I am guessing it probably takes about 3 minutes or so to load the Gig file, which includes everything.
I checked my task manager and it said I am using up 78% of ram.
I am happy with how everything is running. Ideally, I would like to avoid predictive loading or purging Kontakt samples because…well things seem to be working fine and I would like to avoid any changes/complications that might occur if I did used those tools (increased time to load rackspaces that were not predicted, etc.)
But, I guess I am getting close to having to do something about ram usage.
So, first a dumb question, what should I first notice if I am having and issue with ram? Should I get a bit of a heads up or will I create a new song with a new rackspace and GP just wont load?
This question comes up regularly in the community. A link to my last response is HERE, but you should also do a search for RAM in the community articles for numerous other discussions and permutations of the subject.
To address your question more directly, “how much RAM” is right for you can only be determined by you and what you want to run. At face value you have more than enough RAM at present according to any number of articles on the subject, but the reality is that what you need will be determined by the size of the sample libraries and specific plugins you want to use. Two examples I regularly point to is the Berlin Strings library which has a recommended RAM size of 32gb to load the whole thing, and NI Pharlight/Ashlight/(and the other I can’t remember) which can run hundreds of voices and can bring my computer with 64gb of ultrafast ram to it’s knees.
In any case, your comment about loading times is almost certainly not related to RAM capacity, and if you are not having any audio dropout problems when everything is up and running you probable have no problems due to sample libraries and RAM capacity either. There are a number of things that you can do to improve your GP load time, that’s another subject that comes up often and has a lot of replies and suggestions and is covered in the User Manual (available at the “Docs” link in the upper right corner of this webpage) and Knowledge Base section of the GP website. Doing a search for GP load time in the Community articles will provide a wealth of knowledge that will almost certainly cover your experiences.
There were users who used Gig Performer with 85% of RAM used. Personally, I wouldn’t go any further (say 90% or more). Keep in mind that if your Windows computer is not prepared (i.e. optimized) for a performance that other processes may run automatically and interfere with your setup and performance.
Gig Performer will not crash, but you’ll most likely hear glitches due to transferring data from RAM to the Page File.
Thanks, everyone, that was helpful. So, if I start creeping up above 85%, I think my best option is to split my Gig files into separate gig files for each band I’m in. (I tried to follow all the Windows optimization guides for GP).
How many rackspaces and variations are you using? What samplepacks for what virtual instruments are you loading? Are there other opportunities for optimization?
Something tells me there are other factors at play here that lay beneath the surface.
Personally I am in love with the predictive loading feature and number of rackspaces limiting.
I normally run around 35 - 40 rackspaces and was thinking about having to split them up to more manageable sizes until I turned on predictive loading and limited how many it can load. My plan is to just keep on adding rackspaces and running that way until such time I can determine it’s getting detrimental. GP is just so flexible, it never gets in the way of what you are trying to do.
Yes, it is a close call whether to use predictive loading or just split my complete Gig file into separate ones for each band I play in. But, I’m not doing anything right now. Nice to have both options.
Xpansion, I would not be surprised if I am doing something in a way that is not optimal. This is an ongoing learning curve for me.
Right now I have a total 68 Rackspaces and 110 songs (I’ll guess an average of two or three rackspaces per song(?). Song parts vary wildly. I only go “forward” in switching song parts. So, sometimes I will have (let’s say) 12 song parts that are just going back and forth between different rackspaces.
I uses a lot of different sample libraries loaded in the rackspaces. Not sure I will include them all, but:
NI Session Horns (this, and many others, in Kontakt)
Lots of IK Syntronik synths (bought Syntronik Max during the recent group buy, with 32 synth/romplers)
A few Arturia Labs synths
NI Vintage Organs (I have Komplete Collection Select 13)
Friedlander Violin
VG Tenor Tax, Alto Sax and Baritone Sax
Hanon B70 (for new songs I tend to use Vintage Organs instead)
A couple of Lounge Lizard rackspaces (I just have the small Session vst)
Neo Soundstation (nice bread and better, I find more usable than the Kontakt factory library)
I picked up a couple of accordion libraries and a harmonica library
I just picked up Music Sampling Barker Clarinet (use in the Scenes from An Italian Restaurant)
Realitone Banjo
I guess you can see why I love GP. It lets me use all these amazing tools!
Personally I use different gig files for different bands.
Remember also that you can “construct” a gig file quite quickly if you have exported all your songs. Just create an empty gig and drag in the songs you need
Yes, an upside of using different gig file for different bands is if they start skipping songs, I won’t have any issue with loading. (Unless someone starts jamming on a song that I play in another band).
I suppose a downside of creating separate gig files is, if I create a rackspace/song for one band that I would also want to use for another band (there is a fair amount of overlap), I would have go through extra steps of adding it to the other gig file (at that point I would learn exactly what to do).
No, I don’t use predictive loading…never needed it. But I don’t use much in the way of sophisticated samples and I try to use physical modeling when possible. (e.g., for acoustic piano, electric piano, Hammond organ, minimoog,etc)
Hmm, I checked again and it says I am using 45% of memory. (Maybe I misread something).
It seems to fluctuate a lot more than I would have expected. This morning after loading GP, it says 76% (w/ about 20 GB used in GP). After playing through a song and checking again, it now say 54% with 11.5 GB used in GP).
If I start getting concerned about memory, I have decided I would almost certainly start using Predictive Loading. I think it should work really well with Setlist mode, which is typically where I operate.
Setting up different Gig files by each band seems to be more difficult to administer and I will probably mess it up. For example, I usually set up a template file to duplicate and use for future songs. I would have to make sure it is included in different band Gig Files. (There is a fair amount of overlap among bands).
I can achieve basically the same benefit as setting up separate gig files by setting Predictive Loading to 49 rackspaces (the max permitted, as I understand it) and adding all of a particular band’s songs that are not in the setlist to the end of the setlist.
If that does not load all the songs/rackpaces I might use by that band (if we switch around) it will be damn close.
