Yesterday I played a 96kbps opus track on my everyday home laptop (2.8 Northwood / 1GB / XPSP4) with a modern-ish version of mplayer and noticed the sound stuttering badly while loading modern sites in mypal browser.

After closing everything, I noticed the cpu usage is between 13-17% while playing, with occasional spikes into the twenties. When I stop the player, it goes back to 0%.

Playing the same opus track on an iphone 14 also result in the occasional brief stutter here and there.

I find this insane.

I remember playing mp3 files on fast 486 machines back in the day, swallowing all the cpu time unless skipping half of the work. Pentiums did it fine with time to spare and p2’s hardly felt anything at all.

I did not do any speed testing, but anecdotally just eyeballing it, it looks like anything slower than a 500MHz wont work at all. Even faster stuff needs to decode ahead of time into a longer buffer, as the processing needed seems to be rather irregular in real time with occasional spikes that overrun the time in the codec’s frame decode buffer from time to time.

I get it that opus is designed with real time applications in mind, therefore such short latencies are involved, but for static playback it is not needed.

I tried searching for the term opus to see if anything has been posted about this on the forum, but on page 4 the search dingus switches to the term “opera” trying to be helpful. (Search dingus: an opus is a *work*, not an opera)

I will obviously do some testing in this regard in the future, but I was wondering if anybody here also has some anecdotes, experiences or results they could share.