[asterisk-users] Use the NEW ulaw/alaw codecs (slower, but cleaner)
creslin at digium.com
Mon Nov 17 15:39:42 CST 2008
Benny Amorsen wrote:
> Matthew Fredrickson <creslin at digium.com> writes:
>> Actually, with the way caching is done on nearly all modern processors,
>> it is debatable whether or not a look up table is the optimal way to do
>> the conversion, at least on such a simple codec such as ulaw or alaw.
>> In fact, the amount of time it takes to fetch memory from a cache miss
>> can easily ruin the single element lookup performance in a look up
> If the compiler is clever enough, you can embed a small lookup table
> in the instruction stream. Instruction prefecting will automatically
> ensure the page is in I-cache, and even on most processors which can't
> read from I-cache the table will be in 2nd-level cache.
> Low-level optimizations like these are often dependent on processor
> architecture though.
This is very true. Mostly wanted to make sure that people knew that
cache miss penalties can be more of a slow down (and in fact will be for
a simple thing like a lin-to-mu) on a big, multi page table like in a
lin-to-mu lookup than simply executing the instructions from I-cache
(which are much more likely to not cause a miss due to the small number
of instructions involved).
More information about the asterisk-users