ffmpeg/libavcodec/x86/jpeg2000dsp.asm
Lynne bbe95f7353
x86: replace explicit REP_RETs with RETs
From x86inc:
> On AMD cpus <=K10, an ordinary ret is slow if it immediately follows either
> a branch or a branch target. So switch to a 2-byte form of ret in that case.
> We can automatically detect "follows a branch", but not a branch target.
> (SSSE3 is a sufficient condition to know that your cpu doesn't have this problem.)

x86inc can automatically determine whether to use REP_RET rather than
REP in most of these cases, so impact is minimal. Additionally, a few
REP_RETs were used unnecessary, despite the return being nowhere near a
branch.

The only CPUs affected were AMD K10s, made between 2007 and 2011, 16
years ago and 12 years ago, respectively.

In the future, everyone involved with x86inc should consider dropping
REP_RETs altogether.
2023-02-01 04:23:55 +01:00

165 lines
3.9 KiB
NASM

;******************************************************************************
;* SIMD-optimized JPEG2000 DSP functions
;* Copyright (c) 2014 Nicolas Bertrand
;* Copyright (c) 2015 James Almer
;*
;* This file is part of FFmpeg.
;*
;* FFmpeg is free software; you can redistribute it and/or
;* modify it under the terms of the GNU Lesser General Public
;* License as published by the Free Software Foundation; either
;* version 2.1 of the License, or (at your option) any later version.
;*
;* FFmpeg is distributed in the hope that it will be useful,
;* but WITHOUT ANY WARRANTY; without even the implied warranty of
;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
;* Lesser General Public License for more details.
;*
;* You should have received a copy of the GNU Lesser General Public
;* License along with FFmpeg; if not, write to the Free Software
;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
;******************************************************************************
%include "libavutil/x86/x86util.asm"
SECTION_RODATA 32
pf_ict0: times 8 dd 1.402
pf_ict1: times 8 dd 0.34413
pf_ict2: times 8 dd 0.71414
pf_ict3: times 8 dd 1.772
SECTION .text
;***********************************************************************
; ff_ict_float_<opt>(float *src0, float *src1, float *src2, int csize)
;***********************************************************************
%macro ICT_FLOAT 1
cglobal ict_float, 4, 4, %1, src0, src1, src2, csize
shl csized, 2
add src0q, csizeq
add src1q, csizeq
add src2q, csizeq
neg csizeq
movaps m6, [pf_ict0]
movaps m7, [pf_ict1]
%define ICT0 m6
%define ICT1 m7
%if ARCH_X86_64
movaps m8, [pf_ict2]
%define ICT2 m8
%if cpuflag(avx)
movaps m3, [pf_ict3]
%define ICT3 m3
%else
movaps m9, [pf_ict3]
%define ICT3 m9
%endif
%else ; ARCH_X86_32
%define ICT2 [pf_ict2]
%if cpuflag(avx)
movaps m3, [pf_ict3]
%define ICT3 m3
%else
%define ICT3 [pf_ict3]
%endif
%endif ; ARCH
align 16
.loop:
movaps m0, [src0q+csizeq]
movaps m1, [src1q+csizeq]
movaps m2, [src2q+csizeq]
%if cpuflag(fma4) || cpuflag(fma3)
%if cpuflag(fma4)
fnmaddps m5, m1, ICT1, m0
fmaddps m4, m2, ICT0, m0
%else ; fma3
movaps m5, m1
movaps m4, m2
fnmaddps m5, m5, ICT1, m0
fmaddps m4, m4, ICT0, m0
%endif
fmaddps m0, m1, ICT3, m0
fnmaddps m5, m2, ICT2, m5
%else ; non FMA
%if cpuflag(avx)
mulps m5, m1, ICT1
mulps m4, m2, ICT0
mulps m1, m1, ICT3
mulps m2, m2, ICT2
subps m5, m0, m5
%else ; sse
movaps m3, m1
movaps m4, m2
movaps m5, m0
mulps m3, ICT1
mulps m4, ICT0
mulps m1, ICT3
mulps m2, ICT2
subps m5, m3
%endif
addps m4, m4, m0
addps m0, m0, m1
subps m5, m5, m2
%endif
movaps [src0q+csizeq], m4
movaps [src2q+csizeq], m0
movaps [src1q+csizeq], m5
add csizeq, mmsize
jl .loop
RET
%endmacro
INIT_XMM sse
ICT_FLOAT 10
INIT_YMM avx
ICT_FLOAT 9
%if HAVE_FMA4_EXTERNAL
INIT_XMM fma4
ICT_FLOAT 9
%endif
INIT_YMM fma3
ICT_FLOAT 9
;***************************************************************************
; ff_rct_int_<opt>(int32_t *src0, int32_t *src1, int32_t *src2, int csize)
;***************************************************************************
%macro RCT_INT 0
cglobal rct_int, 4, 4, 4, src0, src1, src2, csize
shl csized, 2
add src0q, csizeq
add src1q, csizeq
add src2q, csizeq
neg csizeq
align 16
.loop:
mova m1, [src1q+csizeq]
mova m2, [src2q+csizeq]
mova m0, [src0q+csizeq]
paddd m3, m1, m2
psrad m3, 2
psubd m0, m3
paddd m1, m0
paddd m2, m0
mova [src1q+csizeq], m0
mova [src2q+csizeq], m1
mova [src0q+csizeq], m2
add csizeq, mmsize
jl .loop
RET
%endmacro
INIT_XMM sse2
RCT_INT
%if HAVE_AVX2_EXTERNAL
INIT_YMM avx2
RCT_INT
%endif