David Henningsson
4d84a00b49
SSE/MMX/ARM: Fix high frequency noise with unusual number of channels
...
In the assembly optimized versions of SSE, a noise could occur when the
number of channels were 3,5,6 or 7. For MMX and ARM, this could occur
when the number of channels were 3.
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
2010-10-13 14:52:03 +01:00
Lennart Poettering
54025c96a7
cpu: check for CMOV flag before using this intsruction in assembly
...
http://pulseaudio.org/ticket/776
2010-02-09 22:32:39 +00:00
Wim Taymans
056930cbcd
svolume: fix MMX error
...
We need to sign extend the lower part of the multiplication before adding it to
the higher part. Makes -1 * 0xffff work again.
2009-10-30 05:09:15 +01:00
Lennart Poettering
e7e4e4459e
simd: update test cases
2009-10-29 01:52:39 +01:00
Diego Elio 'Flameeyes' Pettenò
8ec304d2d1
Fix build when using -fweb, accept both register and memory constraints.
...
This was reported as Gentoo bug #287391 by Torsten Kaiser, and the fix was
suggested by Mike Frysinger.
2009-10-07 19:31:11 +02:00
Wim Taymans
5eecd8ea7d
svolume: tweak constraints for 32 bits
...
Tweak the constraints a little so that register starved 32bit systems
can select a stack variable for the channel paramter instead of reusing one of
the registers we're using in the code.
2009-09-23 17:18:20 +02:00
Lennart Poettering
231c17be03
svolume_mmx: disable test accidentaly left on
2009-09-17 01:34:02 +02:00
Wim Taymans
3d5a572694
svolume_mmx: optimize some more
...
We can reorder the algortihm a bit like we do for sse so that we
don't need the contants and masking instructions. Saves 2 instructions
for the mmx code.
2009-09-16 17:14:12 +02:00
Lennart Poettering
71e066c873
simd: be more precise which SIMD optimizations we activate
2009-09-09 04:28:22 +02:00
Wim Taymans
f8ffe0dabc
svolume: cleanups
...
Use PA_MAX
Use pa_rtclock_now() for benchmarks
2009-08-20 12:30:48 +02:00
Wim Taymans
f09b51198f
whitespace fixes
2009-08-20 11:31:04 +02:00
Wim Taymans
370016c0e7
svolume: fix compilation in 32bits
2009-08-20 11:31:04 +02:00
Wim Taymans
bd49d43bd3
svolume: add CPU guards around code
...
Mark code that should only be compiled on x86 CPUs with proper defines.
2009-08-20 11:31:03 +02:00
Wim Taymans
a98fa950d2
svolume: remove unneeded compare
...
We don't need the compare because the sub operation already set the right flags
for us.
2009-08-20 11:31:03 +02:00
Wim Taymans
591baacba5
volume: remove ref functions
2009-08-20 11:31:03 +02:00
Wim Taymans
f24c24c14b
volume: improved comments
2009-08-20 11:31:03 +02:00
Wim Taymans
a1235446a7
volume: make the benchmark more meaningfull
...
MMX is about 6x faster, SSE around 15x on my machine.
2009-08-20 11:31:03 +02:00
Wim Taymans
dcae9a3113
svolume: add some comments
2009-08-20 11:31:03 +02:00
Wim Taymans
563cb2dea9
main: hook up cpu detection code
...
Add CPU detection code to activate the various optimisations.
Move some method definitions around.
Use compatibility macros when we can.
2009-08-20 11:31:03 +02:00
Wim Taymans
5998cf99b0
svolume: improve SSE and MMX code
2009-08-20 11:31:02 +02:00
Wim Taymans
08f3e16c84
volume_mmx: fix mmx code a bit
2009-08-20 11:31:02 +02:00
Wim Taymans
3a0b012ee0
volume: add first mmx optimized function
...
Add code for an mmx optimized version of s16ne volume scaling. Install the
custom function.
2009-08-20 11:31:02 +02:00
