0a949aacae
checkasm/lpc: use fixed length to bench apply_welch_window
...
Signed-off-by: James Almer <jamrial@gmail.com>
2024-05-31 17:06:08 -03:00
16132a810d
checkasm/lpc: test compute_autocorr
...
Signed-off-by: James Almer <jamrial@gmail.com>
2024-05-31 16:36:43 -03:00
668f43af20
tests/checkasm/lpc: correct arithmetic when randomizing buffers
...
Results weren't signed.
2022-09-23 01:50:59 +02:00
6ad39f01df
tests/checkasm/lpc: reduce range and use signed values
...
This is more similar to its regular use, and prevents inaccuracies
of huge float*float multiplications from failing the tests.
2022-09-23 01:42:34 +02:00
9cbfffa0d4
tests/checkasm/lpc: print mismatching values
...
Will help debugging.
Signed-off-by: James Almer <jamrial@gmail.com>
2022-09-22 18:18:52 -03:00
a1c6f4b653
tests/checkasm/lpc: randomize buffer length
...
Simplifies the test, while trying more values and preventing pointlessly
running benchmarks in a loop.
Signed-off-by: James Almer <jamrial@gmail.com>
2022-09-22 18:17:26 -03:00
c8c4a162fc
avcodec/lpc: use ptrdiff_t for length parameters
...
Signed-off-by: James Almer <jamrial@gmail.com>
2022-09-22 18:17:26 -03:00
b67776e12f
x86/lpc: fix even scalar loop overreads/writes
...
Passes checkasm with valgrind, tested to sizes of more than 4000 samples.
2022-09-22 04:27:19 +02:00
3ade6a8644
x86/lpc: implement a new Welch windowing function
...
Old one was written with the assumption only even inputs would be given.
This very messy replacement supports even and odd inputs, and supports
AVX2 for extra speed. The buffers given are usually quite big (4k samples),
so the speedup is worth it.
The new SSE version is still faster than the old inline asm version by 33%.
Also checkasm is provided to make sure this monstrosity works.
This fixes some FATE tests.
2022-09-21 07:12:39 +02:00
James Almer
Rémi Denis-Courmont
Lynne