Fixed-point audio codecs often use saturating arithmetic, and special instructions for these operations are common. Signed-off-by: Mans Rullgard <mans@mansr.com>