virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-12 19:18:44 +02:00
Code Issues Releases Activity
c3d36e1b3d
FFmpeg/libavutil/tx_template.c

1511 lines
62 KiB
C
Raw Normal View History

lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
/*
lavu/tx: refactor power-of-two FFT This commit refactors the power-of-two FFT, making it faster and halving the size of all tables, making the code much smaller on all systems. This removes the big/small pass split, because on modern systems the "big" pass is always faster, and even on older machines there is no measurable speed difference.
2021-04-10 03:47:18 +02:00
* Copyright (c) Lynne
*
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
* Power of two FFT:
lavu/tx: refactor power-of-two FFT This commit refactors the power-of-two FFT, making it faster and halving the size of all tables, making the code much smaller on all systems. This removes the big/small pass split, because on modern systems the "big" pass is always faster, and even on older machines there is no measurable speed difference.
2021-04-10 03:47:18 +02:00
* Copyright (c) Lynne
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
* Copyright (c) 2008 Loren Merritt
* Copyright (c) 2002 Fabrice Bellard
* Partly based on libdjbfft by D. J. Bernstein
*
* 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
*/
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
#define TABLE_DEF(name, size) \
DECLARE_ALIGNED(32, TXSample, TX_TAB(ff_tx_tab_ ##name))[size]
avutil/tx_template: Avoid code duplication Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 14:56:50 +02:00
#define SR_POW2_TABLES \
SR_TABLE(8) \
SR_TABLE(16) \
SR_TABLE(32) \
SR_TABLE(64) \
SR_TABLE(128) \
SR_TABLE(256) \
SR_TABLE(512) \
SR_TABLE(1024) \
SR_TABLE(2048) \
SR_TABLE(4096) \
SR_TABLE(8192) \
SR_TABLE(16384) \
SR_TABLE(32768) \
SR_TABLE(65536) \
SR_TABLE(131072) \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
avutil/tx_template: Avoid code duplication Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 14:56:50 +02:00
#define SR_TABLE(len) \
TABLE_DEF(len, len/4 + 1);
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
/* Power of two tables */
avutil/tx_template: Avoid code duplication Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 14:56:50 +02:00
SR_POW2_TABLES
#undef SR_TABLE
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
/* Other factors' tables */
x86/tx_float: add 15xN PFA FFT AVX SIMD ~4x faster than the C version. The shuffles in the 15pt dim1 are seriously expensive. Not happy with it, but I'm contempt. Can be easily converted to pure AVX by removing all vpermpd/vpermps instructions.
2022-09-19 05:53:01 +02:00
TABLE_DEF(53, 12);
TABLE_DEF( 7, 6);
TABLE_DEF( 9, 8);
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
avutil/tx_template: Move function pointers to const memory This can be achieved by moving the AVOnce out of the structure containing the function pointers; the latter can then be made const. This also has the advantage of eliminating padding in the structure (sizeof(AVOnce) is four here) and allowing the AVOnces to be put into .bss (dependening upon the implementation). Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 15:22:07 +02:00
typedef struct FFTabInitData {
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
void (*func)(void);
avutil/tx_template: Don't waste space for inexistent factors It is possible to avoid the factors array for the power-of-two tables for which said array is unused by using a different structure for initialization for power-of-two tables than for non-power-of-two-tables. This saves 3*15*16B from .data. Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 14:23:07 +02:00
int factors[TX_MAX_SUB]; /* Must be sorted high -> low */
avutil/tx_template: Move function pointers to const memory This can be achieved by moving the AVOnce out of the structure containing the function pointers; the latter can then be made const. This also has the advantage of eliminating padding in the structure (sizeof(AVOnce) is four here) and allowing the AVOnces to be put into .bss (dependening upon the implementation). Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 15:22:07 +02:00
} FFTabInitData;
avutil/tx_template: Don't waste space for inexistent factors It is possible to avoid the factors array for the power-of-two tables for which said array is unused by using a different structure for initialization for power-of-two tables than for non-power-of-two-tables. This saves 3*15*16B from .data. Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 14:23:07 +02:00
avutil/tx_template: Avoid code duplication Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 14:56:50 +02:00
#define SR_TABLE(len) \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static av_cold void TX_TAB(ff_tx_init_tab_ ##len)(void) \
{ \
double freq = 2*M_PI/len; \
TXSample *tab = TX_TAB(ff_tx_tab_ ##len); \
\
for (int i = 0; i < len/4; i++) \
*tab++ = RESCALE(cos(i*freq)); \
\
*tab = 0; \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
}
avutil/tx_template: Avoid code duplication Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 14:56:50 +02:00
SR_POW2_TABLES
#undef SR_TABLE
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
avutil/tx_template: Move function pointers to const memory This can be achieved by moving the AVOnce out of the structure containing the function pointers; the latter can then be made const. This also has the advantage of eliminating padding in the structure (sizeof(AVOnce) is four here) and allowing the AVOnces to be put into .bss (dependening upon the implementation). Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 15:22:07 +02:00
static void (*const sr_tabs_init_funcs[])(void) = {
#define SR_TABLE(len) TX_TAB(ff_tx_init_tab_ ##len),
SR_POW2_TABLES
#undef SR_TABLE
};
static AVOnce sr_tabs_init_once[] = {
#define SR_TABLE(len) AV_ONCE_INIT,
avutil/tx_template: Avoid code duplication Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 14:56:50 +02:00
SR_POW2_TABLES
#undef SR_TABLE
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
};
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
x86/tx_float: add 15xN PFA FFT AVX SIMD ~4x faster than the C version. The shuffles in the 15pt dim1 are seriously expensive. Not happy with it, but I'm contempt. Can be easily converted to pure AVX by removing all vpermpd/vpermps instructions.
2022-09-19 05:53:01 +02:00
static av_cold void TX_TAB(ff_tx_init_tab_53)(void)
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
{
x86/tx_float: add 15xN PFA FFT AVX SIMD ~4x faster than the C version. The shuffles in the 15pt dim1 are seriously expensive. Not happy with it, but I'm contempt. Can be easily converted to pure AVX by removing all vpermpd/vpermps instructions.
2022-09-19 05:53:01 +02:00
/* 5pt, doubled to eliminate AVX lane shuffles */
TX_TAB(ff_tx_tab_53)[0] = RESCALE(cos(2 * M_PI / 5));
TX_TAB(ff_tx_tab_53)[1] = RESCALE(cos(2 * M_PI / 5));
TX_TAB(ff_tx_tab_53)[2] = RESCALE(cos(2 * M_PI / 10));
TX_TAB(ff_tx_tab_53)[3] = RESCALE(cos(2 * M_PI / 10));
TX_TAB(ff_tx_tab_53)[4] = RESCALE(sin(2 * M_PI / 5));
TX_TAB(ff_tx_tab_53)[5] = RESCALE(sin(2 * M_PI / 5));
TX_TAB(ff_tx_tab_53)[6] = RESCALE(sin(2 * M_PI / 10));
TX_TAB(ff_tx_tab_53)[7] = RESCALE(sin(2 * M_PI / 10));
/* 3pt */
TX_TAB(ff_tx_tab_53)[ 8] = RESCALE(cos(2 * M_PI / 12));
TX_TAB(ff_tx_tab_53)[ 9] = RESCALE(cos(2 * M_PI / 12));
TX_TAB(ff_tx_tab_53)[10] = RESCALE(cos(2 * M_PI / 6));
TX_TAB(ff_tx_tab_53)[11] = RESCALE(cos(8 * M_PI / 6));
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
}
x86/tx_float: add 15xN PFA FFT AVX SIMD ~4x faster than the C version. The shuffles in the 15pt dim1 are seriously expensive. Not happy with it, but I'm contempt. Can be easily converted to pure AVX by removing all vpermpd/vpermps instructions.
2022-09-19 05:53:01 +02:00
static av_cold void TX_TAB(ff_tx_init_tab_7)(void)
lavu/tx: add a 7-point FFT and (i)MDCT
2021-04-10 03:51:59 +02:00
{
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TX_TAB(ff_tx_tab_7)[0] = RESCALE(cos(2 * M_PI / 7));
TX_TAB(ff_tx_tab_7)[1] = RESCALE(sin(2 * M_PI / 7));
TX_TAB(ff_tx_tab_7)[2] = RESCALE(sin(2 * M_PI / 28));
TX_TAB(ff_tx_tab_7)[3] = RESCALE(cos(2 * M_PI / 28));
TX_TAB(ff_tx_tab_7)[4] = RESCALE(cos(2 * M_PI / 14));
TX_TAB(ff_tx_tab_7)[5] = RESCALE(sin(2 * M_PI / 14));
lavu/tx: add a 7-point FFT and (i)MDCT
2021-04-10 03:51:59 +02:00
}
x86/tx_float: add 15xN PFA FFT AVX SIMD ~4x faster than the C version. The shuffles in the 15pt dim1 are seriously expensive. Not happy with it, but I'm contempt. Can be easily converted to pure AVX by removing all vpermpd/vpermps instructions.
2022-09-19 05:53:01 +02:00
static av_cold void TX_TAB(ff_tx_init_tab_9)(void)
lavu/tx: add a 9-point FFT and (i)MDCT
2021-04-10 04:19:06 +02:00
{
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TX_TAB(ff_tx_tab_9)[0] = RESCALE(cos(2 * M_PI / 3));
TX_TAB(ff_tx_tab_9)[1] = RESCALE(sin(2 * M_PI / 3));
TX_TAB(ff_tx_tab_9)[2] = RESCALE(cos(2 * M_PI / 9));
TX_TAB(ff_tx_tab_9)[3] = RESCALE(sin(2 * M_PI / 9));
TX_TAB(ff_tx_tab_9)[4] = RESCALE(cos(2 * M_PI / 36));
TX_TAB(ff_tx_tab_9)[5] = RESCALE(sin(2 * M_PI / 36));
TX_TAB(ff_tx_tab_9)[6] = TX_TAB(ff_tx_tab_9)[2] + TX_TAB(ff_tx_tab_9)[5];
TX_TAB(ff_tx_tab_9)[7] = TX_TAB(ff_tx_tab_9)[3] - TX_TAB(ff_tx_tab_9)[4];
lavu/tx: add a 9-point FFT and (i)MDCT
2021-04-10 04:19:06 +02:00
}
avutil/tx_template: Move function pointers to const memory This can be achieved by moving the AVOnce out of the structure containing the function pointers; the latter can then be made const. This also has the advantage of eliminating padding in the structure (sizeof(AVOnce) is four here) and allowing the AVOnces to be put into .bss (dependening upon the implementation). Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 15:22:07 +02:00
static const FFTabInitData nptwo_tabs_init_data[] = {
{ TX_TAB(ff_tx_init_tab_53), { 15, 5, 3 } },
{ TX_TAB(ff_tx_init_tab_9), { 9 } },
{ TX_TAB(ff_tx_init_tab_7), { 7 } },
};
static AVOnce nptwo_tabs_init_once[] = {
AV_ONCE_INIT,
AV_ONCE_INIT,
AV_ONCE_INIT,
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
};
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
av_cold void TX_TAB(ff_tx_init_tabs)(int len)
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
{
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
int factor_2 = ff_ctz(len);
if (factor_2) {
int idx = factor_2 - 3;
for (int i = 0; i <= idx; i++)
avutil/tx_template: Move function pointers to const memory This can be achieved by moving the AVOnce out of the structure containing the function pointers; the latter can then be made const. This also has the advantage of eliminating padding in the structure (sizeof(AVOnce) is four here) and allowing the AVOnces to be put into .bss (dependening upon the implementation). Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 15:22:07 +02:00
ff_thread_once(&sr_tabs_init_once[i],
sr_tabs_init_funcs[i]);
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
len >>= factor_2;
}
avutil/tx_template: Move function pointers to const memory This can be achieved by moving the AVOnce out of the structure containing the function pointers; the latter can then be made const. This also has the advantage of eliminating padding in the structure (sizeof(AVOnce) is four here) and allowing the AVOnces to be put into .bss (dependening upon the implementation). Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 15:22:07 +02:00
for (int i = 0; i < FF_ARRAY_ELEMS(nptwo_tabs_init_data); i++) {
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
int f, f_idx = 0;
if (len <= 1)
return;
avutil/tx_template: Move function pointers to const memory This can be achieved by moving the AVOnce out of the structure containing the function pointers; the latter can then be made const. This also has the advantage of eliminating padding in the structure (sizeof(AVOnce) is four here) and allowing the AVOnces to be put into .bss (dependening upon the implementation). Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 15:22:07 +02:00
while ((f = nptwo_tabs_init_data[i].factors[f_idx++])) {
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
if (f % len)
continue;
avutil/tx_template: Move function pointers to const memory This can be achieved by moving the AVOnce out of the structure containing the function pointers; the latter can then be made const. This also has the advantage of eliminating padding in the structure (sizeof(AVOnce) is four here) and allowing the AVOnces to be put into .bss (dependening upon the implementation). Reviewed-by: Lynne <dev@lynne.ee> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-10-22 15:22:07 +02:00
ff_thread_once(&nptwo_tabs_init_once[i],
nptwo_tabs_init_data[i].func);
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
len /= f;
break;
}
}
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static av_always_inline void fft3(TXComplex *out, TXComplex *in,
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
ptrdiff_t stride)
{
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TXComplex tmp[2];
const TXSample *tab = TX_TAB(ff_tx_tab_53);
lavu/tx: improve 3-point fixed precision There's just no reason not to when its so easy (albeit messy) and its also reducing the precision of all non-power-of-two transforms that use it.
2020-02-14 21:55:00 +02:00
#ifdef TX_INT32
int64_t mtmp[4];
#endif
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
lavu/tx: implement 32 bit fixed point FFT and MDCT Required minimal changes to the code so made sense to implement. FFT and MDCT tested, the output of both was properly rounded. Fun fact: the non-power-of-two fixed-point FFT and MDCT are the fastest ever non-power-of-two fixed-point FFT and MDCT written. This can replace the power of two integer MDCTs in aac and ac3 if the MIPS optimizations are ported across. Unfortunately the ac3 encoder uses a 16-bit fixed point forward transform, unlike the encoder which uses a 32bit inverse transform, so some modifications might be required there. The 3-point FFT is somewhat less accurate than it otherwise could be, having minor rounding errors with bigger transforms. However, this could be improved later, and the way its currently written is the way one would write assembly for it. Similar rounding errors can also be found throughout the power of two FFTs as well, though those are more difficult to correct. Despite this, the integer transforms are more than accurate enough.
2020-02-09 01:13:28 +02:00
BF(tmp[0].re, tmp[1].im, in[1].im, in[2].im);
BF(tmp[0].im, tmp[1].re, in[1].re, in[2].re);
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
out[0*stride].re = in[0].re + tmp[1].re;
out[0*stride].im = in[0].im + tmp[1].im;
lavu/tx: improve 3-point fixed precision There's just no reason not to when its so easy (albeit messy) and its also reducing the precision of all non-power-of-two transforms that use it.
2020-02-14 21:55:00 +02:00
#ifdef TX_INT32
x86/tx_float: add 15xN PFA FFT AVX SIMD ~4x faster than the C version. The shuffles in the 15pt dim1 are seriously expensive. Not happy with it, but I'm contempt. Can be easily converted to pure AVX by removing all vpermpd/vpermps instructions.
2022-09-19 05:53:01 +02:00
mtmp[0] = (int64_t)tab[ 8] * tmp[0].re;
mtmp[1] = (int64_t)tab[ 9] * tmp[0].im;
mtmp[2] = (int64_t)tab[10] * tmp[1].re;
mtmp[3] = (int64_t)tab[10] * tmp[1].im;
lavu/tx: improve 3-point fixed precision There's just no reason not to when its so easy (albeit messy) and its also reducing the precision of all non-power-of-two transforms that use it.
2020-02-14 21:55:00 +02:00
out[1*stride].re = in[0].re - (mtmp[2] + mtmp[0] + 0x40000000 >> 31);
out[1*stride].im = in[0].im - (mtmp[3] - mtmp[1] + 0x40000000 >> 31);
out[2*stride].re = in[0].re - (mtmp[2] - mtmp[0] + 0x40000000 >> 31);
out[2*stride].im = in[0].im - (mtmp[3] + mtmp[1] + 0x40000000 >> 31);
#else
x86/tx_float: add 15xN PFA FFT AVX SIMD ~4x faster than the C version. The shuffles in the 15pt dim1 are seriously expensive. Not happy with it, but I'm contempt. Can be easily converted to pure AVX by removing all vpermpd/vpermps instructions.
2022-09-19 05:53:01 +02:00
tmp[0].re = tab[ 8] * tmp[0].re;
tmp[0].im = tab[ 9] * tmp[0].im;
tmp[1].re = tab[10] * tmp[1].re;
tmp[1].im = tab[10] * tmp[1].im;
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
out[1*stride].re = in[0].re - tmp[1].re + tmp[0].re;
out[1*stride].im = in[0].im - tmp[1].im - tmp[0].im;
out[2*stride].re = in[0].re - tmp[1].re - tmp[0].re;
out[2*stride].im = in[0].im - tmp[1].im + tmp[0].im;
lavu/tx: improve 3-point fixed precision There's just no reason not to when its so easy (albeit messy) and its also reducing the precision of all non-power-of-two transforms that use it.
2020-02-14 21:55:00 +02:00
#endif
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
#define DECL_FFT5(NAME, D0, D1, D2, D3, D4) \
static av_always_inline void NAME(TXComplex *out, TXComplex *in, \
ptrdiff_t stride) \
{ \
TXComplex z0[4], t[6]; \
const TXSample *tab = TX_TAB(ff_tx_tab_53); \
\
BF(t[1].im, t[0].re, in[1].re, in[4].re); \
BF(t[1].re, t[0].im, in[1].im, in[4].im); \
BF(t[3].im, t[2].re, in[2].re, in[3].re); \
BF(t[3].re, t[2].im, in[2].im, in[3].im); \
\
out[D0*stride].re = in[0].re + t[0].re + t[2].re; \
out[D0*stride].im = in[0].im + t[0].im + t[2].im; \
\
x86/tx_float: add 15xN PFA FFT AVX SIMD ~4x faster than the C version. The shuffles in the 15pt dim1 are seriously expensive. Not happy with it, but I'm contempt. Can be easily converted to pure AVX by removing all vpermpd/vpermps instructions.
2022-09-19 05:53:01 +02:00
SMUL(t[4].re, t[0].re, tab[0], tab[2], t[2].re, t[0].re); \
SMUL(t[4].im, t[0].im, tab[0], tab[2], t[2].im, t[0].im); \
CMUL(t[5].re, t[1].re, tab[4], tab[6], t[3].re, t[1].re); \
CMUL(t[5].im, t[1].im, tab[4], tab[6], t[3].im, t[1].im); \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
\
BF(z0[0].re, z0[3].re, t[0].re, t[1].re); \
BF(z0[0].im, z0[3].im, t[0].im, t[1].im); \
BF(z0[2].re, z0[1].re, t[4].re, t[5].re); \
BF(z0[2].im, z0[1].im, t[4].im, t[5].im); \
\
out[D1*stride].re = in[0].re + z0[3].re; \
out[D1*stride].im = in[0].im + z0[0].im; \
out[D2*stride].re = in[0].re + z0[2].re; \
out[D2*stride].im = in[0].im + z0[1].im; \
out[D3*stride].re = in[0].re + z0[1].re; \
out[D3*stride].im = in[0].im + z0[2].im; \
out[D4*stride].re = in[0].re + z0[0].re; \
out[D4*stride].im = in[0].im + z0[3].im; \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
}
DECL_FFT5(fft5, 0, 1, 2, 3, 4)
DECL_FFT5(fft5_m1, 0, 6, 12, 3, 9)
DECL_FFT5(fft5_m2, 10, 1, 7, 13, 4)
DECL_FFT5(fft5_m3, 5, 11, 2, 8, 14)
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static av_always_inline void fft7(TXComplex *out, TXComplex *in,
lavu/tx: add a 7-point FFT and (i)MDCT
2021-04-10 03:51:59 +02:00
ptrdiff_t stride)
{
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TXComplex t[6], z[3];
const TXComplex *tab = (const TXComplex *)TX_TAB(ff_tx_tab_7);
lavu/tx: add a 7-point FFT and (i)MDCT
2021-04-10 03:51:59 +02:00
#ifdef TX_INT32
int64_t mtmp[12];
#endif
BF(t[1].re, t[0].re, in[1].re, in[6].re);
BF(t[1].im, t[0].im, in[1].im, in[6].im);
BF(t[3].re, t[2].re, in[2].re, in[5].re);
BF(t[3].im, t[2].im, in[2].im, in[5].im);
BF(t[5].re, t[4].re, in[3].re, in[4].re);
BF(t[5].im, t[4].im, in[3].im, in[4].im);
out[0*stride].re = in[0].re + t[0].re + t[2].re + t[4].re;
out[0*stride].im = in[0].im + t[0].im + t[2].im + t[4].im;
#ifdef TX_INT32 /* NOTE: it's possible to do this with 16 mults but 72 adds */
mtmp[ 0] = ((int64_t)tab[0].re)*t[0].re - ((int64_t)tab[2].re)*t[4].re;
mtmp[ 1] = ((int64_t)tab[0].re)*t[4].re - ((int64_t)tab[1].re)*t[0].re;
mtmp[ 2] = ((int64_t)tab[0].re)*t[2].re - ((int64_t)tab[2].re)*t[0].re;
mtmp[ 3] = ((int64_t)tab[0].re)*t[0].im - ((int64_t)tab[1].re)*t[2].im;
mtmp[ 4] = ((int64_t)tab[0].re)*t[4].im - ((int64_t)tab[1].re)*t[0].im;
mtmp[ 5] = ((int64_t)tab[0].re)*t[2].im - ((int64_t)tab[2].re)*t[0].im;
mtmp[ 6] = ((int64_t)tab[2].im)*t[1].im + ((int64_t)tab[1].im)*t[5].im;
mtmp[ 7] = ((int64_t)tab[0].im)*t[5].im + ((int64_t)tab[2].im)*t[3].im;
mtmp[ 8] = ((int64_t)tab[2].im)*t[5].im + ((int64_t)tab[1].im)*t[3].im;
mtmp[ 9] = ((int64_t)tab[0].im)*t[1].re + ((int64_t)tab[1].im)*t[3].re;
mtmp[10] = ((int64_t)tab[2].im)*t[3].re + ((int64_t)tab[0].im)*t[5].re;
mtmp[11] = ((int64_t)tab[2].im)*t[1].re + ((int64_t)tab[1].im)*t[5].re;
z[0].re = (int32_t)(mtmp[ 0] - ((int64_t)tab[1].re)*t[2].re + 0x40000000 >> 31);
z[1].re = (int32_t)(mtmp[ 1] - ((int64_t)tab[2].re)*t[2].re + 0x40000000 >> 31);
z[2].re = (int32_t)(mtmp[ 2] - ((int64_t)tab[1].re)*t[4].re + 0x40000000 >> 31);
z[0].im = (int32_t)(mtmp[ 3] - ((int64_t)tab[2].re)*t[4].im + 0x40000000 >> 31);
z[1].im = (int32_t)(mtmp[ 4] - ((int64_t)tab[2].re)*t[2].im + 0x40000000 >> 31);
z[2].im = (int32_t)(mtmp[ 5] - ((int64_t)tab[1].re)*t[4].im + 0x40000000 >> 31);
t[0].re = (int32_t)(mtmp[ 6] - ((int64_t)tab[0].im)*t[3].im + 0x40000000 >> 31);
t[2].re = (int32_t)(mtmp[ 7] - ((int64_t)tab[1].im)*t[1].im + 0x40000000 >> 31);
t[4].re = (int32_t)(mtmp[ 8] + ((int64_t)tab[0].im)*t[1].im + 0x40000000 >> 31);
t[0].im = (int32_t)(mtmp[ 9] + ((int64_t)tab[2].im)*t[5].re + 0x40000000 >> 31);
t[2].im = (int32_t)(mtmp[10] - ((int64_t)tab[1].im)*t[1].re + 0x40000000 >> 31);
t[4].im = (int32_t)(mtmp[11] - ((int64_t)tab[0].im)*t[3].re + 0x40000000 >> 31);
#else
z[0].re = tab[0].re*t[0].re - tab[2].re*t[4].re - tab[1].re*t[2].re;
z[1].re = tab[0].re*t[4].re - tab[1].re*t[0].re - tab[2].re*t[2].re;
z[2].re = tab[0].re*t[2].re - tab[2].re*t[0].re - tab[1].re*t[4].re;
z[0].im = tab[0].re*t[0].im - tab[1].re*t[2].im - tab[2].re*t[4].im;
z[1].im = tab[0].re*t[4].im - tab[1].re*t[0].im - tab[2].re*t[2].im;
z[2].im = tab[0].re*t[2].im - tab[2].re*t[0].im - tab[1].re*t[4].im;
/* It's possible to do t[4].re and t[0].im with 2 multiplies only by
* multiplying the sum of all with the average of the twiddles */
t[0].re = tab[2].im*t[1].im + tab[1].im*t[5].im - tab[0].im*t[3].im;
t[2].re = tab[0].im*t[5].im + tab[2].im*t[3].im - tab[1].im*t[1].im;
t[4].re = tab[2].im*t[5].im + tab[1].im*t[3].im + tab[0].im*t[1].im;
t[0].im = tab[0].im*t[1].re + tab[1].im*t[3].re + tab[2].im*t[5].re;
t[2].im = tab[2].im*t[3].re + tab[0].im*t[5].re - tab[1].im*t[1].re;
t[4].im = tab[2].im*t[1].re + tab[1].im*t[5].re - tab[0].im*t[3].re;
#endif
BF(t[1].re, z[0].re, z[0].re, t[4].re);
BF(t[3].re, z[1].re, z[1].re, t[2].re);
BF(t[5].re, z[2].re, z[2].re, t[0].re);
BF(t[1].im, z[0].im, z[0].im, t[0].im);
BF(t[3].im, z[1].im, z[1].im, t[2].im);
BF(t[5].im, z[2].im, z[2].im, t[4].im);
out[1*stride].re = in[0].re + z[0].re;
out[1*stride].im = in[0].im + t[1].im;
out[2*stride].re = in[0].re + t[3].re;
out[2*stride].im = in[0].im + z[1].im;
out[3*stride].re = in[0].re + z[2].re;
out[3*stride].im = in[0].im + t[5].im;
out[4*stride].re = in[0].re + t[5].re;
out[4*stride].im = in[0].im + z[2].im;
out[5*stride].re = in[0].re + z[1].re;
out[5*stride].im = in[0].im + t[3].im;
out[6*stride].re = in[0].re + t[1].re;
out[6*stride].im = in[0].im + z[0].im;
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static av_always_inline void fft9(TXComplex *out, TXComplex *in,
lavu/tx: add a 9-point FFT and (i)MDCT
2021-04-10 04:19:06 +02:00
ptrdiff_t stride)
{
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
const TXComplex *tab = (const TXComplex *)TX_TAB(ff_tx_tab_9);
TXComplex t[16], w[4], x[5], y[5], z[2];
lavu/tx: add a 9-point FFT and (i)MDCT
2021-04-10 04:19:06 +02:00
#ifdef TX_INT32
int64_t mtmp[12];
#endif
BF(t[1].re, t[0].re, in[1].re, in[8].re);
BF(t[1].im, t[0].im, in[1].im, in[8].im);
BF(t[3].re, t[2].re, in[2].re, in[7].re);
BF(t[3].im, t[2].im, in[2].im, in[7].im);
BF(t[5].re, t[4].re, in[3].re, in[6].re);
BF(t[5].im, t[4].im, in[3].im, in[6].im);
BF(t[7].re, t[6].re, in[4].re, in[5].re);
BF(t[7].im, t[6].im, in[4].im, in[5].im);
w[0].re = t[0].re - t[6].re;
w[0].im = t[0].im - t[6].im;
w[1].re = t[2].re - t[6].re;
w[1].im = t[2].im - t[6].im;
w[2].re = t[1].re - t[7].re;
w[2].im = t[1].im - t[7].im;
w[3].re = t[3].re + t[7].re;
w[3].im = t[3].im + t[7].im;
z[0].re = in[0].re + t[4].re;
z[0].im = in[0].im + t[4].im;
z[1].re = t[0].re + t[2].re + t[6].re;
z[1].im = t[0].im + t[2].im + t[6].im;
out[0*stride].re = z[0].re + z[1].re;
out[0*stride].im = z[0].im + z[1].im;
#ifdef TX_INT32
mtmp[0] = t[1].re - t[3].re + t[7].re;
mtmp[1] = t[1].im - t[3].im + t[7].im;
y[3].re = (int32_t)(((int64_t)tab[0].im)*mtmp[0] + 0x40000000 >> 31);
y[3].im = (int32_t)(((int64_t)tab[0].im)*mtmp[1] + 0x40000000 >> 31);
mtmp[0] = (int32_t)(((int64_t)tab[0].re)*z[1].re + 0x40000000 >> 31);
mtmp[1] = (int32_t)(((int64_t)tab[0].re)*z[1].im + 0x40000000 >> 31);
mtmp[2] = (int32_t)(((int64_t)tab[0].re)*t[4].re + 0x40000000 >> 31);
mtmp[3] = (int32_t)(((int64_t)tab[0].re)*t[4].im + 0x40000000 >> 31);
x[3].re = z[0].re + (int32_t)mtmp[0];
x[3].im = z[0].im + (int32_t)mtmp[1];
z[0].re = in[0].re + (int32_t)mtmp[2];
z[0].im = in[0].im + (int32_t)mtmp[3];
mtmp[0] = ((int64_t)tab[1].re)*w[0].re;
mtmp[1] = ((int64_t)tab[1].re)*w[0].im;
mtmp[2] = ((int64_t)tab[2].im)*w[0].re;
mtmp[3] = ((int64_t)tab[2].im)*w[0].im;
mtmp[4] = ((int64_t)tab[1].im)*w[2].re;
mtmp[5] = ((int64_t)tab[1].im)*w[2].im;
mtmp[6] = ((int64_t)tab[2].re)*w[2].re;
mtmp[7] = ((int64_t)tab[2].re)*w[2].im;
x[1].re = (int32_t)(mtmp[0] + ((int64_t)tab[2].im)*w[1].re + 0x40000000 >> 31);
x[1].im = (int32_t)(mtmp[1] + ((int64_t)tab[2].im)*w[1].im + 0x40000000 >> 31);
x[2].re = (int32_t)(mtmp[2] - ((int64_t)tab[3].re)*w[1].re + 0x40000000 >> 31);
x[2].im = (int32_t)(mtmp[3] - ((int64_t)tab[3].re)*w[1].im + 0x40000000 >> 31);
y[1].re = (int32_t)(mtmp[4] + ((int64_t)tab[2].re)*w[3].re + 0x40000000 >> 31);
y[1].im = (int32_t)(mtmp[5] + ((int64_t)tab[2].re)*w[3].im + 0x40000000 >> 31);
y[2].re = (int32_t)(mtmp[6] - ((int64_t)tab[3].im)*w[3].re + 0x40000000 >> 31);
y[2].im = (int32_t)(mtmp[7] - ((int64_t)tab[3].im)*w[3].im + 0x40000000 >> 31);
y[0].re = (int32_t)(((int64_t)tab[0].im)*t[5].re + 0x40000000 >> 31);
y[0].im = (int32_t)(((int64_t)tab[0].im)*t[5].im + 0x40000000 >> 31);
#else
y[3].re = tab[0].im*(t[1].re - t[3].re + t[7].re);
y[3].im = tab[0].im*(t[1].im - t[3].im + t[7].im);
x[3].re = z[0].re + tab[0].re*z[1].re;
x[3].im = z[0].im + tab[0].re*z[1].im;
z[0].re = in[0].re + tab[0].re*t[4].re;
z[0].im = in[0].im + tab[0].re*t[4].im;
x[1].re = tab[1].re*w[0].re + tab[2].im*w[1].re;
x[1].im = tab[1].re*w[0].im + tab[2].im*w[1].im;
x[2].re = tab[2].im*w[0].re - tab[3].re*w[1].re;
x[2].im = tab[2].im*w[0].im - tab[3].re*w[1].im;
y[1].re = tab[1].im*w[2].re + tab[2].re*w[3].re;
y[1].im = tab[1].im*w[2].im + tab[2].re*w[3].im;
y[2].re = tab[2].re*w[2].re - tab[3].im*w[3].re;
y[2].im = tab[2].re*w[2].im - tab[3].im*w[3].im;
y[0].re = tab[0].im*t[5].re;
y[0].im = tab[0].im*t[5].im;
#endif
x[4].re = x[1].re + x[2].re;
x[4].im = x[1].im + x[2].im;
y[4].re = y[1].re - y[2].re;
y[4].im = y[1].im - y[2].im;
x[1].re = z[0].re + x[1].re;
x[1].im = z[0].im + x[1].im;
y[1].re = y[0].re + y[1].re;
y[1].im = y[0].im + y[1].im;
x[2].re = z[0].re + x[2].re;
x[2].im = z[0].im + x[2].im;
y[2].re = y[2].re - y[0].re;
y[2].im = y[2].im - y[0].im;
x[4].re = z[0].re - x[4].re;
x[4].im = z[0].im - x[4].im;
y[4].re = y[0].re - y[4].re;
y[4].im = y[0].im - y[4].im;
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
out[1*stride] = (TXComplex){ x[1].re + y[1].im, x[1].im - y[1].re };
out[2*stride] = (TXComplex){ x[2].re + y[2].im, x[2].im - y[2].re };
out[3*stride] = (TXComplex){ x[3].re + y[3].im, x[3].im - y[3].re };
out[4*stride] = (TXComplex){ x[4].re + y[4].im, x[4].im - y[4].re };
out[5*stride] = (TXComplex){ x[4].re - y[4].im, x[4].im + y[4].re };
out[6*stride] = (TXComplex){ x[3].re - y[3].im, x[3].im + y[3].re };
out[7*stride] = (TXComplex){ x[2].re - y[2].im, x[2].im + y[2].re };
out[8*stride] = (TXComplex){ x[1].re - y[1].im, x[1].im + y[1].re };
lavu/tx: add a 9-point FFT and (i)MDCT
2021-04-10 04:19:06 +02:00
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static av_always_inline void fft15(TXComplex *out, TXComplex *in,
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
ptrdiff_t stride)
{
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TXComplex tmp[15];
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
for (int i = 0; i < 5; i++)
fft3(tmp + i, in + i*3, 5);
fft5_m1(out, tmp + 0, stride);
fft5_m2(out, tmp + 5, stride);
fft5_m3(out, tmp + 10, stride);
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
#define BUTTERFLIES(a0, a1, a2, a3) \
lavu/tx: refactor power-of-two FFT This commit refactors the power-of-two FFT, making it faster and halving the size of all tables, making the code much smaller on all systems. This removes the big/small pass split, because on modern systems the "big" pass is always faster, and even on older machines there is no measurable speed difference.
2021-04-10 03:47:18 +02:00
do { \
r0=a0.re; \
i0=a0.im; \
r1=a1.re; \
i1=a1.im; \
BF(t3, t5, t5, t1); \
BF(a2.re, a0.re, r0, t5); \
BF(a3.im, a1.im, i1, t3); \
BF(t4, t6, t2, t6); \
BF(a3.re, a1.re, r1, t4); \
BF(a2.im, a0.im, i0, t6); \
} while (0)
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
#define TRANSFORM(a0, a1, a2, a3, wre, wim) \
lavu/tx: refactor power-of-two FFT This commit refactors the power-of-two FFT, making it faster and halving the size of all tables, making the code much smaller on all systems. This removes the big/small pass split, because on modern systems the "big" pass is always faster, and even on older machines there is no measurable speed difference.
2021-04-10 03:47:18 +02:00
do { \
CMUL(t1, t2, a2.re, a2.im, wre, -wim); \
CMUL(t5, t6, a3.re, a3.im, wre, wim); \
BUTTERFLIES(a0, a1, a2, a3); \
} while (0)
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
/* z[0...8n-1], w[1...2n-1] */
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static inline void TX_NAME(ff_tx_fft_sr_combine)(TXComplex *z,
const TXSample *cos, int len)
lavu/tx: refactor power-of-two FFT This commit refactors the power-of-two FFT, making it faster and halving the size of all tables, making the code much smaller on all systems. This removes the big/small pass split, because on modern systems the "big" pass is always faster, and even on older machines there is no measurable speed difference.
2021-04-10 03:47:18 +02:00
{
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
int o1 = 2*len;
int o2 = 4*len;
int o3 = 6*len;
const TXSample *wim = cos + o1 - 7;
TXSample t1, t2, t3, t4, t5, t6, r0, i0, r1, i1;
lavu/tx: refactor power-of-two FFT This commit refactors the power-of-two FFT, making it faster and halving the size of all tables, making the code much smaller on all systems. This removes the big/small pass split, because on modern systems the "big" pass is always faster, and even on older machines there is no measurable speed difference.
2021-04-10 03:47:18 +02:00
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
for (int i = 0; i < len; i += 4) {
lavu/tx: refactor power-of-two FFT This commit refactors the power-of-two FFT, making it faster and halving the size of all tables, making the code much smaller on all systems. This removes the big/small pass split, because on modern systems the "big" pass is always faster, and even on older machines there is no measurable speed difference.
2021-04-10 03:47:18 +02:00
TRANSFORM(z[0], z[o1 + 0], z[o2 + 0], z[o3 + 0], cos[0], wim[7]);
TRANSFORM(z[2], z[o1 + 2], z[o2 + 2], z[o3 + 2], cos[2], wim[5]);
TRANSFORM(z[4], z[o1 + 4], z[o2 + 4], z[o3 + 4], cos[4], wim[3]);
TRANSFORM(z[6], z[o1 + 6], z[o2 + 6], z[o3 + 6], cos[6], wim[1]);
TRANSFORM(z[1], z[o1 + 1], z[o2 + 1], z[o3 + 1], cos[1], wim[6]);
TRANSFORM(z[3], z[o1 + 3], z[o2 + 3], z[o3 + 3], cos[3], wim[4]);
TRANSFORM(z[5], z[o1 + 5], z[o2 + 5], z[o3 + 5], cos[5], wim[2]);
TRANSFORM(z[7], z[o1 + 7], z[o2 + 7], z[o3 + 7], cos[7], wim[0]);
z += 2*4;
cos += 2*4;
wim -= 2*4;
}
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static av_cold int TX_NAME(ff_tx_fft_sr_codelet_init)(AVTXContext *s,
const FFTXCodelet *cd,
uint64_t flags,
FFTXCodeletOptions *opts,
int len, int inv,
const void *scale)
{
TX_TAB(ff_tx_init_tabs)(len);
return ff_tx_gen_ptwo_revtab(s, opts ? opts->invert_lookup : 1);
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
#define DECL_SR_CODELET_DEF(n) \
static const FFTXCodelet TX_NAME(ff_tx_fft##n##_ns_def) = { \
.name = TX_NAME_STR("fft" #n "_ns"), \
.function = TX_NAME(ff_tx_fft##n##_ns), \
.type = TX_TYPE(FFT), \
.flags = AV_TX_INPLACE | AV_TX_UNALIGNED | \
FF_TX_PRESHUFFLE, \
.factors[0] = 2, \
.min_len = n, \
.max_len = n, \
.init = TX_NAME(ff_tx_fft_sr_codelet_init), \
.cpu_flags = FF_TX_CPU_FLAGS_ALL, \
.prio = FF_TX_PRIO_BASE, \
};
#define DECL_SR_CODELET(n, n2, n4) \
static void TX_NAME(ff_tx_fft##n##_ns)(AVTXContext *s, void *dst, \
void *src, ptrdiff_t stride) \
{ \
TXComplex *z = dst; \
const TXSample *cos = TX_TAB(ff_tx_tab_##n); \
\
TX_NAME(ff_tx_fft##n2##_ns)(s, z, z, stride); \
TX_NAME(ff_tx_fft##n4##_ns)(s, z + n4*2, z + n4*2, stride); \
TX_NAME(ff_tx_fft##n4##_ns)(s, z + n4*3, z + n4*3, stride); \
TX_NAME(ff_tx_fft_sr_combine)(z, cos, n4 >> 1); \
} \
\
DECL_SR_CODELET_DEF(n)
static void TX_NAME(ff_tx_fft2_ns)(AVTXContext *s, void *dst,
void *src, ptrdiff_t stride)
lavu/tx: add 2-point FFT transform By itself, this allows 6-point, 10-point and 30-point transforms. When the 9-point transform is added it allows for 18-point FFT, and also for a 36-point MDCT (used by MP3).
2020-02-25 08:49:26 +02:00
{
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TXComplex *z = dst;
TXComplex tmp;
lavu/tx: add 2-point FFT transform By itself, this allows 6-point, 10-point and 30-point transforms. When the 9-point transform is added it allows for 18-point FFT, and also for a 36-point MDCT (used by MP3).
2020-02-25 08:49:26 +02:00
BF(tmp.re, z[0].re, z[0].re, z[1].re);
BF(tmp.im, z[0].im, z[0].im, z[1].im);
z[1] = tmp;
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static void TX_NAME(ff_tx_fft4_ns)(AVTXContext *s, void *dst,
void *src, ptrdiff_t stride)
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
{
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TXComplex *z = dst;
TXSample t1, t2, t3, t4, t5, t6, t7, t8;
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
BF(t3, t1, z[0].re, z[1].re);
BF(t8, t6, z[3].re, z[2].re);
BF(z[2].re, z[0].re, t1, t6);
BF(t4, t2, z[0].im, z[1].im);
BF(t7, t5, z[2].im, z[3].im);
BF(z[3].im, z[1].im, t4, t8);
BF(z[3].re, z[1].re, t3, t7);
BF(z[2].im, z[0].im, t2, t5);
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static void TX_NAME(ff_tx_fft8_ns)(AVTXContext *s, void *dst,
void *src, ptrdiff_t stride)
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
{
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TXComplex *z = dst;
TXSample t1, t2, t3, t4, t5, t6, r0, i0, r1, i1;
const TXSample cos = TX_TAB(ff_tx_tab_8)[1];
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TX_NAME(ff_tx_fft4_ns)(s, z, z, stride);
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
BF(t1, z[5].re, z[4].re, -z[5].re);
BF(t2, z[5].im, z[4].im, -z[5].im);
BF(t5, z[7].re, z[6].re, -z[7].re);
BF(t6, z[7].im, z[6].im, -z[7].im);
lavu/tx: refactor power-of-two FFT This commit refactors the power-of-two FFT, making it faster and halving the size of all tables, making the code much smaller on all systems. This removes the big/small pass split, because on modern systems the "big" pass is always faster, and even on older machines there is no measurable speed difference.
2021-04-10 03:47:18 +02:00
BUTTERFLIES(z[0], z[2], z[4], z[6]);
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TRANSFORM(z[1], z[3], z[5], z[7], cos, cos);
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static void TX_NAME(ff_tx_fft16_ns)(AVTXContext *s, void *dst,
void *src, ptrdiff_t stride)
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
{
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TXComplex *z = dst;
const TXSample *cos = TX_TAB(ff_tx_tab_16);
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TXSample t1, t2, t3, t4, t5, t6, r0, i0, r1, i1;
TXSample cos_16_1 = cos[1];
TXSample cos_16_2 = cos[2];
TXSample cos_16_3 = cos[3];
TX_NAME(ff_tx_fft8_ns)(s, z + 0, z + 0, stride);
TX_NAME(ff_tx_fft4_ns)(s, z + 8, z + 8, stride);
TX_NAME(ff_tx_fft4_ns)(s, z + 12, z + 12, stride);
lavu/tx: refactor power-of-two FFT This commit refactors the power-of-two FFT, making it faster and halving the size of all tables, making the code much smaller on all systems. This removes the big/small pass split, because on modern systems the "big" pass is always faster, and even on older machines there is no measurable speed difference.
2021-04-10 03:47:18 +02:00
t1 = z[ 8].re;
t2 = z[ 8].im;
t5 = z[12].re;
t6 = z[12].im;
BUTTERFLIES(z[0], z[4], z[8], z[12]);
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
lavu/tx: refactor power-of-two FFT This commit refactors the power-of-two FFT, making it faster and halving the size of all tables, making the code much smaller on all systems. This removes the big/small pass split, because on modern systems the "big" pass is always faster, and even on older machines there is no measurable speed difference.
2021-04-10 03:47:18 +02:00
TRANSFORM(z[ 2], z[ 6], z[10], z[14], cos_16_2, cos_16_2);
TRANSFORM(z[ 1], z[ 5], z[ 9], z[13], cos_16_1, cos_16_3);
TRANSFORM(z[ 3], z[ 7], z[11], z[15], cos_16_3, cos_16_1);
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
DECL_SR_CODELET_DEF(2)
DECL_SR_CODELET_DEF(4)
DECL_SR_CODELET_DEF(8)
DECL_SR_CODELET_DEF(16)
DECL_SR_CODELET(32,16,8)
DECL_SR_CODELET(64,32,16)
DECL_SR_CODELET(128,64,32)
DECL_SR_CODELET(256,128,64)
DECL_SR_CODELET(512,256,128)
DECL_SR_CODELET(1024,512,256)
DECL_SR_CODELET(2048,1024,512)
DECL_SR_CODELET(4096,2048,1024)
DECL_SR_CODELET(8192,4096,2048)
DECL_SR_CODELET(16384,8192,4096)
DECL_SR_CODELET(32768,16384,8192)
DECL_SR_CODELET(65536,32768,16384)
DECL_SR_CODELET(131072,65536,32768)
static av_cold int TX_NAME(ff_tx_fft_sr_init)(AVTXContext *s,
const FFTXCodelet *cd,
uint64_t flags,
FFTXCodeletOptions *opts,
int len, int inv,
const void *scale)
{
int ret;
int is_inplace = !!(flags & AV_TX_INPLACE);
FFTXCodeletOptions sub_opts = { .invert_lookup = !is_inplace };
flags &= ~FF_TX_OUT_OF_PLACE; /* We want the subtransform to be */
flags |= AV_TX_INPLACE; /* in-place */
flags |= FF_TX_PRESHUFFLE; /* This function handles the permute step */
if ((ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts, len, inv, scale)))
return ret;
if (is_inplace && (ret = ff_tx_gen_ptwo_inplace_revtab_idx(s)))
return ret;
return 0;
}
static void TX_NAME(ff_tx_fft_sr)(AVTXContext *s, void *_dst,
void *_src, ptrdiff_t stride)
{
TXComplex *src = _src;
TXComplex *dst = _dst;
int *map = s->sub[0].map;
int len = s->len;
/* Compilers can't vectorize this anyway without assuming AVX2, which they
* generally don't, at least without -march=native -mtune=native */
for (int i = 0; i < len; i++)
dst[i] = src[map[i]];
s->fn[0](&s->sub[0], dst, dst, stride);
}
static void TX_NAME(ff_tx_fft_sr_inplace)(AVTXContext *s, void *_dst,
void *_src, ptrdiff_t stride)
{
TXComplex *dst = _dst;
TXComplex tmp;
const int *map = s->sub->map;
const int *inplace_idx = s->map;
int src_idx, dst_idx;
src_idx = *inplace_idx++;
do {
tmp = dst[src_idx];
dst_idx = map[src_idx];
do {
FFSWAP(TXComplex, tmp, dst[dst_idx]);
dst_idx = map[dst_idx];
} while (dst_idx != src_idx); /* Can be > as well, but was less predictable */
dst[dst_idx] = tmp;
} while ((src_idx = *inplace_idx++));
s->fn[0](&s->sub[0], dst, dst, stride);
}
static const FFTXCodelet TX_NAME(ff_tx_fft_sr_def) = {
.name = TX_NAME_STR("fft_sr"),
.function = TX_NAME(ff_tx_fft_sr),
.type = TX_TYPE(FFT),
.flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE,
.factors[0] = 2,
.min_len = 2,
.max_len = TX_LEN_UNLIMITED,
.init = TX_NAME(ff_tx_fft_sr_init),
.cpu_flags = FF_TX_CPU_FLAGS_ALL,
.prio = FF_TX_PRIO_BASE,
};
static const FFTXCodelet TX_NAME(ff_tx_fft_sr_inplace_def) = {
.name = TX_NAME_STR("fft_sr_inplace"),
.function = TX_NAME(ff_tx_fft_sr_inplace),
.type = TX_TYPE(FFT),
.flags = AV_TX_UNALIGNED | AV_TX_INPLACE,
.factors[0] = 2,
.min_len = 2,
.max_len = TX_LEN_UNLIMITED,
.init = TX_NAME(ff_tx_fft_sr_init),
.cpu_flags = FF_TX_CPU_FLAGS_ALL,
.prio = FF_TX_PRIO_BASE,
};
static void TX_NAME(ff_tx_fft_naive)(AVTXContext *s, void *_dst, void *_src,
ptrdiff_t stride)
{
TXComplex *src = _src;
TXComplex *dst = _dst;
const int n = s->len;
double phase = s->inv ? 2.0*M_PI/n : -2.0*M_PI/n;
for(int i = 0; i < n; i++) {
TXComplex tmp = { 0 };
for(int j = 0; j < n; j++) {
const double factor = phase*i*j;
const TXComplex mult = {
RESCALE(cos(factor)),
RESCALE(sin(factor)),
};
TXComplex res;
CMUL3(res, src[j], mult);
tmp.re += res.re;
tmp.im += res.im;
}
dst[i] = tmp;
}
}
static const FFTXCodelet TX_NAME(ff_tx_fft_naive_def) = {
.name = TX_NAME_STR("fft_naive"),
.function = TX_NAME(ff_tx_fft_naive),
.type = TX_TYPE(FFT),
.flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE,
.factors[0] = TX_FACTOR_ANY,
.min_len = 2,
.max_len = TX_LEN_UNLIMITED,
.init = NULL,
.cpu_flags = FF_TX_CPU_FLAGS_ALL,
.prio = FF_TX_PRIO_MIN,
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
};
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static av_cold int TX_NAME(ff_tx_fft_pfa_init)(AVTXContext *s,
const FFTXCodelet *cd,
uint64_t flags,
FFTXCodeletOptions *opts,
int len, int inv,
const void *scale)
{
int ret;
int sub_len = len / cd->factors[0];
FFTXCodeletOptions sub_opts = { .invert_lookup = 0 };
flags &= ~FF_TX_OUT_OF_PLACE; /* We want the subtransform to be */
flags |= AV_TX_INPLACE; /* in-place */
flags |= FF_TX_PRESHUFFLE; /* This function handles the permute step */
if ((ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts,
sub_len, inv, scale)))
return ret;
if ((ret = ff_tx_gen_compound_mapping(s, cd->factors[0], sub_len)))
return ret;
if (!(s->tmp = av_malloc(len*sizeof(*s->tmp))))
return AVERROR(ENOMEM);
TX_TAB(ff_tx_init_tabs)(len / sub_len);
return 0;
}
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
#define DECL_COMP_FFT(N) \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static void TX_NAME(ff_tx_fft_pfa_##N##xM)(AVTXContext *s, void *_out, \
void *_in, ptrdiff_t stride) \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
{ \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
const int m = s->sub->len; \
const int *in_map = s->map, *out_map = in_map + s->len; \
const int *sub_map = s->sub->map; \
TXComplex *in = _in; \
TXComplex *out = _out; \
TXComplex fft##N##in[N]; \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
\
for (int i = 0; i < m; i++) { \
for (int j = 0; j < N; j++) \
fft##N##in[j] = in[in_map[i*N + j]]; \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
fft##N(s->tmp + sub_map[i], fft##N##in, m); \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
} \
\
for (int i = 0; i < N; i++) \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
s->fn[0](&s->sub[0], s->tmp + m*i, s->tmp + m*i, sizeof(TXComplex)); \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
\
for (int i = 0; i < N*m; i++) \
out[i] = s->tmp[out_map[i]]; \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
} \
\
static const FFTXCodelet TX_NAME(ff_tx_fft_pfa_##N##xM_def) = { \
.name = TX_NAME_STR("fft_pfa_" #N "xM"), \
.function = TX_NAME(ff_tx_fft_pfa_##N##xM), \
.type = TX_TYPE(FFT), \
lavu/tx: add the inplace flag to PFA FFTs They support in-place, because they have to use a temporary buffer.
2022-09-10 02:26:49 +02:00
.flags = AV_TX_UNALIGNED | AV_TX_INPLACE | FF_TX_OUT_OF_PLACE, \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
.factors = { N, TX_FACTOR_ANY }, \
.min_len = N*2, \
.max_len = TX_LEN_UNLIMITED, \
.init = TX_NAME(ff_tx_fft_pfa_init), \
.cpu_flags = FF_TX_CPU_FLAGS_ALL, \
.prio = FF_TX_PRIO_BASE, \
};
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
DECL_COMP_FFT(3)
DECL_COMP_FFT(5)
lavu/tx: add a 7-point FFT and (i)MDCT
2021-04-10 03:51:59 +02:00
DECL_COMP_FFT(7)
lavu/tx: add a 9-point FFT and (i)MDCT
2021-04-10 04:19:06 +02:00
DECL_COMP_FFT(9)
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
DECL_COMP_FFT(15)
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static av_cold int TX_NAME(ff_tx_mdct_naive_init)(AVTXContext *s,
const FFTXCodelet *cd,
uint64_t flags,
FFTXCodeletOptions *opts,
int len, int inv,
const void *scale)
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
{
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
s->scale_d = *((SCALE_TYPE *)scale);
s->scale_f = s->scale_d;
return 0;
}
static void TX_NAME(ff_tx_mdct_naive_fwd)(AVTXContext *s, void *_dst,
void *_src, ptrdiff_t stride)
{
TXSample *src = _src;
TXSample *dst = _dst;
double scale = s->scale_d;
int len = s->len;
const double phase = M_PI/(4.0*len);
lavu/tx: support in-place FFT transforms This commit adds support for in-place FFT transforms. Since our internal transforms were all in-place anyway, this only changes the permutation on the input. Unfortunately, research papers were of no help here. All focused on dry hardware implementations, where permutes are free, or on software implementations where binary bloat is of no concern so storing dozen times the transforms for each permutation and version is not considered bad practice. Still, for a pure C implementation, it's only around 28% slower than the multi-megabyte FFTW3 in unaligned mode. Unlike a closed permutation like with PFA, split-radix FFT bit-reversals contain multiple NOPs, multiple simple swaps, and a few chained swaps, so regular single-loop single-state permute loops were not possible. Instead, we filter out parts of the input indices which are redundant. This allows for a single branch, and with some clever AVX512 asm, could possibly be SIMD'd without refactoring. The inplace_idx array is guaranteed to never be larger than the revtab array, and in practice only requires around log2(len) entries. The power-of-two MDCTs can be done in-place as well. And it's possible to eliminate a copy in the compound MDCTs too, however it'll be slower than doing them out of place, and we'd need to dirty the input array.
2021-02-10 18:58:22 +02:00
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
stride /= sizeof(*dst);
for (int i = 0; i < len; i++) {
double sum = 0.0;
for (int j = 0; j < len*2; j++) {
int a = (2*j + 1 + len) * (2*i + 1);
sum += UNSCALE(src[j]) * cos(a * phase);
}
dst[i*stride] = RESCALE(sum*scale);
}
}
lavu/tx: support in-place FFT transforms This commit adds support for in-place FFT transforms. Since our internal transforms were all in-place anyway, this only changes the permutation on the input. Unfortunately, research papers were of no help here. All focused on dry hardware implementations, where permutes are free, or on software implementations where binary bloat is of no concern so storing dozen times the transforms for each permutation and version is not considered bad practice. Still, for a pure C implementation, it's only around 28% slower than the multi-megabyte FFTW3 in unaligned mode. Unlike a closed permutation like with PFA, split-radix FFT bit-reversals contain multiple NOPs, multiple simple swaps, and a few chained swaps, so regular single-loop single-state permute loops were not possible. Instead, we filter out parts of the input indices which are redundant. This allows for a single branch, and with some clever AVX512 asm, could possibly be SIMD'd without refactoring. The inplace_idx array is guaranteed to never be larger than the revtab array, and in practice only requires around log2(len) entries. The power-of-two MDCTs can be done in-place as well. And it's possible to eliminate a copy in the compound MDCTs too, however it'll be slower than doing them out of place, and we'd need to dirty the input array.
2021-02-10 18:58:22 +02:00
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static void TX_NAME(ff_tx_mdct_naive_inv)(AVTXContext *s, void *_dst,
void *_src, ptrdiff_t stride)
{
TXSample *src = _src;
TXSample *dst = _dst;
double scale = s->scale_d;
int len = s->len >> 1;
int len2 = len*2;
const double phase = M_PI/(4.0*len2);
lavu/tx: support in-place FFT transforms This commit adds support for in-place FFT transforms. Since our internal transforms were all in-place anyway, this only changes the permutation on the input. Unfortunately, research papers were of no help here. All focused on dry hardware implementations, where permutes are free, or on software implementations where binary bloat is of no concern so storing dozen times the transforms for each permutation and version is not considered bad practice. Still, for a pure C implementation, it's only around 28% slower than the multi-megabyte FFTW3 in unaligned mode. Unlike a closed permutation like with PFA, split-radix FFT bit-reversals contain multiple NOPs, multiple simple swaps, and a few chained swaps, so regular single-loop single-state permute loops were not possible. Instead, we filter out parts of the input indices which are redundant. This allows for a single branch, and with some clever AVX512 asm, could possibly be SIMD'd without refactoring. The inplace_idx array is guaranteed to never be larger than the revtab array, and in practice only requires around log2(len) entries. The power-of-two MDCTs can be done in-place as well. And it's possible to eliminate a copy in the compound MDCTs too, however it'll be slower than doing them out of place, and we'd need to dirty the input array.
2021-02-10 18:58:22 +02:00
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
stride /= sizeof(*src);
for (int i = 0; i < len; i++) {
double sum_d = 0.0;
double sum_u = 0.0;
double i_d = phase * (4*len - 2*i - 1);
double i_u = phase * (3*len2 + 2*i + 1);
for (int j = 0; j < len2; j++) {
double a = (2 * j + 1);
double a_d = cos(a * i_d);
double a_u = cos(a * i_u);
double val = UNSCALE(src[j*stride]);
sum_d += a_d * val;
sum_u += a_u * val;
}
dst[i + 0] = RESCALE( sum_d*scale);
dst[i + len] = RESCALE(-sum_u*scale);
lavu/tx: support in-place FFT transforms This commit adds support for in-place FFT transforms. Since our internal transforms were all in-place anyway, this only changes the permutation on the input. Unfortunately, research papers were of no help here. All focused on dry hardware implementations, where permutes are free, or on software implementations where binary bloat is of no concern so storing dozen times the transforms for each permutation and version is not considered bad practice. Still, for a pure C implementation, it's only around 28% slower than the multi-megabyte FFTW3 in unaligned mode. Unlike a closed permutation like with PFA, split-radix FFT bit-reversals contain multiple NOPs, multiple simple swaps, and a few chained swaps, so regular single-loop single-state permute loops were not possible. Instead, we filter out parts of the input indices which are redundant. This allows for a single branch, and with some clever AVX512 asm, could possibly be SIMD'd without refactoring. The inplace_idx array is guaranteed to never be larger than the revtab array, and in practice only requires around log2(len) entries. The power-of-two MDCTs can be done in-place as well. And it's possible to eliminate a copy in the compound MDCTs too, however it'll be slower than doing them out of place, and we'd need to dirty the input array.
2021-02-10 18:58:22 +02:00
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
}
static const FFTXCodelet TX_NAME(ff_tx_mdct_naive_fwd_def) = {
.name = TX_NAME_STR("mdct_naive_fwd"),
.function = TX_NAME(ff_tx_mdct_naive_fwd),
.type = TX_TYPE(MDCT),
.flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE | FF_TX_FORWARD_ONLY,
.factors = { 2, TX_FACTOR_ANY }, /* MDCTs need an even length */
.min_len = 2,
.max_len = TX_LEN_UNLIMITED,
.init = TX_NAME(ff_tx_mdct_naive_init),
.cpu_flags = FF_TX_CPU_FLAGS_ALL,
.prio = FF_TX_PRIO_MIN,
};
static const FFTXCodelet TX_NAME(ff_tx_mdct_naive_inv_def) = {
.name = TX_NAME_STR("mdct_naive_inv"),
.function = TX_NAME(ff_tx_mdct_naive_inv),
.type = TX_TYPE(MDCT),
.flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE | FF_TX_INVERSE_ONLY,
.factors = { 2, TX_FACTOR_ANY },
.min_len = 2,
.max_len = TX_LEN_UNLIMITED,
.init = TX_NAME(ff_tx_mdct_naive_init),
.cpu_flags = FF_TX_CPU_FLAGS_ALL,
.prio = FF_TX_PRIO_MIN,
};
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
static av_cold int TX_NAME(ff_tx_mdct_init)(AVTXContext *s,
const FFTXCodelet *cd,
uint64_t flags,
FFTXCodeletOptions *opts,
int len, int inv,
const void *scale)
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
{
int ret;
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
FFTXCodeletOptions sub_opts = { .invert_lookup = inv };
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
s->scale_d = *((SCALE_TYPE *)scale);
s->scale_f = s->scale_d;
lavu/tx: support in-place FFT transforms This commit adds support for in-place FFT transforms. Since our internal transforms were all in-place anyway, this only changes the permutation on the input. Unfortunately, research papers were of no help here. All focused on dry hardware implementations, where permutes are free, or on software implementations where binary bloat is of no concern so storing dozen times the transforms for each permutation and version is not considered bad practice. Still, for a pure C implementation, it's only around 28% slower than the multi-megabyte FFTW3 in unaligned mode. Unlike a closed permutation like with PFA, split-radix FFT bit-reversals contain multiple NOPs, multiple simple swaps, and a few chained swaps, so regular single-loop single-state permute loops were not possible. Instead, we filter out parts of the input indices which are redundant. This allows for a single branch, and with some clever AVX512 asm, could possibly be SIMD'd without refactoring. The inplace_idx array is guaranteed to never be larger than the revtab array, and in practice only requires around log2(len) entries. The power-of-two MDCTs can be done in-place as well. And it's possible to eliminate a copy in the compound MDCTs too, however it'll be slower than doing them out of place, and we'd need to dirty the input array.
2021-02-10 18:58:22 +02:00
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
flags &= ~FF_TX_OUT_OF_PLACE; /* We want the subtransform to be */
flags |= AV_TX_INPLACE; /* in-place */
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
flags |= FF_TX_PRESHUFFLE; /* First try with an in-place transform */
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
if ((ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts, len >> 1,
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
inv, scale))) {
flags &= ~FF_TX_PRESHUFFLE; /* Now try with a generic FFT */
if ((ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts, len >> 1,
inv, scale)))
return ret;
}
lavu/tx: do not steal lookup tables of subcontexts in the iMDCT As it happens, some still need their contexts.
2022-09-22 08:25:52 +02:00
s->map = av_malloc((len >> 1)*sizeof(*s->map));
if (!s->map)
return AVERROR(ENOMEM);
/* If we need to preshuffle copy the map from the subcontext */
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
if (s->sub[0].flags & FF_TX_PRESHUFFLE) {
lavu/tx: do not steal lookup tables of subcontexts in the iMDCT As it happens, some still need their contexts.
2022-09-22 08:25:52 +02:00
memcpy(s->map, s->sub->map, (len >> 1)*sizeof(*s->map));
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
} else {
for (int i = 0; i < len >> 1; i++)
s->map[i] = i;
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
if ((ret = TX_TAB(ff_tx_mdct_gen_exp)(s, inv ? s->map : NULL)))
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
return ret;
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
/* Saves a multiply in a hot path. */
if (inv)
for (int i = 0; i < (s->len >> 1); i++)
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
s->map[i] <<= 1;
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
return 0;
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
}
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
static void TX_NAME(ff_tx_mdct_fwd)(AVTXContext *s, void *_dst, void *_src,
ptrdiff_t stride)
lavu: support arbitrary-point FFTs and all even (i)MDCT transforms This patch adds support for arbitrary-point FFTs and all even MDCT transforms. Odd MDCTs are not supported yet as they're based on the DCT-II and DCT-III and they're very niche. With this we can now write tests.
2021-01-12 09:11:47 +02:00
{
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TXSample *src = _src, *dst = _dst;
TXComplex *exp = s->exp, tmp, *z = _dst;
const int len2 = s->len >> 1;
const int len4 = s->len >> 2;
const int len3 = len2 * 3;
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
const int *sub_map = s->map;
lavu: support arbitrary-point FFTs and all even (i)MDCT transforms This patch adds support for arbitrary-point FFTs and all even MDCT transforms. Odd MDCTs are not supported yet as they're based on the DCT-II and DCT-III and they're very niche. With this we can now write tests.
2021-01-12 09:11:47 +02:00
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
stride /= sizeof(*dst);
for (int i = 0; i < len2; i++) { /* Folding and pre-reindexing */
const int k = 2*i;
const int idx = sub_map[i];
if (k < len2) {
tmp.re = FOLD(-src[ len2 + k], src[1*len2 - 1 - k]);
tmp.im = FOLD(-src[ len3 + k], -src[1*len3 - 1 - k]);
} else {
tmp.re = FOLD(-src[ len2 + k], -src[5*len2 - 1 - k]);
tmp.im = FOLD( src[-len2 + k], -src[1*len3 - 1 - k]);
lavu: support arbitrary-point FFTs and all even (i)MDCT transforms This patch adds support for arbitrary-point FFTs and all even MDCT transforms. Odd MDCTs are not supported yet as they're based on the DCT-II and DCT-III and they're very niche. With this we can now write tests.
2021-01-12 09:11:47 +02:00
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
CMUL(z[idx].im, z[idx].re, tmp.re, tmp.im, exp[i].re, exp[i].im);
}
s->fn[0](&s->sub[0], z, z, sizeof(TXComplex));
for (int i = 0; i < len4; i++) {
const int i0 = len4 + i, i1 = len4 - i - 1;
TXComplex src1 = { z[i1].re, z[i1].im };
TXComplex src0 = { z[i0].re, z[i0].im };
CMUL(dst[2*i1*stride + stride], dst[2*i0*stride], src0.re, src0.im,
exp[i0].im, exp[i0].re);
CMUL(dst[2*i0*stride + stride], dst[2*i1*stride], src1.re, src1.im,
exp[i1].im, exp[i1].re);
lavu: support arbitrary-point FFTs and all even (i)MDCT transforms This patch adds support for arbitrary-point FFTs and all even MDCT transforms. Odd MDCTs are not supported yet as they're based on the DCT-II and DCT-III and they're very niche. With this we can now write tests.
2021-01-12 09:11:47 +02:00
}
}
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
static void TX_NAME(ff_tx_mdct_inv)(AVTXContext *s, void *_dst, void *_src,
ptrdiff_t stride)
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
{
TXComplex *z = _dst, *exp = s->exp;
const TXSample *src = _src, *in1, *in2;
const int len2 = s->len >> 1;
const int len4 = s->len >> 2;
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
const int *sub_map = s->map;
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
stride /= sizeof(*src);
in1 = src;
in2 = src + ((len2*2) - 1) * stride;
for (int i = 0; i < len2; i++) {
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
int k = sub_map[i];
TXComplex tmp = { in2[-k*stride], in1[k*stride] };
CMUL3(z[i], tmp, exp[i]);
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
}
s->fn[0](&s->sub[0], z, z, sizeof(TXComplex));
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
exp += len2;
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
for (int i = 0; i < len4; i++) {
const int i0 = len4 + i, i1 = len4 - i - 1;
TXComplex src1 = { z[i1].im, z[i1].re };
TXComplex src0 = { z[i0].im, z[i0].re };
CMUL(z[i1].re, z[i0].im, src1.re, src1.im, exp[i1].im, exp[i1].re);
CMUL(z[i0].re, z[i1].im, src0.re, src0.im, exp[i0].im, exp[i0].re);
}
}
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
static const FFTXCodelet TX_NAME(ff_tx_mdct_fwd_def) = {
.name = TX_NAME_STR("mdct_fwd"),
.function = TX_NAME(ff_tx_mdct_fwd),
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
.type = TX_TYPE(MDCT),
.flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE | FF_TX_FORWARD_ONLY,
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
.factors = { 2, TX_FACTOR_ANY },
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
.min_len = 2,
.max_len = TX_LEN_UNLIMITED,
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
.init = TX_NAME(ff_tx_mdct_init),
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
.cpu_flags = FF_TX_CPU_FLAGS_ALL,
.prio = FF_TX_PRIO_BASE,
};
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
static const FFTXCodelet TX_NAME(ff_tx_mdct_inv_def) = {
.name = TX_NAME_STR("mdct_inv"),
.function = TX_NAME(ff_tx_mdct_inv),
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
.type = TX_TYPE(MDCT),
.flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE | FF_TX_INVERSE_ONLY,
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
.factors = { 2, TX_FACTOR_ANY },
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
.min_len = 2,
.max_len = TX_LEN_UNLIMITED,
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
.init = TX_NAME(ff_tx_mdct_init),
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
.cpu_flags = FF_TX_CPU_FLAGS_ALL,
.prio = FF_TX_PRIO_BASE,
};
static av_cold int TX_NAME(ff_tx_mdct_inv_full_init)(AVTXContext *s,
const FFTXCodelet *cd,
uint64_t flags,
FFTXCodeletOptions *opts,
int len, int inv,
const void *scale)
{
int ret;
s->scale_d = *((SCALE_TYPE *)scale);
s->scale_f = s->scale_d;
flags &= ~AV_TX_FULL_IMDCT;
if ((ret = ff_tx_init_subtx(s, TX_TYPE(MDCT), flags, NULL, len, 1, scale)))
return ret;
return 0;
}
static void TX_NAME(ff_tx_mdct_inv_full)(AVTXContext *s, void *_dst,
void *_src, ptrdiff_t stride)
{
int len = s->len << 1;
int len2 = len >> 1;
int len4 = len >> 2;
TXSample *dst = _dst;
s->fn[0](&s->sub[0], dst + len4, _src, stride);
stride /= sizeof(*dst);
for (int i = 0; i < len4; i++) {
dst[ i*stride] = -dst[(len2 - i - 1)*stride];
dst[(len - i - 1)*stride] = dst[(len2 + i + 0)*stride];
}
}
static const FFTXCodelet TX_NAME(ff_tx_mdct_inv_full_def) = {
.name = TX_NAME_STR("mdct_inv_full"),
.function = TX_NAME(ff_tx_mdct_inv_full),
.type = TX_TYPE(MDCT),
.flags = AV_TX_UNALIGNED | AV_TX_INPLACE |
FF_TX_OUT_OF_PLACE | AV_TX_FULL_IMDCT,
.factors = { 2, TX_FACTOR_ANY },
.min_len = 2,
.max_len = TX_LEN_UNLIMITED,
.init = TX_NAME(ff_tx_mdct_inv_full_init),
.cpu_flags = FF_TX_CPU_FLAGS_ALL,
.prio = FF_TX_PRIO_BASE,
};
static av_cold int TX_NAME(ff_tx_mdct_pfa_init)(AVTXContext *s,
const FFTXCodelet *cd,
uint64_t flags,
FFTXCodeletOptions *opts,
int len, int inv,
const void *scale)
{
int ret, sub_len;
FFTXCodeletOptions sub_opts = { .invert_lookup = 0 };
len >>= 1;
sub_len = len / cd->factors[0];
s->scale_d = *((SCALE_TYPE *)scale);
s->scale_f = s->scale_d;
flags &= ~FF_TX_OUT_OF_PLACE; /* We want the subtransform to be */
flags |= AV_TX_INPLACE; /* in-place */
flags |= FF_TX_PRESHUFFLE; /* This function handles the permute step */
if ((ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts,
sub_len, inv, scale)))
return ret;
if ((ret = ff_tx_gen_compound_mapping(s, cd->factors[0], sub_len)))
return ret;
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
if ((ret = TX_TAB(ff_tx_mdct_gen_exp)(s, inv ? s->map : NULL)))
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
return ret;
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
/* Saves multiplies in loops. */
for (int i = 0; i < len; i++)
s->map[i] <<= 1;
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
if (!(s->tmp = av_malloc(len*sizeof(*s->tmp))))
return AVERROR(ENOMEM);
TX_TAB(ff_tx_init_tabs)(len / sub_len);
return 0;
}
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
#define DECL_COMP_IMDCT(N) \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static void TX_NAME(ff_tx_mdct_pfa_##N##xM_inv)(AVTXContext *s, void *_dst, \
void *_src, ptrdiff_t stride) \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
{ \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TXComplex fft##N##in[N]; \
TXComplex *z = _dst, *exp = s->exp; \
const TXSample *src = _src, *in1, *in2; \
const int len4 = s->len >> 2; \
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
const int len2 = s->len >> 1; \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
const int m = s->sub->len; \
const int *in_map = s->map, *out_map = in_map + N*m; \
const int *sub_map = s->sub->map; \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
\
stride /= sizeof(*src); /* To convert it from bytes */ \
in1 = src; \
in2 = src + ((N*m*2) - 1) * stride; \
\
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
for (int i = 0; i < len2; i += N) { \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
for (int j = 0; j < N; j++) { \
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
const int k = in_map[j]; \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TXComplex tmp = { in2[-k*stride], in1[k*stride] }; \
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
CMUL3(fft##N##in[j], tmp, exp[j]); \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
} \
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
fft##N(s->tmp + *(sub_map++), fft##N##in, m); \
exp += N; \
in_map += N; \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
} \
\
for (int i = 0; i < N; i++) \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
s->fn[0](&s->sub[0], s->tmp + m*i, s->tmp + m*i, sizeof(TXComplex)); \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
\
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
for (int i = 0; i < len4; i++) { \
const int i0 = len4 + i, i1 = len4 - i - 1; \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
const int s0 = out_map[i0], s1 = out_map[i1]; \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TXComplex src1 = { s->tmp[s1].im, s->tmp[s1].re }; \
TXComplex src0 = { s->tmp[s0].im, s->tmp[s0].re }; \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
\
CMUL(z[i1].re, z[i0].im, src1.re, src1.im, exp[i1].im, exp[i1].re); \
CMUL(z[i0].re, z[i1].im, src0.re, src0.im, exp[i0].im, exp[i0].re); \
} \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
} \
\
static const FFTXCodelet TX_NAME(ff_tx_mdct_pfa_##N##xM_inv_def) = { \
.name = TX_NAME_STR("mdct_pfa_" #N "xM_inv"), \
.function = TX_NAME(ff_tx_mdct_pfa_##N##xM_inv), \
.type = TX_TYPE(MDCT), \
.flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE | FF_TX_INVERSE_ONLY, \
.factors = { N, TX_FACTOR_ANY }, \
.min_len = N*2, \
.max_len = TX_LEN_UNLIMITED, \
.init = TX_NAME(ff_tx_mdct_pfa_init), \
.cpu_flags = FF_TX_CPU_FLAGS_ALL, \
.prio = FF_TX_PRIO_BASE, \
};
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
DECL_COMP_IMDCT(3)
DECL_COMP_IMDCT(5)
lavu/tx: add a 7-point FFT and (i)MDCT
2021-04-10 03:51:59 +02:00
DECL_COMP_IMDCT(7)
lavu/tx: add a 9-point FFT and (i)MDCT
2021-04-10 04:19:06 +02:00
DECL_COMP_IMDCT(9)
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
DECL_COMP_IMDCT(15)
#define DECL_COMP_MDCT(N) \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
static void TX_NAME(ff_tx_mdct_pfa_##N##xM_fwd)(AVTXContext *s, void *_dst, \
void *_src, ptrdiff_t stride) \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
{ \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TXComplex fft##N##in[N]; \
TXSample *src = _src, *dst = _dst; \
TXComplex *exp = s->exp, tmp; \
const int m = s->sub->len; \
const int len4 = N*m; \
const int len3 = len4 * 3; \
const int len8 = s->len >> 2; \
const int *in_map = s->map, *out_map = in_map + N*m; \
const int *sub_map = s->sub->map; \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
\
stride /= sizeof(*dst); \
\
for (int i = 0; i < m; i++) { /* Folding and pre-reindexing */ \
for (int j = 0; j < N; j++) { \
const int k = in_map[i*N + j]; \
if (k < len4) { \
lavu/tx: implement 32 bit fixed point FFT and MDCT Required minimal changes to the code so made sense to implement. FFT and MDCT tested, the output of both was properly rounded. Fun fact: the non-power-of-two fixed-point FFT and MDCT are the fastest ever non-power-of-two fixed-point FFT and MDCT written. This can replace the power of two integer MDCTs in aac and ac3 if the MIPS optimizations are ported across. Unfortunately the ac3 encoder uses a 16-bit fixed point forward transform, unlike the encoder which uses a 32bit inverse transform, so some modifications might be required there. The 3-point FFT is somewhat less accurate than it otherwise could be, having minor rounding errors with bigger transforms. However, this could be improved later, and the way its currently written is the way one would write assembly for it. Similar rounding errors can also be found throughout the power of two FFTs as well, though those are more difficult to correct. Despite this, the integer transforms are more than accurate enough.
2020-02-09 01:13:28 +02:00
tmp.re = FOLD(-src[ len4 + k], src[1*len4 - 1 - k]); \
tmp.im = FOLD(-src[ len3 + k], -src[1*len3 - 1 - k]); \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
} else { \
lavu/tx: implement 32 bit fixed point FFT and MDCT Required minimal changes to the code so made sense to implement. FFT and MDCT tested, the output of both was properly rounded. Fun fact: the non-power-of-two fixed-point FFT and MDCT are the fastest ever non-power-of-two fixed-point FFT and MDCT written. This can replace the power of two integer MDCTs in aac and ac3 if the MIPS optimizations are ported across. Unfortunately the ac3 encoder uses a 16-bit fixed point forward transform, unlike the encoder which uses a 32bit inverse transform, so some modifications might be required there. The 3-point FFT is somewhat less accurate than it otherwise could be, having minor rounding errors with bigger transforms. However, this could be improved later, and the way its currently written is the way one would write assembly for it. Similar rounding errors can also be found throughout the power of two FFTs as well, though those are more difficult to correct. Despite this, the integer transforms are more than accurate enough.
2020-02-09 01:13:28 +02:00
tmp.re = FOLD(-src[ len4 + k], -src[5*len4 - 1 - k]); \
tmp.im = FOLD( src[-len4 + k], -src[1*len3 - 1 - k]); \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
} \
CMUL(fft##N##in[j].im, fft##N##in[j].re, tmp.re, tmp.im, \
exp[k >> 1].re, exp[k >> 1].im); \
} \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
fft##N(s->tmp + sub_map[i], fft##N##in, m); \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
} \
\
for (int i = 0; i < N; i++) \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
s->fn[0](&s->sub[0], s->tmp + m*i, s->tmp + m*i, sizeof(TXComplex)); \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
\
for (int i = 0; i < len8; i++) { \
const int i0 = len8 + i, i1 = len8 - i - 1; \
const int s0 = out_map[i0], s1 = out_map[i1]; \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
TXComplex src1 = { s->tmp[s1].re, s->tmp[s1].im }; \
TXComplex src0 = { s->tmp[s0].re, s->tmp[s0].im }; \
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
\
CMUL(dst[2*i1*stride + stride], dst[2*i0*stride], src0.re, src0.im, \
exp[i0].im, exp[i0].re); \
CMUL(dst[2*i0*stride + stride], dst[2*i1*stride], src1.re, src1.im, \
exp[i1].im, exp[i1].re); \
} \
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
} \
\
static const FFTXCodelet TX_NAME(ff_tx_mdct_pfa_##N##xM_fwd_def) = { \
.name = TX_NAME_STR("mdct_pfa_" #N "xM_fwd"), \
.function = TX_NAME(ff_tx_mdct_pfa_##N##xM_fwd), \
.type = TX_TYPE(MDCT), \
.flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE | FF_TX_FORWARD_ONLY, \
.factors = { N, TX_FACTOR_ANY }, \
.min_len = N*2, \
.max_len = TX_LEN_UNLIMITED, \
.init = TX_NAME(ff_tx_mdct_pfa_init), \
.cpu_flags = FF_TX_CPU_FLAGS_ALL, \
.prio = FF_TX_PRIO_BASE, \
};
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
DECL_COMP_MDCT(3)
DECL_COMP_MDCT(5)
lavu/tx: add a 7-point FFT and (i)MDCT
2021-04-10 03:51:59 +02:00
DECL_COMP_MDCT(7)
lavu/tx: add a 9-point FFT and (i)MDCT
2021-04-10 04:19:06 +02:00
DECL_COMP_MDCT(9)
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
DECL_COMP_MDCT(15)
lavu/tx: add an RDFT implementation RDFTs are full of conventions that vary between implementations. What I've gone for here is what's most common between both fftw, avcodec's rdft and what we use, the equivalent of which is DFT_R2C for forward and IDFT_C2R for inverse. The other 2 conventions (IDFT_R2C and DFT_C2R) were not used at all in our code, and their names are also not appropriate. If there's a use for either, we can easily add a flag which would just flip the sign on one exptab. For some unknown reason, possibly to allow reusing FFT's exp tables, av_rdft's C2R output is 0.5x lower than what it should be to ensure a proper back-and-forth conversion. This code outputs its real samples at the correct level, which matches FFTW's level, and allows the user to change the level and insert arbitrary multiplies for free by setting the scale option.
2022-01-21 08:50:53 +02:00
static av_cold int TX_NAME(ff_tx_rdft_init)(AVTXContext *s,
const FFTXCodelet *cd,
uint64_t flags,
FFTXCodeletOptions *opts,
int len, int inv,
const void *scale)
{
int ret;
double f, m;
TXSample *tab;
s->scale_d = *((SCALE_TYPE *)scale);
s->scale_f = s->scale_d;
if ((ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, NULL, len >> 1, inv, scale)))
return ret;
if (!(s->exp = av_mallocz((8 + (len >> 2) - 1)*sizeof(*s->exp))))
return AVERROR(ENOMEM);
tab = (TXSample *)s->exp;
f = 2*M_PI/len;
m = (inv ? 2*s->scale_d : s->scale_d);
*tab++ = RESCALE((inv ? 0.5 : 1.0) * m);
*tab++ = RESCALE(inv ? 0.5*m : 1.0);
*tab++ = RESCALE( m);
*tab++ = RESCALE(-m);
*tab++ = RESCALE( (0.5 - 0.0) * m);
*tab++ = RESCALE( (0.0 - 0.5) * m);
*tab++ = RESCALE( (0.5 - inv) * m);
*tab++ = RESCALE(-(0.5 - inv) * m);
for (int i = 0; i < len >> 2; i++)
*tab++ = RESCALE(cos(i*f));
for (int i = len >> 2; i >= 0; i--)
*tab++ = RESCALE(cos(i*f) * (inv ? +1.0 : -1.0));
return 0;
}
#define DECL_RDFT(name, inv) \
static void TX_NAME(ff_tx_rdft_ ##name)(AVTXContext *s, void *_dst, \
void *_src, ptrdiff_t stride) \
{ \
const int len2 = s->len >> 1; \
const int len4 = s->len >> 2; \
const TXSample *fact = (void *)s->exp; \
const TXSample *tcos = fact + 8; \
const TXSample *tsin = tcos + len4; \
TXComplex *data = inv ? _src : _dst; \
TXComplex t[3]; \
\
if (!inv) \
s->fn[0](&s->sub[0], data, _src, sizeof(TXComplex)); \
else \
data[0].im = data[len2].re; \
\
/* The DC value's both components are real, but we need to change them \
* into complex values. Also, the middle of the array is special-cased. \
* These operations can be done before or after the loop. */ \
t[0].re = data[0].re; \
data[0].re = t[0].re + data[0].im; \
data[0].im = t[0].re - data[0].im; \
data[ 0].re = MULT(fact[0], data[ 0].re); \
data[ 0].im = MULT(fact[1], data[ 0].im); \
data[len4].re = MULT(fact[2], data[len4].re); \
data[len4].im = MULT(fact[3], data[len4].im); \
\
for (int i = 1; i < len4; i++) { \
/* Separate even and odd FFTs */ \
t[0].re = MULT(fact[4], (data[i].re + data[len2 - i].re)); \
t[0].im = MULT(fact[5], (data[i].im - data[len2 - i].im)); \
t[1].re = MULT(fact[6], (data[i].im + data[len2 - i].im)); \
t[1].im = MULT(fact[7], (data[i].re - data[len2 - i].re)); \
\
/* Apply twiddle factors to the odd FFT and add to the even FFT */ \
CMUL(t[2].re, t[2].im, t[1].re, t[1].im, tcos[i], tsin[i]); \
\
data[ i].re = t[0].re + t[2].re; \
data[ i].im = t[2].im - t[0].im; \
data[len2 - i].re = t[0].re - t[2].re; \
data[len2 - i].im = t[2].im + t[0].im; \
} \
\
if (inv) { \
s->fn[0](&s->sub[0], _dst, data, sizeof(TXComplex)); \
} else { \
/* Move [0].im to the last position, as convention requires */ \
data[len2].re = data[0].im; \
data[ 0].im = 0; \
} \
}
DECL_RDFT(r2c, 0)
DECL_RDFT(c2r, 1)
static const FFTXCodelet TX_NAME(ff_tx_rdft_r2c_def) = {
.name = TX_NAME_STR("rdft_r2c"),
.function = TX_NAME(ff_tx_rdft_r2c),
.type = TX_TYPE(RDFT),
.flags = AV_TX_UNALIGNED | AV_TX_INPLACE |
FF_TX_OUT_OF_PLACE | FF_TX_FORWARD_ONLY,
.factors = { 2, TX_FACTOR_ANY },
.min_len = 2,
.max_len = TX_LEN_UNLIMITED,
.init = TX_NAME(ff_tx_rdft_init),
.cpu_flags = FF_TX_CPU_FLAGS_ALL,
.prio = FF_TX_PRIO_BASE,
};
static const FFTXCodelet TX_NAME(ff_tx_rdft_c2r_def) = {
.name = TX_NAME_STR("rdft_c2r"),
.function = TX_NAME(ff_tx_rdft_c2r),
.type = TX_TYPE(RDFT),
.flags = AV_TX_UNALIGNED | AV_TX_INPLACE |
FF_TX_OUT_OF_PLACE | FF_TX_INVERSE_ONLY,
.factors = { 2, TX_FACTOR_ANY },
.min_len = 2,
.max_len = TX_LEN_UNLIMITED,
.init = TX_NAME(ff_tx_rdft_init),
.cpu_flags = FF_TX_CPU_FLAGS_ALL,
.prio = FF_TX_PRIO_BASE,
};
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
int TX_TAB(ff_tx_mdct_gen_exp)(AVTXContext *s, int *pre_tab)
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
{
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
int off = 0;
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
int len4 = s->len >> 1;
double scale = s->scale_d;
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
const double theta = (scale < 0 ? len4 : 0) + 1.0/8.0;
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
size_t alloc = pre_tab ? 2*len4 : len4;
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
if (!(s->exp = av_malloc_array(alloc, sizeof(*s->exp))))
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
return AVERROR(ENOMEM);
scale = sqrt(fabs(scale));
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
if (pre_tab)
off = len4;
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
for (int i = 0; i < len4; i++) {
const double alpha = M_PI_2 * (i + theta) / len4;
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
s->exp[off + i] = (TXComplex){ RESCALE(cos(alpha) * scale),
RESCALE(sin(alpha) * scale) };
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
}
lavu/tx: optimize and simplify inverse MDCTs Convert the input from a scatter to a gather instead, which is faster and better for SIMD. Also, add a pre-shuffled exptab version to avoid gathering there at all. This doubles the exptab size, but the speedup makes it worth it. In SIMD, the exptab will likely be purged to a higher cache anyway because of the FFT in the middle, and the amount of loads stays identical. For a 960-point inverse MDCT, the speedup is 10%. This makes it possible to write sane and fast SIMD versions of inverse MDCTs.
2022-08-16 01:11:40 +02:00
if (pre_tab)
for (int i = 0; i < len4; i++)
s->exp[i] = s->exp[len4 + pre_tab[i]];
lavu/tx: add support for double precision FFT and MDCT Simply moves and templates the actual transforms to support an additional data type. Unlike the float version, which is equal or better than libfftw3f, double precision output is bit identical with libfftw3.
2019-07-27 19:54:20 +02:00
return 0;
}
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
const FFTXCodelet * const TX_NAME(ff_tx_codelet_list)[] = {
/* Split-Radix codelets */
&TX_NAME(ff_tx_fft2_ns_def),
&TX_NAME(ff_tx_fft4_ns_def),
&TX_NAME(ff_tx_fft8_ns_def),
&TX_NAME(ff_tx_fft16_ns_def),
&TX_NAME(ff_tx_fft32_ns_def),
&TX_NAME(ff_tx_fft64_ns_def),
&TX_NAME(ff_tx_fft128_ns_def),
&TX_NAME(ff_tx_fft256_ns_def),
&TX_NAME(ff_tx_fft512_ns_def),
&TX_NAME(ff_tx_fft1024_ns_def),
&TX_NAME(ff_tx_fft2048_ns_def),
&TX_NAME(ff_tx_fft4096_ns_def),
&TX_NAME(ff_tx_fft8192_ns_def),
&TX_NAME(ff_tx_fft16384_ns_def),
&TX_NAME(ff_tx_fft32768_ns_def),
&TX_NAME(ff_tx_fft65536_ns_def),
&TX_NAME(ff_tx_fft131072_ns_def),
/* Standalone transforms */
&TX_NAME(ff_tx_fft_sr_def),
&TX_NAME(ff_tx_fft_sr_inplace_def),
&TX_NAME(ff_tx_fft_pfa_3xM_def),
&TX_NAME(ff_tx_fft_pfa_5xM_def),
&TX_NAME(ff_tx_fft_pfa_7xM_def),
&TX_NAME(ff_tx_fft_pfa_9xM_def),
&TX_NAME(ff_tx_fft_pfa_15xM_def),
&TX_NAME(ff_tx_fft_naive_def),
lavu/tx: generalize MDCTs The same code can perform any-length MDCTs with minimal changes.
2022-09-10 02:28:10 +02:00
&TX_NAME(ff_tx_mdct_fwd_def),
&TX_NAME(ff_tx_mdct_inv_def),
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
&TX_NAME(ff_tx_mdct_pfa_3xM_fwd_def),
&TX_NAME(ff_tx_mdct_pfa_5xM_fwd_def),
&TX_NAME(ff_tx_mdct_pfa_7xM_fwd_def),
&TX_NAME(ff_tx_mdct_pfa_9xM_fwd_def),
&TX_NAME(ff_tx_mdct_pfa_15xM_fwd_def),
&TX_NAME(ff_tx_mdct_pfa_3xM_inv_def),
&TX_NAME(ff_tx_mdct_pfa_5xM_inv_def),
&TX_NAME(ff_tx_mdct_pfa_7xM_inv_def),
&TX_NAME(ff_tx_mdct_pfa_9xM_inv_def),
&TX_NAME(ff_tx_mdct_pfa_15xM_inv_def),
&TX_NAME(ff_tx_mdct_naive_fwd_def),
&TX_NAME(ff_tx_mdct_naive_inv_def),
&TX_NAME(ff_tx_mdct_inv_full_def),
lavu/tx: add an RDFT implementation RDFTs are full of conventions that vary between implementations. What I've gone for here is what's most common between both fftw, avcodec's rdft and what we use, the equivalent of which is DFT_R2C for forward and IDFT_C2R for inverse. The other 2 conventions (IDFT_R2C and DFT_C2R) were not used at all in our code, and their names are also not appropriate. If there's a use for either, we can easily add a flag which would just flip the sign on one exptab. For some unknown reason, possibly to allow reusing FFT's exp tables, av_rdft's C2R output is 0.5x lower than what it should be to ensure a proper back-and-forth conversion. This code outputs its real samples at the correct level, which matches FFTW's level, and allows the user to change the level and insert arbitrary multiplies for free by setting the scale option.
2022-01-21 08:50:53 +02:00
&TX_NAME(ff_tx_rdft_r2c_def),
&TX_NAME(ff_tx_rdft_c2r_def),
lavu: support arbitrary-point FFTs and all even (i)MDCT transforms This patch adds support for arbitrary-point FFTs and all even MDCT transforms. Odd MDCTs are not supported yet as they're based on the DCT-II and DCT-III and they're very niche. With this we can now write tests.
2021-01-12 09:11:47 +02:00
lavu/tx: rewrite internal code as a tree-based codelet constructor This commit rewrites the internal transform code into a constructor that stitches transforms (codelets). This allows for transforms to reuse arbitrary parts of other transforms, and allows transforms to be stacked onto one another (such as a full iMDCT using a half-iMDCT which in turn uses an FFT). It also permits for each step to be individually replaced by assembly or a custom implementation (such as an ASIC).
2022-01-20 08:14:46 +02:00
NULL,
};
Reference in New Issue Copy Permalink