mirror/JUCE
1
0
Fork 0
mirror of https://github.com/juce-framework/JUCE.git synced 2026-01-10 23:44:24 +00:00
Code Activity

Removed some legacy support for non-SSE2 CPUs. Added function FloatVectorOperations::disableDenormalisedNumberSupport()

Browse source
This commit is contained in:
jules 2016-03-01 16:32:08 +00:00
parent 889a6e200f
commit beac2e4b3a
2 changed files with 15 additions and 22 deletions
Download patch file Download diff file Expand all files Collapse all files

6
modules/juce_audio_basics/buffers/juce_FloatVectorOperations.h
View file

@ -198,6 +198,12 @@ public:
Effectively, this is a wrapper around a call to _MM_SET_FLUSH_ZERO_MODE
*/
static void JUCE_CALLTYPE enableFlushToZeroMode (bool shouldEnable) noexcept;
/** On Intel CPUs, this method enables the SSE flush-to-zero and denormalised-are-zero modes.
This effectively sets the DAZ and FZ bits of the MXCSR register. It's a convenient thing to
call before audio processing code where you really want to avoid denormalisation performance hits.
*/
static void JUCE_CALLTYPE disableDenormalisedNumberSupport() noexcept;
};