It is happening on one thread of one core
1 Like
Thanks for that clarification 
Just to add my experience:
The critical GP CPU reading before clicks and pops occur in the audio may vary from system to system.
For me it is approximately 60% until drop-outs start.
To be on the save side I design my sounds such that CPU reading is well below 50% while heaviest reasonable playing of notes. The “both arms full length all keyboards cross-check” would be the worst case but not realistic, at least for my style of playing …
One question aside: I have set NI Kontakt to multicore CPU support. Is this multicore usage considered in the GP CPU reading?
3 Likes
I don’t know what you mean by “considered”
As explained above, the GP CPU meter is measuring one thread, the one used for audio processing. Clearly if a plugin is using multiple threads, i.e, in multiple cores, those threads are obviously not the audio processing thread. Consequently less work is being done by the audio processing thread and so the meter should report a smaller number. But again, this gets very complicated - there are other delays incurred when dealing with multi-thread processing, it’s not a panacea.
2 Likes
Just for investigation and diagnosing: when you change in task manager the affinity mask of Gigperformer4.exe such that it uses only 1 core, you will see in task manager (provided you’ve changed the cpu graph to ‘logical processors’) a much more 1 to 1 relation between gp’s cpu usage display and that of the taskmanager. You can’t, however, get this completely right because this way the gui processing and the audio processing are bound to one core (and of course this way the audio thread will not get enough cpu when the gui thread takes to much).
Maybe this way you get a better understanding how the cpu-scheduling in windows works…
1 Like
Here an example of how my system behaves with everything on one core:
You can see me overloading the core (bottom left corner: core clips at 100%), GP says cpu 58% (but it does not take in account the gui processing). For this to happen I actually used two instances of a graphic-intensive plugin with its gui visible.
Of course, normally you shall not bind GP to just 1 core.
2 Likes
I used a Behringer until about 2 weeks ago, where I also had pops/crackles starting at around 50%, now I’m using a FocusRite which doesn’t crackle only when (close) at 100%.
Thanks for the explanation.
Just checked, and my affinity for GP is set to all CPUs.
I’ll give my two cents . Everyone has their own threshold of perceived latency. I personally use 256 buffer at 44.1 sample rate with hardly ever any issues (or perceived latency to me). The only plugin that has issues is PianoTeq when too many notes are played too fast. PianoTeq has a setting for “Internal sample rate” that I can dial back for the songs were I notice this issue.
Another issue to consider is plugins can act differently on different computers. I brought up that one of the GP supplied plugins was crushing my CPU usage and was unusable with many replies saying they did not have the issue. It was never resolved but simply said I can not use that plugin!
https://community.gigperformer.com/t/bare-gp-installer-2/10672?u=aharry
Honestly, that sounds more like an audio device driver issue.
It always is for every process (so you will not have to worry about it), unless you change it. And when you change it, the setting only persists until the process stops. You would have to do some effort to change automatically the affinity-mask from all cores to a specific amount when gp starts. But you should not want to do that for gp, so no worries 
.
Bottomline: Do nothing about it and (in this respect) everything is Galaktisch Gut (as Michael Tippach from asio4all would say😁)
1 Like
Good to hear … I thought already that the option was at a strange location (I mean, not part of the properties of the application itself, but as item in the task manager, thus not persistent after restarting GP).
1 Like
Well, RME Babyface Pro, in my case. Using 96 samples buffer @44.1 kHz.
1 Like
These are actually great results.
Yes … I used just as a kind of mixer before a Behringer, so I bought two extra (one for the rehearsal room and one for my VST laptop), but it turned out it didn’t work well for audio production, so after buying two audio interfaces ‘for nothing’, I eventually got the right ones (for me) … luckily I didn’t need to replace the laptop (it’s not the fastests, but it does the job good enough).