Ive been using Predictive loading set to 1 recently, and it’s working very well. Time between songs is not noticeable with the band (faster than my guitarist changing patches on his gear).
I have noticed that before I close GP after a session where we have played about 40+ songs that Activity monitor is saying GP is using about 21GB of memory. When it first opens to 1 song it’s using about 5GB.
Is this how it works? Ive not had any issues, but was surprised at how much the used ram had grown.
This is on a MacBook Pro running Mac OS 15.6, running the most up to date version of GP 5.
RAM usage is controlled by the OS at the end.
I have the same issue, but in Windows. I use a large gig file of more than 550 rackspaces with predictive set to 1. My ram usage does creep up also, to the extent that whenever I quit any gig file, the next one refused to load despite clicking on the icon many times. I then had to quit GP and reload it before I could proceed with what I wanted to do. I was advised by a forum member to check the task manager to see what memory was available. Not only was it way up to 80% or 90%, there were other instances of GP still there from previous editing sessions. The answer was to clear out the ram, so I found a small Windows app called “PC Manager” which does that when I click on it. I keep it sitting on the task bar and use it after editing. Not sure if there is anything for Mac that will do the same job.
I find the Mac is ok with clearing the ram when GP is shut down, it’s more the fact that I assumed predictive loading was for those with less ram. If it ends up using as much as the whole gig file would have by the end of the night im not sure what the advantage is.
I did a gig on Sat when I had it set to 1, and was going through the songs sequentially. When I finished the gig ram usage was sitting at 29GB for GP. Everything seemed to run smooth enough though.
“Gretchfrage” - key question: what’s better then? I made the same observation when switching to predictive loading:
- my gig-file used up to 21-22GB of RAM on a 16GB MBP which was obviously handled by macOS. I had some glitches/random crackling when switching to rackspaces which contain Kontakt with a larger instruments or Pianoteq
- I switched to predictive loading (5) and the glitches have gone. However I lose some flexibilty, as predicting may cause some delay, if I do not stick to my setlist and jump bigger distances. I also see growing RAM usage after start and moving on in the setlist
So finally, back to the initial question: when does the RAM get freed again?
Won’t you need to clear it periodically as the night goes on so you don’t run out of memory?
Regarding predictive set to 1. It was added at my request, and to me, it has proven to be very useful (Many thanks developers!) I use bass and drums in most of my rackspaces along with other instruments added. So the rs can be quite large. I save each song as a rackspace and my big gigfile (500+ rackspaces) would easily run out of ram long before many songs were loaded. I asked for the predictive 1 setting because I wanted to play large sets of songs at random and not in the songs list order. This allows me to I jump around according crowd requests etc. Previously, the lowest predictive setting had been 3, and I didn’t see the point of loading/unloading extra songs that I wasn’t going to play.The only disadvantage is the several extra few seconds of waiting time before the song is ready to be played. But I have gotten used to that. The longest wait has been about 10 seconds.
I’d like the answer to that too!
I don’t really have an issue with ram. Normally at gigs i would use predictive loading set to about 30. We have been rehearsing a lot and jumping around the set, so i lowered it to 1 as i assumed this would be more useful, as i sometimes find it takes a while to settle down and let me play when using a higher number and jumping around (pretty much what you have said).
As it was working so well and the gaps were short enough i thought I’d stick with it for this gig as I want too sure of the full running order.
I don have a problem as such, just wondered why the ram used was so high when predictive loading was designed for computers that have lower ram (I believe).