2022-08-09 23:53:10 +00:00
|
|
|
/*
|
|
|
|
* This file is part of FFmpeg.
|
|
|
|
*
|
|
|
|
* FFmpeg is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
|
|
|
* version 2.1 of the License, or (at your option) any later version.
|
|
|
|
*
|
|
|
|
* FFmpeg is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* Lesser General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU Lesser General Public
|
|
|
|
* License along with FFmpeg; if not, write to the Free Software
|
|
|
|
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "libavutil/float2half.h"
|
|
|
|
|
|
|
|
void ff_init_float2half_tables(Float2HalfTables *t)
|
|
|
|
{
|
avutil/half2float: use native _Float16 if available
_Float16 support was available on arm/aarch64 for a while, and with gcc
12 was enabled on x86 as long as SSE2 is supported.
If the target arch supports f16c, gcc emits fairly efficient assembly,
taking advantage of it. This is the case on x86-64-v3 or higher.
Same goes on arm, which has native float16 support.
On x86, without f16c, it emulates it in software using sse2 instructions.
This has shown to perform rather poorly:
_Float16 full SSE2 emulation:
frame=50074 fps=848 q=-0.0 size=N/A time=00:33:22.96 bitrate=N/A speed=33.9x
_Float16 f16c accelerated (Zen2, --cpu=znver2):
frame=50636 fps=1965 q=-0.0 Lsize=N/A time=00:33:45.40 bitrate=N/A speed=78.6x
classic half2float full software implementation:
frame=49926 fps=1605 q=-0.0 Lsize=N/A time=00:33:17.00 bitrate=N/A speed=64.2x
Hence an additional check was introduced, that only enables use of
_Float16 on x86 if f16c is being utilized.
On aarch64, a similar uplift in performance is seen:
RPi4 half2float full software implementation:
frame= 6088 fps=126 q=-0.0 Lsize=N/A time=00:04:03.48 bitrate=N/A speed=5.06x
RPi4 _Float16:
frame= 6103 fps=158 q=-0.0 Lsize=N/A time=00:04:04.08 bitrate=N/A speed=6.32x
Since arm/aarch64 always natively support 16 bit floats, it can always
be considered fast there.
I'm not aware of any additional platforms that currently support
_Float16. And if there are, they should be considered non-fast until
proven fast.
2022-08-10 00:23:26 +00:00
|
|
|
#if !HAVE_FAST_FLOAT16
|
2022-08-09 23:53:10 +00:00
|
|
|
for (int i = 0; i < 256; i++) {
|
|
|
|
int e = i - 127;
|
|
|
|
|
|
|
|
if (e < -24) { // Very small numbers map to zero
|
|
|
|
t->basetable[i|0x000] = 0x0000;
|
|
|
|
t->basetable[i|0x100] = 0x8000;
|
|
|
|
t->shifttable[i|0x000] = 24;
|
|
|
|
t->shifttable[i|0x100] = 24;
|
|
|
|
} else if (e < -14) { // Small numbers map to denorms
|
|
|
|
t->basetable[i|0x000] = (0x0400>>(-e-14));
|
|
|
|
t->basetable[i|0x100] = (0x0400>>(-e-14)) | 0x8000;
|
|
|
|
t->shifttable[i|0x000] = -e-1;
|
|
|
|
t->shifttable[i|0x100] = -e-1;
|
|
|
|
} else if (e <= 15) { // Normal numbers just lose precision
|
|
|
|
t->basetable[i|0x000] = ((e + 15) << 10);
|
|
|
|
t->basetable[i|0x100] = ((e + 15) << 10) | 0x8000;
|
|
|
|
t->shifttable[i|0x000] = 13;
|
|
|
|
t->shifttable[i|0x100] = 13;
|
|
|
|
} else if (e < 128) { // Large numbers map to Infinity
|
|
|
|
t->basetable[i|0x000] = 0x7C00;
|
|
|
|
t->basetable[i|0x100] = 0xFC00;
|
|
|
|
t->shifttable[i|0x000] = 24;
|
|
|
|
t->shifttable[i|0x100] = 24;
|
|
|
|
} else { // Infinity and NaN's stay Infinity and NaN's
|
|
|
|
t->basetable[i|0x000] = 0x7C00;
|
|
|
|
t->basetable[i|0x100] = 0xFC00;
|
|
|
|
t->shifttable[i|0x000] = 13;
|
|
|
|
t->shifttable[i|0x100] = 13;
|
|
|
|
}
|
|
|
|
}
|
avutil/half2float: use native _Float16 if available
_Float16 support was available on arm/aarch64 for a while, and with gcc
12 was enabled on x86 as long as SSE2 is supported.
If the target arch supports f16c, gcc emits fairly efficient assembly,
taking advantage of it. This is the case on x86-64-v3 or higher.
Same goes on arm, which has native float16 support.
On x86, without f16c, it emulates it in software using sse2 instructions.
This has shown to perform rather poorly:
_Float16 full SSE2 emulation:
frame=50074 fps=848 q=-0.0 size=N/A time=00:33:22.96 bitrate=N/A speed=33.9x
_Float16 f16c accelerated (Zen2, --cpu=znver2):
frame=50636 fps=1965 q=-0.0 Lsize=N/A time=00:33:45.40 bitrate=N/A speed=78.6x
classic half2float full software implementation:
frame=49926 fps=1605 q=-0.0 Lsize=N/A time=00:33:17.00 bitrate=N/A speed=64.2x
Hence an additional check was introduced, that only enables use of
_Float16 on x86 if f16c is being utilized.
On aarch64, a similar uplift in performance is seen:
RPi4 half2float full software implementation:
frame= 6088 fps=126 q=-0.0 Lsize=N/A time=00:04:03.48 bitrate=N/A speed=5.06x
RPi4 _Float16:
frame= 6103 fps=158 q=-0.0 Lsize=N/A time=00:04:04.08 bitrate=N/A speed=6.32x
Since arm/aarch64 always natively support 16 bit floats, it can always
be considered fast there.
I'm not aware of any additional platforms that currently support
_Float16. And if there are, they should be considered non-fast until
proven fast.
2022-08-10 00:23:26 +00:00
|
|
|
#endif
|
2022-08-09 23:53:10 +00:00
|
|
|
}
|