I know the topics of CPU utilization, multi-threading and audio glitching has been discussed here ad nauseam, and I have read the related postings. So I guess I could use some advice on how to balance the number of rackspaces (to reduce load/startup time) and CPU utilization (to reduce audio glitching). Here is my setup.
Without playing a single note, GP’s CPU meter is reading 69%. On my four-core i7 laptop, Windows Task Manager only shows 26% CPU load and Windows Resource Monitor shows that the load is distributed across all four processors. Although CPU 2 has the balance of the load, it is still at around 60%. When I start playing notes on my MIDI controller, the CPU meter in GP quickly spikes to 90-100% and audio starts glitching badly.
In this configuration, SampleTank is loaded with 11 instruments that I switch between for pianos, clav, guitars and strings. The B-3X plugin is just a straight B3 through a Leslie 147 with some drawbar widgets on the front panel. I was hoping to use this rackspace with multiple variations to cover about 2/3 of my setlist. This rackspace loads in under 30 seconds, so I thought this was better than the > three minutes I was seeing when I had a rackspace per song. However, this rackspace is unplayable due to the constant audio glitching.
I assume that I need to pair back the number of instruments loaded into SampleTank, but I was hoping for other options that would allow me to maintain this configuration or something close since it otherwise meets my needs.
What are your audio buffer settings in GP? Look in options>preferences>audio setup. I don’t have the plugins you’re using, but others on the forum have noticed that B-3X is CPU-heavy.
Even when I replace the B-3X plugin with the Arturia V Collection B3 (which is MUCH lighter), I’m still glitching, although the CPU at rest is now at 40% rather than 70%. Still seems high considering I’m only using two output plugins.
GP’s CPU meter drops to ~ 25% when I remove the SampleTank 4 plugin from the rack.
SampleTank running stand-alone (outside of GP) with all 16 instrument slots loaded consumes about 20% of my computer’s CPU when idle and about 27% when I am playing.
Since the SampleTank and B-3X plugins are so heavy, looks like I need to go back to creating more rackspaces, which will increase load time and memory utilization. As I’ve stated in other posts, my band doesn’t work off a setlist so I need to be able to call up any song at a moment’s notice. For that reason, predictive loading doesn’t help me to manage load time and memory. Oh well…
First of all, Gig Performer is showing only the CPU cycles used by the audio processing. That’s completely different from overall CPU usage which you would see using the Windows Task Manager (say)
Secondly, SampleTank is consuming an absurd amount of CPU cycles even when no audio is being produced. A well-behaved plugin should barely have any “audio processing” CPU activity when there’s no audio being produced.
(See movie for example, using DS Thorn plugin, watch the (slightly lagging) meter as the audio comes and goes)
Agreed. The IK plugins are resource hogs, but they sound great!!
To compare, I created a new rackspace using only Arturia plugins from V Collection 7. This setup, which lets me switch between acoustic piano, Rhodes, Wurly and Clavinet on the bottom controller along with B3 on the top controller, uses under 20% at rest and under 60% when playing with everything active. No glitching whatsoever. When I use “bypass” widgets to select only one of the pianos, as I normally would within a variation, the numbers drop to 6% and 20%.
The IK plugins are the issue, I believe. I already submitted a ticket for the B-3X resource issue on the IK forums. Have not done so for SampleTank 4, although it’s not bad when you only have one or two instruments loaded.
Does anyone else have experience with the SampleTank 4 plugin in GP? If so, are you loading multiple instruments within a single instance per rackspace, or creating multiple instances within a rackspace, one for each instrument? Are you seeing the same resource-intensive audio processing utilization?
I have tried both ST3 and ST4 on a Mac and have found them to be so CPU intensive that I just won’t use them in a live situation. If there was a sound I really wanted, I would just sample it and throw it into Kontakt.
Through trial and error, I am finding that ST4 is usable as long as I don’t load more than two instruments per instance. In rackspaces where I need multiple ST4 instruments, I create separate plugins and bypass those not required for a particular variation using widgets (switches). Same for the B-3X; I bypass when not using it and the CPU goes way down. Not the best option, but it works.
Yeah, I get a similar result now loading just one instrument at a time. Initially, I was trying to create one mega rackspace with ST4 loaded up with 12 instruments, but just way too resource intensive (particularly with IK’s B-3X plugin as well).
I believe the lesson learned here is that audio processing load is directly related to the number of active plugins/instruments, and possibly at an exponential rate. I also see between 4-12% with only one plugin active. When I had ST4 loaded with 12 instruments and B-3X (which is very heavy), it was trying to process audio for 13 plugins simultaneously, which exceeded GP’s or my audio interface’s capacity.
There are also obvious differences between Windows and Mac to consider, with the Mac having a clear advantage in audio processing in my opinion.
