Renoise Cpu Meter Reads Significantly Higher Than System (Linux X64)

As of recently I’ve been trying to max out my audio performance on Ubuntu Studio 12.04 and have realized that the CPU meter in Renoise reports using more of the CPU than all other monitors report. I was wondering if this is normal. Does the meter read CPU usage differently on purpose or is renoise only allotted a certain amount of processor resources and reports that percentage? It’s weird because if the Renoise meter hits 99.9% I still get cracks and slowdowns.

I have a quad core Intel Core i7 running at a fixed 2.66GHz.

I’ve included an example screenshot. Three separate load monitors report 23% use when renoise reports 67.2%.

Please keep in mind I am not trying to report a bug and this isn’t really causing me any issues, I’m just curious. ;)

3552 renoisecpu.jpg

Energy saving may throttle down the CPU frequency, which Renoises CPU meter does not take into account. This is very confusing, so we probably should try to fix that at some point. What the meter in Renoise displays is the “real, current” load of the CPU and not the possible one when it’s running at full speed, which is what the OS process viewer shows.

See for example From Os X To Ubuntu on how to force the CPU to run at full speed.

For audio production, throttled down CPUs can be a bad thing, cause it may take to long to speed up the CPU frequency again when you need it. When for example a heavy DSP kicks in at a certain point in the song…

I actually did think of that as being a possiblity before, which is why shortly before taking that screenshot and posting this thread, I locked the CPU at full speed in the BIOS and turned off any throttling and “boosting” features, so something else must be causing the difference…

IC. This is indeed strange then. Does it perform better when setting Renoise to 4 cores only?

There seems to be no noticable difference in performance between 8 and 4 “cores”, however the xfce task manager shows a slightly lower CPU use on 4 than 8, even though renoise reports the same. The only setting that seems to make a noticeable difference is 1 core, which spikes the CPU up an extra 20-30% in Renoise, but again, slightly reduces the reported use (even further than 4 “cores”) in other CPU monitors…

Do you see more realistic values when using Jack? I had a similar issue on OS X with another application, the issue ended up residing in a poor Port Audio driver, something which went away when using Jack.

Using Jack yields the same results. I can’t find any settings or tweaks that seem to change this, so I’m starting to think it has to do with my sound chip.