And the denorm opcode is good for adding that low level noise. Very handy.
Oeyvind

2013/11/19 Michael Gogins :
> This is called the "denormal problem." It is caused when floating point
> numbers have an exponent that is too small for the number of bits allocated
> to the exponent, so put leading zeros into the significand. You can obviate
> the problem by adding a bit of low amplitude white noise to the signal. This
> may or may not actually be a problem for you. You can look up your CPU to
> see if denormal numers are handled in hardware, thus no problem, or in
> software, thus there is a problem. Or you can write a test to experimentally
> determine if performance suffers with denomals.
>
> Hope this helps,
> Mike
>
>
> -----------------------------------------------------
> Michael Gogins
> Irreducible Productions
> http://michaelgogins.tumblr.com
> Michael dot Gogins at gmail dot com
>
>
> On Tue, Nov 19, 2013 at 6:50 AM, Jeff Taylor 
> wrote:
>>
>> I recall reading somewhere, and I think it was on this mailing list, about
>> an interesting quirk of how numbers are computed on computers.  Signals
>> satisfying certain conditions cause an increase in CPU usage, and these
>> conditions are ones that could easily occur in everyday usage.  Thus you
>> would want to keep the effect in mind as you program.  I was thinking it was
>> something like very low amplitude signals or slight DC offsets.
>>
>> Does anyone have any idea what I am talking about?  I've been looking
>> through my notes but I can't seem to find this particular info and I would
>> like to refresh my memory on the topic.  I will accept the answer that I
>> dreamed it all up!
>>
>> --
>> Electronically,
>> Jeff Taylor
>
>



-- 

Oeyvind Brandtsegg
Professor of Music Technology
NTNU
7491 Trondheim
Norway
Cell: +47 92 203 205

http://flyndresang.no/
http://www.partikkelaudio.com/
http://soundcloud.com/brandtsegg
http://soundcloud.com/t-emp

They are called denormal numbers.

http://en.wikipedia.org/wiki/Denormal_number

cheers,
Jan

On 11/19/13 12:50 PM, Jeff Taylor wrote:
> I recall reading somewhere, and I think it was on this mailing list,
> about an interesting quirk of how numbers are computed on computers.
> Signals satisfying certain conditions cause an increase in CPU usage,
> and these conditions are ones that could easily occur in everyday
> usage.  Thus you would want to keep the effect in mind as you program.
> I was thinking it was something like very low amplitude signals or
> slight DC offsets.
>
> Does anyone have any idea what I am talking about?  I've been looking
> through my notes but I can't seem to find this particular info and I
> would like to refresh my memory on the topic.  I will accept the answer
> that I dreamed it all up!
>
> --
> Electronically,
> Jeff Taylor