[FFmpeg-devel] [PATCH] x86/flacdsp: add SSE2 and AVX decorrelate functions
Two to four times faster depending on instruction set, block size and channel
count.
Signed-off-by: James Almer jamr...@gmail.com
---
Now also with 16 bits indep4 and indep6.
libavcodec/arm/flacdsp_init_arm.c | 2 +-
libavcodec/flacdec.c | 6 +-
libavcodec/flacdsp.c | 6 +-
libavcodec/flacdsp.h | 6 +-
libavcodec/flacenc.c | 2 +-
libavcodec/x86/flacdsp.asm| 261 ++
libavcodec/x86/flacdsp_init.c | 52 +++-
7 files changed, 323 insertions(+), 12 deletions(-)
diff --git a/libavcodec/arm/flacdsp_init_arm.c
b/libavcodec/arm/flacdsp_init_arm.c
index 9ddb268..df1b19c 100644
--- a/libavcodec/arm/flacdsp_init_arm.c
+++ b/libavcodec/arm/flacdsp_init_arm.c
@@ -24,7 +24,7 @@
void ff_flac_lpc_16_arm(int32_t *samples, const int coeffs[32], int order,
int qlevel, int len);
-av_cold void ff_flacdsp_init_arm(FLACDSPContext *c, enum AVSampleFormat fmt,
+av_cold void ff_flacdsp_init_arm(FLACDSPContext *c, enum AVSampleFormat fmt,
int channels,
int bps)
{
if (bps = 16 CONFIG_FLAC_DECODER)
diff --git a/libavcodec/flacdec.c b/libavcodec/flacdec.c
index 1333972..01ae717 100644
--- a/libavcodec/flacdec.c
+++ b/libavcodec/flacdec.c
@@ -111,7 +111,7 @@ static av_cold int flac_decode_init(AVCodecContext *avctx)
if (ret 0)
return ret;
flac_set_bps(s);
-ff_flacdsp_init(s-dsp, avctx-sample_fmt, s-bps);
+ff_flacdsp_init(s-dsp, avctx-sample_fmt, s-channels, s-bps);
s-got_streaminfo = 1;
return 0;
@@ -173,7 +173,7 @@ static int parse_streaminfo(FLACContext *s, const uint8_t
*buf, int buf_size)
if (ret 0)
return ret;
flac_set_bps(s);
-ff_flacdsp_init(s-dsp, s-avctx-sample_fmt, s-bps);
+ff_flacdsp_init(s-dsp, s-avctx-sample_fmt, s-channels, s-bps);
s-got_streaminfo = 1;
return 0;
@@ -472,7 +472,7 @@ static int decode_frame(FLACContext *s)
ret = allocate_buffers(s);
if (ret 0)
return ret;
-ff_flacdsp_init(s-dsp, s-avctx-sample_fmt, s-bps);
+ff_flacdsp_init(s-dsp, s-avctx-sample_fmt, s-channels, s-bps);
s-got_streaminfo = 1;
dump_headers(s-avctx, (FLACStreaminfo *)s);
}
diff --git a/libavcodec/flacdsp.c b/libavcodec/flacdsp.c
index b15bc74..a83eb83 100644
--- a/libavcodec/flacdsp.c
+++ b/libavcodec/flacdsp.c
@@ -85,7 +85,7 @@ static void flac_lpc_32_c(int32_t *decoded, const int
coeffs[32],
}
-av_cold void ff_flacdsp_init(FLACDSPContext *c, enum AVSampleFormat fmt,
+av_cold void ff_flacdsp_init(FLACDSPContext *c, enum AVSampleFormat fmt, int
channels,
int bps)
{
if (bps 16) {
@@ -127,7 +127,7 @@ av_cold void ff_flacdsp_init(FLACDSPContext *c, enum
AVSampleFormat fmt,
}
if (ARCH_ARM)
-ff_flacdsp_init_arm(c, fmt, bps);
+ff_flacdsp_init_arm(c, fmt, channels, bps);
if (ARCH_X86)
-ff_flacdsp_init_x86(c, fmt, bps);
+ff_flacdsp_init_x86(c, fmt, channels, bps);
}
diff --git a/libavcodec/flacdsp.h b/libavcodec/flacdsp.h
index 14f3466..417381c 100644
--- a/libavcodec/flacdsp.h
+++ b/libavcodec/flacdsp.h
@@ -31,8 +31,8 @@ typedef struct FLACDSPContext {
const int32_t coefs[32], int shift);
} FLACDSPContext;
-void ff_flacdsp_init(FLACDSPContext *c, enum AVSampleFormat fmt, int bps);
-void ff_flacdsp_init_arm(FLACDSPContext *c, enum AVSampleFormat fmt, int bps);
-void ff_flacdsp_init_x86(FLACDSPContext *c, enum AVSampleFormat fmt, int bps);
+void ff_flacdsp_init(FLACDSPContext *c, enum AVSampleFormat fmt, int channels,
int bps);
+void ff_flacdsp_init_arm(FLACDSPContext *c, enum AVSampleFormat fmt, int
channels, int bps);
+void ff_flacdsp_init_x86(FLACDSPContext *c, enum AVSampleFormat fmt, int
channels, int bps);
#endif /* AVCODEC_FLACDSP_H */
diff --git a/libavcodec/flacenc.c b/libavcodec/flacenc.c
index 3b72888..e66ef3d 100644
--- a/libavcodec/flacenc.c
+++ b/libavcodec/flacenc.c
@@ -428,7 +428,7 @@ static av_cold int flac_encode_init(AVCodecContext *avctx)
s-options.max_prediction_order, FF_LPC_TYPE_LEVINSON);
ff_bswapdsp_init(s-bdsp);
-ff_flacdsp_init(s-flac_dsp, avctx-sample_fmt,
+ff_flacdsp_init(s-flac_dsp, avctx-sample_fmt, channels,
avctx-bits_per_raw_sample);
dprint_compression_options(s);
diff --git a/libavcodec/x86/flacdsp.asm b/libavcodec/x86/flacdsp.asm
index 37ee87b..afbe199 100644
--- a/libavcodec/x86/flacdsp.asm
+++ b/libavcodec/x86/flacdsp.asm
@@ -72,3 +72,264 @@ ALIGN 16
LPC_32 xop
%endif
LPC_32 sse4
+
+;-
+;void ff_flac_decorrelate_{ls,rs,ms,indep2}_16_sse2(uint8_t **out, int32_t
**in, int channels,
+; int len, int shift);
[FFmpeg-devel] [PATCH] x86/flacdsp: add SSE2 and AVX decorrelate functions
Two to four times faster depending on instruction set, block size and channel
count.
Signed-off-by: James Almer jamr...@gmail.com
---
TODO: 16 bits indep for 4, 6 and 8 channels. 24/32 bits indep for 8 channels.
AVX2 and maybe MMX versions.
Planar?
libavcodec/arm/flacdsp_init_arm.c | 2 +-
libavcodec/flacdec.c | 6 +-
libavcodec/flacdsp.c | 6 +-
libavcodec/flacdsp.h | 6 +-
libavcodec/flacenc.c | 2 +-
libavcodec/x86/flacdsp.asm| 206 ++
libavcodec/x86/flacdsp_init.c | 48 -
7 files changed, 264 insertions(+), 12 deletions(-)
diff --git a/libavcodec/arm/flacdsp_init_arm.c
b/libavcodec/arm/flacdsp_init_arm.c
index 9ddb268..df1b19c 100644
--- a/libavcodec/arm/flacdsp_init_arm.c
+++ b/libavcodec/arm/flacdsp_init_arm.c
@@ -24,7 +24,7 @@
void ff_flac_lpc_16_arm(int32_t *samples, const int coeffs[32], int order,
int qlevel, int len);
-av_cold void ff_flacdsp_init_arm(FLACDSPContext *c, enum AVSampleFormat fmt,
+av_cold void ff_flacdsp_init_arm(FLACDSPContext *c, enum AVSampleFormat fmt,
int channels,
int bps)
{
if (bps = 16 CONFIG_FLAC_DECODER)
diff --git a/libavcodec/flacdec.c b/libavcodec/flacdec.c
index 1333972..01ae717 100644
--- a/libavcodec/flacdec.c
+++ b/libavcodec/flacdec.c
@@ -111,7 +111,7 @@ static av_cold int flac_decode_init(AVCodecContext *avctx)
if (ret 0)
return ret;
flac_set_bps(s);
-ff_flacdsp_init(s-dsp, avctx-sample_fmt, s-bps);
+ff_flacdsp_init(s-dsp, avctx-sample_fmt, s-channels, s-bps);
s-got_streaminfo = 1;
return 0;
@@ -173,7 +173,7 @@ static int parse_streaminfo(FLACContext *s, const uint8_t
*buf, int buf_size)
if (ret 0)
return ret;
flac_set_bps(s);
-ff_flacdsp_init(s-dsp, s-avctx-sample_fmt, s-bps);
+ff_flacdsp_init(s-dsp, s-avctx-sample_fmt, s-channels, s-bps);
s-got_streaminfo = 1;
return 0;
@@ -472,7 +472,7 @@ static int decode_frame(FLACContext *s)
ret = allocate_buffers(s);
if (ret 0)
return ret;
-ff_flacdsp_init(s-dsp, s-avctx-sample_fmt, s-bps);
+ff_flacdsp_init(s-dsp, s-avctx-sample_fmt, s-channels, s-bps);
s-got_streaminfo = 1;
dump_headers(s-avctx, (FLACStreaminfo *)s);
}
diff --git a/libavcodec/flacdsp.c b/libavcodec/flacdsp.c
index b15bc74..a83eb83 100644
--- a/libavcodec/flacdsp.c
+++ b/libavcodec/flacdsp.c
@@ -85,7 +85,7 @@ static void flac_lpc_32_c(int32_t *decoded, const int
coeffs[32],
}
-av_cold void ff_flacdsp_init(FLACDSPContext *c, enum AVSampleFormat fmt,
+av_cold void ff_flacdsp_init(FLACDSPContext *c, enum AVSampleFormat fmt, int
channels,
int bps)
{
if (bps 16) {
@@ -127,7 +127,7 @@ av_cold void ff_flacdsp_init(FLACDSPContext *c, enum
AVSampleFormat fmt,
}
if (ARCH_ARM)
-ff_flacdsp_init_arm(c, fmt, bps);
+ff_flacdsp_init_arm(c, fmt, channels, bps);
if (ARCH_X86)
-ff_flacdsp_init_x86(c, fmt, bps);
+ff_flacdsp_init_x86(c, fmt, channels, bps);
}
diff --git a/libavcodec/flacdsp.h b/libavcodec/flacdsp.h
index 14f3466..417381c 100644
--- a/libavcodec/flacdsp.h
+++ b/libavcodec/flacdsp.h
@@ -31,8 +31,8 @@ typedef struct FLACDSPContext {
const int32_t coefs[32], int shift);
} FLACDSPContext;
-void ff_flacdsp_init(FLACDSPContext *c, enum AVSampleFormat fmt, int bps);
-void ff_flacdsp_init_arm(FLACDSPContext *c, enum AVSampleFormat fmt, int bps);
-void ff_flacdsp_init_x86(FLACDSPContext *c, enum AVSampleFormat fmt, int bps);
+void ff_flacdsp_init(FLACDSPContext *c, enum AVSampleFormat fmt, int channels,
int bps);
+void ff_flacdsp_init_arm(FLACDSPContext *c, enum AVSampleFormat fmt, int
channels, int bps);
+void ff_flacdsp_init_x86(FLACDSPContext *c, enum AVSampleFormat fmt, int
channels, int bps);
#endif /* AVCODEC_FLACDSP_H */
diff --git a/libavcodec/flacenc.c b/libavcodec/flacenc.c
index 3b72888..e66ef3d 100644
--- a/libavcodec/flacenc.c
+++ b/libavcodec/flacenc.c
@@ -428,7 +428,7 @@ static av_cold int flac_encode_init(AVCodecContext *avctx)
s-options.max_prediction_order, FF_LPC_TYPE_LEVINSON);
ff_bswapdsp_init(s-bdsp);
-ff_flacdsp_init(s-flac_dsp, avctx-sample_fmt,
+ff_flacdsp_init(s-flac_dsp, avctx-sample_fmt, channels,
avctx-bits_per_raw_sample);
dprint_compression_options(s);
diff --git a/libavcodec/x86/flacdsp.asm b/libavcodec/x86/flacdsp.asm
index 37ee87b..35f9b81 100644
--- a/libavcodec/x86/flacdsp.asm
+++ b/libavcodec/x86/flacdsp.asm
@@ -72,3 +72,209 @@ ALIGN 16
LPC_32 xop
%endif
LPC_32 sse4
+
+;-
+;void ff_flac_decorrelate_{ls,rs,ms,indep2}_16_sse2(uint8_t **out, int32_t
**in, int channels,
+;
Re: [FFmpeg-devel] [PATCH] x86/flacdsp: add SSE2 and AVX decorrelate functions
On Sun, Nov 02, 2014 at 07:31:48PM -0300, James Almer wrote: Two to four times faster depending on instruction set, block size and channel count. Signed-off-by: James Almer jamr...@gmail.com --- TODO: 16 bits indep for 4, 6 and 8 channels. 24/32 bits indep for 8 channels. AVX2 and maybe MMX versions. Planar? libavcodec/arm/flacdsp_init_arm.c | 2 +- libavcodec/flacdec.c | 6 +- libavcodec/flacdsp.c | 6 +- libavcodec/flacdsp.h | 6 +- libavcodec/flacenc.c | 2 +- libavcodec/x86/flacdsp.asm| 206 ++ libavcodec/x86/flacdsp_init.c | 48 - 7 files changed, 264 insertions(+), 12 deletions(-) [...] +mova m0, [in0q] +mova m1, [in0q+in1q] +%if %1 2 +mova m2, [in0q+in2q] +mova m3, [in0q+in3q] +%if %1 4 +mova m4, [in0q+in4q] +mova m5, [in0q+in5q] +%endif +%endif +pslld m0, m%2 +pslld m1, m%2 +%if %1 2 +pslld m2, m%2 +pslld m3, m%2 +%if %1 4 +pslld m4, m%2 +pslld m5, m%2 +%endif +%endif Can't you do something like this? (untested) pslld m0, [in0q], m%2 %assign i 0 %rep %1 pslld m%i, [in0q+in%iq], m%2 %assigni i+1 %endrep [...] -- Clément B. pgphQeDWNz8un.pgp Description: PGP signature ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
Re: [FFmpeg-devel] [PATCH] x86/flacdsp: add SSE2 and AVX decorrelate functions
On 02/11/14 7:43 PM, Clément Bœsch wrote: On Sun, Nov 02, 2014 at 07:31:48PM -0300, James Almer wrote: Two to four times faster depending on instruction set, block size and channel count. Signed-off-by: James Almer jamr...@gmail.com --- TODO: 16 bits indep for 4, 6 and 8 channels. 24/32 bits indep for 8 channels. AVX2 and maybe MMX versions. Planar? libavcodec/arm/flacdsp_init_arm.c | 2 +- libavcodec/flacdec.c | 6 +- libavcodec/flacdsp.c | 6 +- libavcodec/flacdsp.h | 6 +- libavcodec/flacenc.c | 2 +- libavcodec/x86/flacdsp.asm| 206 ++ libavcodec/x86/flacdsp_init.c | 48 - 7 files changed, 264 insertions(+), 12 deletions(-) [...] +mova m0, [in0q] +mova m1, [in0q+in1q] +%if %1 2 +mova m2, [in0q+in2q] +mova m3, [in0q+in3q] +%if %1 4 +mova m4, [in0q+in4q] +mova m5, [in0q+in5q] +%endif +%endif +pslld m0, m%2 +pslld m1, m%2 +%if %1 2 +pslld m2, m%2 +pslld m3, m%2 +%if %1 4 +pslld m4, m%2 +pslld m5, m%2 +%endif +%endif Can't you do something like this? (untested) pslld m0, [in0q], m%2 %assign i 0 %rep %1 pslld m%i, [in0q+in%iq], m%2 %assigni i+1 %endrep YASMlibavcodec/x86/flacdsp.o D:/MinGW/msys/1.0/ffmpeg/src/libavcodec/x86/flacdsp.asm:271: error: undefined symbol `m' (first use) D:/MinGW/msys/1.0/ffmpeg/src/libavcodec/x86/flacdsp.asm:271: error: undefined symbol `i' (first use) D:/MinGW/msys/1.0/ffmpeg/src/libavcodec/x86/flacdsp.asm:271: error: undefined symbol `in' (first use) D:/MinGW/msys/1.0/ffmpeg/src/libavcodec/x86/flacdsp.asm:271: error: undefined symbol `iq' (first use) D:/MinGW/msys/1.0/ffmpeg/src/libavcodec/x86/flacdsp.asm:271: error: (Each undefined symbol is reported only once.) make: *** [libavcodec/x86/flacdsp.o] Error 1 A %rep like that is only four lines shorter. Do you consider it more readable than the alternative to justify trying to get it working? ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
Re: [FFmpeg-devel] [PATCH] x86/flacdsp: add SSE2 and AVX decorrelate functions
On Sun, Nov 02, 2014 at 07:55:35PM -0300, James Almer wrote: On 02/11/14 7:43 PM, Clément Bœsch wrote: On Sun, Nov 02, 2014 at 07:31:48PM -0300, James Almer wrote: Two to four times faster depending on instruction set, block size and channel count. Signed-off-by: James Almer jamr...@gmail.com --- TODO: 16 bits indep for 4, 6 and 8 channels. 24/32 bits indep for 8 channels. AVX2 and maybe MMX versions. Planar? libavcodec/arm/flacdsp_init_arm.c | 2 +- libavcodec/flacdec.c | 6 +- libavcodec/flacdsp.c | 6 +- libavcodec/flacdsp.h | 6 +- libavcodec/flacenc.c | 2 +- libavcodec/x86/flacdsp.asm| 206 ++ libavcodec/x86/flacdsp_init.c | 48 - 7 files changed, 264 insertions(+), 12 deletions(-) [...] +mova m0, [in0q] +mova m1, [in0q+in1q] +%if %1 2 +mova m2, [in0q+in2q] +mova m3, [in0q+in3q] +%if %1 4 +mova m4, [in0q+in4q] +mova m5, [in0q+in5q] +%endif +%endif +pslld m0, m%2 +pslld m1, m%2 +%if %1 2 +pslld m2, m%2 +pslld m3, m%2 +%if %1 4 +pslld m4, m%2 +pslld m5, m%2 +%endif +%endif Can't you do something like this? (untested) pslld m0, [in0q], m%2 %assign i 0 %rep %1 pslld m%i, [in0q+in%iq], m%2 %assigni i+1 %endrep YASMlibavcodec/x86/flacdsp.o D:/MinGW/msys/1.0/ffmpeg/src/libavcodec/x86/flacdsp.asm:271: error: undefined symbol `m' (first use) D:/MinGW/msys/1.0/ffmpeg/src/libavcodec/x86/flacdsp.asm:271: error: undefined symbol `i' (first use) D:/MinGW/msys/1.0/ffmpeg/src/libavcodec/x86/flacdsp.asm:271: error: undefined symbol `in' (first use) D:/MinGW/msys/1.0/ffmpeg/src/libavcodec/x86/flacdsp.asm:271: error: undefined symbol `iq' (first use) D:/MinGW/msys/1.0/ffmpeg/src/libavcodec/x86/flacdsp.asm:271: error: (Each undefined symbol is reported only once.) make: *** [libavcodec/x86/flacdsp.o] Error 1 A %rep like that is only four lines shorter. Do you consider it more readable than the alternative to justify trying to get it working? Totally up to you, it looked easier to maintain and obvious than several nested ifdefery. -- Clément B. pgpzLgMCq7wMk.pgp Description: PGP signature ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
[FFmpeg-devel] [PATCH] x86/flacdsp: add SSE2 and AVX decorrelate functions
Two to four times faster depending on instruction set, block size and channel
count.
Signed-off-by: James Almer jamr...@gmail.com
---
libavcodec/arm/flacdsp_init_arm.c | 2 +-
libavcodec/flacdec.c | 6 +-
libavcodec/flacdsp.c | 6 +-
libavcodec/flacdsp.h | 6 +-
libavcodec/flacenc.c | 2 +-
libavcodec/x86/flacdsp.asm| 193 ++
libavcodec/x86/flacdsp_init.c | 48 +-
7 files changed, 251 insertions(+), 12 deletions(-)
diff --git a/libavcodec/arm/flacdsp_init_arm.c
b/libavcodec/arm/flacdsp_init_arm.c
index 9ddb268..df1b19c 100644
--- a/libavcodec/arm/flacdsp_init_arm.c
+++ b/libavcodec/arm/flacdsp_init_arm.c
@@ -24,7 +24,7 @@
void ff_flac_lpc_16_arm(int32_t *samples, const int coeffs[32], int order,
int qlevel, int len);
-av_cold void ff_flacdsp_init_arm(FLACDSPContext *c, enum AVSampleFormat fmt,
+av_cold void ff_flacdsp_init_arm(FLACDSPContext *c, enum AVSampleFormat fmt,
int channels,
int bps)
{
if (bps = 16 CONFIG_FLAC_DECODER)
diff --git a/libavcodec/flacdec.c b/libavcodec/flacdec.c
index 1333972..01ae717 100644
--- a/libavcodec/flacdec.c
+++ b/libavcodec/flacdec.c
@@ -111,7 +111,7 @@ static av_cold int flac_decode_init(AVCodecContext *avctx)
if (ret 0)
return ret;
flac_set_bps(s);
-ff_flacdsp_init(s-dsp, avctx-sample_fmt, s-bps);
+ff_flacdsp_init(s-dsp, avctx-sample_fmt, s-channels, s-bps);
s-got_streaminfo = 1;
return 0;
@@ -173,7 +173,7 @@ static int parse_streaminfo(FLACContext *s, const uint8_t
*buf, int buf_size)
if (ret 0)
return ret;
flac_set_bps(s);
-ff_flacdsp_init(s-dsp, s-avctx-sample_fmt, s-bps);
+ff_flacdsp_init(s-dsp, s-avctx-sample_fmt, s-channels, s-bps);
s-got_streaminfo = 1;
return 0;
@@ -472,7 +472,7 @@ static int decode_frame(FLACContext *s)
ret = allocate_buffers(s);
if (ret 0)
return ret;
-ff_flacdsp_init(s-dsp, s-avctx-sample_fmt, s-bps);
+ff_flacdsp_init(s-dsp, s-avctx-sample_fmt, s-channels, s-bps);
s-got_streaminfo = 1;
dump_headers(s-avctx, (FLACStreaminfo *)s);
}
diff --git a/libavcodec/flacdsp.c b/libavcodec/flacdsp.c
index b15bc74..a83eb83 100644
--- a/libavcodec/flacdsp.c
+++ b/libavcodec/flacdsp.c
@@ -85,7 +85,7 @@ static void flac_lpc_32_c(int32_t *decoded, const int
coeffs[32],
}
-av_cold void ff_flacdsp_init(FLACDSPContext *c, enum AVSampleFormat fmt,
+av_cold void ff_flacdsp_init(FLACDSPContext *c, enum AVSampleFormat fmt, int
channels,
int bps)
{
if (bps 16) {
@@ -127,7 +127,7 @@ av_cold void ff_flacdsp_init(FLACDSPContext *c, enum
AVSampleFormat fmt,
}
if (ARCH_ARM)
-ff_flacdsp_init_arm(c, fmt, bps);
+ff_flacdsp_init_arm(c, fmt, channels, bps);
if (ARCH_X86)
-ff_flacdsp_init_x86(c, fmt, bps);
+ff_flacdsp_init_x86(c, fmt, channels, bps);
}
diff --git a/libavcodec/flacdsp.h b/libavcodec/flacdsp.h
index 14f3466..417381c 100644
--- a/libavcodec/flacdsp.h
+++ b/libavcodec/flacdsp.h
@@ -31,8 +31,8 @@ typedef struct FLACDSPContext {
const int32_t coefs[32], int shift);
} FLACDSPContext;
-void ff_flacdsp_init(FLACDSPContext *c, enum AVSampleFormat fmt, int bps);
-void ff_flacdsp_init_arm(FLACDSPContext *c, enum AVSampleFormat fmt, int bps);
-void ff_flacdsp_init_x86(FLACDSPContext *c, enum AVSampleFormat fmt, int bps);
+void ff_flacdsp_init(FLACDSPContext *c, enum AVSampleFormat fmt, int channels,
int bps);
+void ff_flacdsp_init_arm(FLACDSPContext *c, enum AVSampleFormat fmt, int
channels, int bps);
+void ff_flacdsp_init_x86(FLACDSPContext *c, enum AVSampleFormat fmt, int
channels, int bps);
#endif /* AVCODEC_FLACDSP_H */
diff --git a/libavcodec/flacenc.c b/libavcodec/flacenc.c
index 3b72888..e66ef3d 100644
--- a/libavcodec/flacenc.c
+++ b/libavcodec/flacenc.c
@@ -428,7 +428,7 @@ static av_cold int flac_encode_init(AVCodecContext *avctx)
s-options.max_prediction_order, FF_LPC_TYPE_LEVINSON);
ff_bswapdsp_init(s-bdsp);
-ff_flacdsp_init(s-flac_dsp, avctx-sample_fmt,
+ff_flacdsp_init(s-flac_dsp, avctx-sample_fmt, channels,
avctx-bits_per_raw_sample);
dprint_compression_options(s);
diff --git a/libavcodec/x86/flacdsp.asm b/libavcodec/x86/flacdsp.asm
index 37ee87b..4091e16 100644
--- a/libavcodec/x86/flacdsp.asm
+++ b/libavcodec/x86/flacdsp.asm
@@ -72,3 +72,196 @@ ALIGN 16
LPC_32 xop
%endif
LPC_32 sse4
+
+;-
+;void ff_flac_decorrelate_{ls,rs,ms,indep2}_16_sse2(uint8_t **out, int32_t
**in, int channels,
+; int len, int shift);
