lavc/flacenc: add sse4 version of the 16-bit lpc encoder

From 1.8 to 2.4 times faster.  Runtime is reduced by 2 to 39%.  The
speed-up generally increases with compression_level.

This lpc encoder is not used with levels < 3 so it provides no speed-up
in these cases.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

LICENSE.md

@@ -15,6 +15,7 @@ Specifically, the GPL parts of FFmpeg are:
 - libpostproc
 - libmpcodecs
 - optional x86 optimizations in the files
+  libavcodec/x86/flac_dsp_gpl.asm
   libavcodec/x86/idct_mmx.c
 - libutvideo encoding/decoding wrappers in
   libavcodec/libutvideo*.cpp

libavcodec/flacenc.c

@@ -80,7 +80,7 @@ typedef struct FlacSubframe {
     int shift;
     RiceContext rc;
     int32_t samples[FLAC_MAX_BLOCKSIZE];
-    int32_t residual[FLAC_MAX_BLOCKSIZE+1];
+    int32_t residual[FLAC_MAX_BLOCKSIZE+3];
 } FlacSubframe;
 
 typedef struct FlacFrame {

libavcodec/x86/Makefile

@@ -83,6 +83,9 @@ YASM-OBJS-$(CONFIG_DIRAC_DECODER)      += x86/diracdsp_mmx.o x86/diracdsp_yasm.o
 YASM-OBJS-$(CONFIG_DNXHD_ENCODER)      += x86/dnxhdenc.o
 YASM-OBJS-$(CONFIG_FFT)                += x86/fft.o
 YASM-OBJS-$(CONFIG_FLAC_DECODER)       += x86/flacdsp.o
+ifdef CONFIG_GPL
+YASM-OBJS-$(CONFIG_FLAC_ENCODER)       += x86/flac_dsp_gpl.o
+endif
 YASM-OBJS-$(CONFIG_H263DSP)            += x86/h263_loopfilter.o
 YASM-OBJS-$(CONFIG_H264CHROMA)         += x86/h264_chromamc.o           \
                                           x86/h264_chromamc_10bit.o

libavcodec/x86/flac_dsp_gpl.asm

+;******************************************************************************
+;* FLAC DSP functions
+;*
+;* Copyright (c) 2014 James Darnley <james.darnley@gmail.com>
+;*
+;* This file is part of FFmpeg.
+;*
+;* FFmpeg is free software; you can redistribute it and/or modify
+;* it under the terms of the GNU General Public License as published by
+;* the Free Software Foundation; either version 2 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 General Public License for more details.
+;*
+;* You should have received a copy of the GNU 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_TEXT
+
+INIT_XMM sse4
+%if ARCH_X86_64
+    cglobal flac_enc_lpc_16, 5, 7, 4, 0, res, smp, len, order, coefs
+    DECLARE_REG_TMP 5, 6
+    %define length r2d
+
+    movsxd orderq, orderd
+%else
+    cglobal flac_enc_lpc_16, 5, 6, 4, 0, res, smp, len, order, coefs
+    DECLARE_REG_TMP 2, 5
+    %define length r2mp
+%endif
+
+; Here we assume that the maximum order value is 32.  This means that we only
+; need to copy a maximum of 32 samples.  Therefore we let the preprocessor
+; unroll this loop and copy all 32.
+%assign iter 0
+%rep 32/(mmsize/4)
+    movu  m0,         [smpq+iter]
+    movu [resq+iter],  m0
+    %assign iter iter+mmsize
+%endrep
+
+lea  resq,   [resq+orderq*4]
+lea  smpq,   [smpq+orderq*4]
+lea  coefsq, [coefsq+orderq*4]
+sub  length,  orderd
+movd m3,      r5m
+neg  orderq
+
+%define posj t0q
+%define negj t1q
+
+.looplen:
+    pxor m0,   m0
+    mov  posj, orderq
+    xor  negj, negj
+
+    .looporder:
+        movd   m2, [coefsq+posj*4] ; c = coefs[j]
+        SPLATD m2
+        movu   m1, [smpq+negj*4-4] ; s = smp[i-j-1]
+        pmulld m1,  m2
+        paddd  m0,  m1             ; p += c * s
+
+        dec    negj
+        inc    posj
+    jnz .looporder
+
+    psrad  m0,     m3              ; p >>= shift
+    movu   m1,    [smpq]
+    psubd  m1,     m0              ; smp[i] - p
+    movu  [resq],  m1              ; res[i] = smp[i] - (p >> shift)
+
+    add resq,   mmsize
+    add smpq,   mmsize
+    sub length, mmsize/4
+jg .looplen
+RET

libavcodec/x86/flacdsp_init.c

@@ -27,6 +27,8 @@ void ff_flac_lpc_32_sse4(int32_t *samples, const int coeffs[32], int order,
 void ff_flac_lpc_32_xop(int32_t *samples, const int coeffs[32], int order,
                         int qlevel, int len);
 
+void ff_flac_enc_lpc_16_sse4(int32_t *, const int32_t *, int, int, const int32_t *,int);
+
 av_cold void ff_flacdsp_init_x86(FLACDSPContext *c, enum AVSampleFormat fmt,
                                  int bps)
 {
@@ -36,6 +38,8 @@ av_cold void ff_flacdsp_init_x86(FLACDSPContext *c, enum AVSampleFormat fmt,
     if (EXTERNAL_SSE4(cpu_flags)) {
         if (bps > 16 && CONFIG_FLAC_DECODER)
             c->lpc = ff_flac_lpc_32_sse4;
+        if (bps == 16 && CONFIG_FLAC_ENCODER && CONFIG_GPL)
+            c->lpc_encode = ff_flac_enc_lpc_16_sse4;
     }
     if (EXTERNAL_XOP(cpu_flags)) {
         if (bps > 16 && CONFIG_FLAC_DECODER)