I recently added a peak meter to my chosen wps config on my iPod Video,
and got a shock: the music filled with pauses when playback started,
even when playing back 128kbps VBR MP3s, which should hardly be a
strain.

Looking at the audio threads shows the CPU boost at 100% and the codec
thread not keeping up until hard drive access is over and the buffers
have filled. With this toy .wps, I see this problem:

%ac%fbkbit %fc %?fv<(VBR)|(CBR)>
%pm

With this one, I do not, and CPU boost sits around 75%, even while the
buffers are filling, and after they've filled, the boost drops to 70%:

%ac%fbkbit %fc %?fv<(VBR)|(CBR)>

Is the peak meter really meant to eat >25% of CPU time on a softcodec
box? I'd have expected it to be rather a lot cheaper, even on a
not-yet-optimized port like the iPod.

(I suppose in the long run, if this stays this expensive, it can use
the iPod's second CPU...)

-- 
`On a scale of 1-10, X's "brokenness rating" is 1.1, but that's only
 because bringing Windows into the picture rescaled "brokenness" by
 a factor of 10.' --- Peter da Silva