First, I must say that I'm generally very impressed with the current version - does just about everything I want.

However, I have found that it can exhibit high CPU usage at times - say 10-15% on a dual core Athlon 64 (2 GHz). I've also observed it on a P4 machine. Both running XP SP3.

This occurs even when minimised to the task tray. Usage stays consistently high until I fiddle around - change song, or change the number of spectrum bands etc, after which it sometimes drops to the usual level of 0-1%.

I haven't installed any third party add-ons, and run fb2k with the "Album List + Properties (tabbed) + Visualisations" Quick Setup.

Visualizations require significant CPU cycles.

Visualizations stop being calculated after about a minute after the window is minimized.

You could also remove the visualizations from your layout.

Thanks - I've changed to "Album List + Properties (tabbed)", and I'll see how it goes.

However, I have had the window maximised with 160 spectrum bands, and CPU usage around the 1-2% mark. I think they can eat up more CPU time than they should.

I think what's more disconcerting and problematic is the memory usage of the spectrogram.

In my use, with a 24bit/96kHz 5.1 FLAC and Secret Rabbit Code resampler set to Medium Sinc Interpolator and 96000Hz (not sure if it actually does anything in this case) and without the Convert 5.1 to stereo DSP active, the spectrogram, with FFT Size set to 16384 and a window size of 363 by 345 pixels, gobbles up 37MB of RAM. When full-screened to 1280 by 1024, it easily uses over 225MB of RAM and could use even more if there weren't skips in the image.

With Milkdrop running on my main screen at the same time, my computer takes a big steamy crap, and I've got a very capable rig.

God only knows that ridiculous amounts of eye candy are absolutely essential for full enjoyment of music playback.

I'm not entirely sure if this post should go in this thread, but I've also noticed some odd cpu behavior.

I noticed that when any other window is open in front of the spectrogram, and the window is small enough for the spectrogram to be seen on both sides of the window in front of it, one of my cpu's (Core2Duo machine) maxes out. However, if any part of the window in front of the spectrogram is covering the exiting edge of the spectrogram, the cpu usage drops to it's normal value (~1%).

But wait, theres more...If the window is covering the entering edge of the spectrogram, and the top or bottom of the spectrogram is also visible, the amount of cpu usage seems to scale with the amount of area the window covers. If the window only covers a small area of the entering side of the spectrogram, the cpu usage is correspondingly low. As I move (or resize) the window to cover more of the spectrogram, the cpu usage scales with that area until at some point the cpu is maxed out.

It seems to be tied to how many columns of pixels the window covers, as long as the window doesn't pass beyond the exiting edge of the spectrogram...

I don't have time right now to do more investigation, but hopefully others can confirm and investigate further.

I am using version 9.5.3.