptr
movq [rax+16], mm7
; next 8
movq mm4, [rsi+24]
mov rax, arg(1) ;zbin_ptr
movq mm5, [rax+24]
movq mm7, mm4
psraw mm4, 15
pxor mm7, mm4
psubw mm7, mm4 ; abs
movq mm6, mm7
pcmpgtw mm5, mm6
pandn mm5, mm6
movq mm7, mm5
movq mm5, [rdx+24]
movq mm6, [rcx+24]
paddw mm7, mm6
pmulhuw mm7, mm5
pxor mm7, mm4
psubw mm7, mm4;gain the sign back
mov rdi, arg(2) ;qcoeff_ptr
movq mm1, mm7
movq [rdi+24], mm7
mov rax, arg(3) ;dequant_ptr
movq mm6, [rax+24]
pmullw mm7, mm6
mov rax, arg(7) ;dqcoeff_ptr
movq [rax+24], mm7
mov rdi, arg(4) ;scan_mask
mov rsi, arg(2) ;qcoeff_ptr
pxor mm5, mm5
pxor mm7, mm7
movq mm0, [rsi]
movq mm1, [rsi+8]
movq mm2, [rdi]
movq mm3, [rdi+8];
pcmpeqw mm0, mm7
pcmpeqw mm1, mm7
pcmpeqw mm6, mm6
pxor mm0, mm6
pxor mm1, mm6
psrlw mm0, 15
psrlw mm1, 15
pmaddwd mm0, mm2
pmaddwd mm1, mm3
movq mm5, mm0
paddd mm5, mm1
movq mm0, [rsi+16]
movq mm1, [rsi+24]
movq mm2, [rdi+16]
movq mm3, [rdi+24];
pcmpeqw mm0, mm7
pcmpeqw mm1, mm7
pcmpeqw mm6, mm6
pxor mm0, mm6
pxor mm1, mm6
psrlw mm0, 15
psrlw mm1, 15
pmaddwd mm0, mm2
pmaddwd mm1, mm3
paddd mm5, mm0
paddd mm5, mm1
movq mm0, mm5
psrlq mm5, 32
paddd mm0, mm5
; eob adjustment begins here
nasm: match instruction length (movd/movq) to parameters nasm requires the instruction length (movd/movq) to match to its parameters. I find it more clear to really use 64bit instructions when we use 64bit registers in the assembly. Provide nasm compatibility. No binary change by this patch with yasm on {x86_64,i686}-fedora13-linux-gnu. Few longer opcodes with nasm on {x86_64,i686}-fedora13-linux-gnu have been checked as safe. Change-Id: Id9b1a5cdfb1bc05697e523c317a296df43d42a91
2010-10-04 23:19:33 +02:00
movq rcx, mm0
Initial WebM release
2010-05-18 17:58:33 +02:00
and rcx, 0xffff
xor rdx, rdx
sub rdx, rcx ; rdx=-rcx
bsr rax, rcx
inc rax
sar rdx, 31
and rax, rdx
; Substitute the sse assembly for the old mmx mixed assembly/C. The
; following is kept as reference
nasm: match instruction length (movd/movq) to parameters nasm requires the instruction length (movd/movq) to match to its parameters. I find it more clear to really use 64bit instructions when we use 64bit registers in the assembly. Provide nasm compatibility. No binary change by this patch with yasm on {x86_64,i686}-fedora13-linux-gnu. Few longer opcodes with nasm on {x86_64,i686}-fedora13-linux-gnu have been checked as safe. Change-Id: Id9b1a5cdfb1bc05697e523c317a296df43d42a91
2010-10-04 23:19:33 +02:00
; movq rcx, mm0
Initial WebM release
2010-05-18 17:58:33 +02:00
; bsr rax, rcx
;
; mov eob, rax
; mov eee, rcx
;
;if(eee==0)
;{
; eob=-1;
;}
;else if(eee<0)
;{
; eob=15;
;}
;d->eob = eob+1;
; begin epilog
pop rdi
pop rsi
UNSHADOW_ARGS
pop rbp
ret
Reference in New Issue Copy Permalink