Keeping in line with my recent tips on how to optimize Kontakt, here’s a great way of conserving memory. This is something that could prove to be invaluable for composers in particular due to the nature of the instruments they’ll typically be hosting in their templates. Large multi-sampled orchestral libraries can potentially use large amounts of RAM. Keeping that memory footprint as small as possible is important for obvious reasons.
Utilizing the purge function can free up RAM by unloading any unused samples. Some libraries allow you to do this from within the library UI whereby you’re able to deselect certain articulations or disable certain mic positions. This can be seen in Spitfire Audio’s Albion One Library below.
Here’s the available functions available in the Purge menu:
- Reset Markers – When a sample is played it marks it as being used by Kontakt. This function removes the markers but doesn’t purge anything so keeps the samples loaded.
- Update Sample Pool – Purges all the unused samples
- Purge All Samples – Unloads all samples
- Reloads All Samples – Loads all samples contained within the instrument
There’s a few ways you can go about implementing the integrated purge function. You could start with all samples within an instrument loaded (like it loads up by default). Once you’ve successfully laid down your part you could then simply reset the markers, run through the part from start to finish, then update the sample pool. This will unload any unused samples.
Alternatively, you could start with all samples purged within an instrument. Then as you input your MIDI it’ll load the used samples in on the fly. It’s important to bear in mind if you decide to use this method and you’re using hard disk drives to host your samples, you may experience a few clicks/pops/missing notes on the first run through as the samples load from the disk. I’d certainly recommend using SSDs if you use this method.
DFD (Direct From Disk) settings should also be considered here. When using DFD, only the first part of the sample is loaded into the RAM. The DFD buffer setting determines how much of the sample is pre-loaded into the memory. Lower settings load less of the sample into the memory so will decrease your memory footprint. As more of the sample is being loaded directly from the disk it goes without saying that SSDs will perform better than mechanical disks.
Working with a large template and having every sample in every patch loaded may not be the most efficient way of working. Using this method, you could theoretically build a template consisting of many patches, each with the samples unloaded whilst keeping your memory footprint relatively small. You’d just simply load in the samples as and when you need them.
Hopefully the methods I’ve discussed over recent weeks will help you get the most out of Kontakt. If you missed my previous articles, here they are for quick reference: