Help with Kontakt Load Times in Studio One 7

[THW]

Hello and Happy Holidays!

I'm looking for some guidance on improving Kontakt load times in S1. I've been searching the net, including VI Control and the NI website and forums, and running into conflicting information, and hoping to get some insight from other S1 users for my specific setup. I am not very technical and having trouble wrapping my head around it.

I run a large template with many instances of Kontakt 7. I have batch resaved and typically purge samples. I have also encountered slow load times in Kontakt 7 and 8. I rarely use Kontakt in standalone mode. I have also experimented with a simple routing template, and dragging over track presets for instruments as needed, but because of the slow Kontakt load times, this ideal workflow is unfortunately is not working as well as I'd like.

I am wondering if there is anything I can do that would speed things up in S1 specifically. As an example, I was watching a great tutorial from Lukas and noticed how quickly instances of CSS, as an example, loaded when dragging over from track presets.

I think my rig is pretty solid, but perhaps not? I did a trial of Cubase and was struck by how much faster Kontakt loaded than in S1. I am using mainly orchestral libraries (multiple mic positions) and some other resource heavy libraries from Keepforest and Heavyocity. Musical Sampling Libraries (e.g. Soaring Strings) take a VERY long time to load. On the flip side, libraries like BBCSO and standalone applications are very quick and load in seconds. I am getting the sense that something is off with Kontakt. It takes minutes for my template to load (and I load with most tracks disabled).

Current System Settings:
Windows 11
AMD Ryzen 9 5950X 16 Core
64GB RAM
All libraries are on SSD (Samsung 860, 870 EVO, and SSD 980 Pro)


In particular, I am wondering if I would benefit from adjusting the "instrument preload buffer size" -- I have been using the default setting of 60kb. Any benefit if I were to turn this down to the minimum of 6kb?

There is an option for multiprocessor support (standalone): This is currently set to off. I have a 16 core processor...would changing this setting have any benefit when running in S1?

Greatly appreciate any insight on how I can speed things up!

Thanks,
Thomas

 

[Vocalpoint]

I have yet to find any tip, tweak or instruction that will speed things up. And all my libraries are on NVMe x4 and NMVe x5 drives.

Kontakt is just slow for some reason on some libraries and quicker on others. I have long since abandoned dragging Kontakt itself into any track these days in Studio One. Much prefer using Komplete Kontrol and THEN choosing a Kontakt patch is more my way of working.

Do remember - the guts of this thing are 20+years old - there is only so much that can be done. AND Kontakt is forever at the mercy of those who make the libraries - none are very good at optimizing anything.

And yes - Cubase will always be well ahead of S1 in the "large template" department. That is why more folks who dabble with large templates continue to gravitate to Cubase.

But to counter any need for a huge template in the first place - while I do not do large orchestral scoring - I lean more heavily on S1's superior Track Templates - which after some thought and setup - could easily erase any need for a large template.

I simply drag in what I need - when I need it - all tracks pre-setup and ready to go - and get working.

Never did quite understand how anyone could tolerate the navigation (or non-ease of use) of any template (in any DAW) with 500-1000 empty tracks - but that's probably why I don't do large scoring stuff

VP

 

[THW]

I have yet to find any tip, tweak or instruction that will speed things up. And all my libraries are on NVMe x4 and NMVe x5 drives.

Kontakt is just slow for some reason on some libraries and quicker on others. I have long since abandoned dragging Kontakt itself into any track these days in Studio One. Much prefer using Komplete Kontrol and THEN choosing a Kontakt patch is more my way of working.

Do remember - the guts of this thing are 20+years old - there is only so much that can be done. AND Kontakt is forever at the mercy of those who make the libraries - none are very good at optimizing anything.

And yes - Cubase will always be well ahead of S1 in the "large template" department. That is why more folks who dabble with large templates continue to gravitate to Cubase.

But to counter any need for a huge template in the first place - while I do not do large orchestral scoring - I lean more heavily on S1's superior Track Templates - which after some thought and setup - could easily erase any need for a large template.

I simply drag in what I need - when I need it - all tracks pre-setup and ready to go - and get working.

Never did quite understand how anyone could tolerate the navigation (or non-ease of use) of any template (in any DAW) with 500-1000 empty tracks - but that's probably why I don't do large scoring stuff

VP

Hey thanks Vocalpoint, appreciate your comment! Solid points made. I was hesitant to make this post as I know Kontakt can be a source of frustration for many -- but figured, hey maybe I'm missing something!

I edited my comment shortly after posting to note that I do utilize track presets. That is a great, standout feature which I do use. Unfortunately, the slow Kontakt load times I'm encountering make this less efficient than I'd like. I'm managing OK -- just a bit frustrating, especially when I'm in the flow.

Did you happen to try adjusting any of the kontakt settings like "preload buffer size" or multiprocessor support? Am wondering if adjustments to those settings might help...

 

[Vocalpoint]

Did you happen to try adjusting any of the kontakt settings like "preload buffer size" or multiprocessor support? Am wondering if adjustments to those settings might help...

None of those will help to my knowledge.

This is a well known hassle for many - if there was a magic solution - pretty sure we would have known by now.

VP

 

[BobF]

Sorry to say I've found nothing to speed up K7 either. I get the same load times with stand-alone and SO use, so I blame NI If you find a magic potion, please post back

 

[THW]

I have made some tweaks and am noticing improvements. Will spend more time to make sure this is working right with how I set things up -- but I am quite pleased at the moment. I was not expecting this to go so well.

TLDR:
I am experiencing MUCH faster K7 load times in the lates version of S1 V7.

I was hesitant to start tweaking because everything is "working" right now and I didn't want to break anything. But tonight...felt like going for it and happy I did.

Below is what I did over the past hour so...some very loose/casual experimenting.

All changes were done on Kontakt 7 (I imagine I will need to repeat some of the steps for K8.

First, I loaded my usual template without any changes and with some programs running in background as I would typically (e.g. spotify and a web browser):

The template used has roughly 115 separate K7 instances (mainly CSS, Jaeger, CB, KF, Heavyocity), no multis, and 50 or so other standalones, (as as well a number of fx plugins on each channel and mixbus -- though these fx plugs typically load very quick despite being resource hogs). Most Kontakt instruments, especially resource intensive libraries with multiple mic libraries have been batch resaved.

Sample Rate: 48.0 kHz
Block Size: 128

For this "experiment" I closed/restarted S1 between loading the template.

Load Times:

1st Load Time (default K7 settings): 2:09 min

Second Load Time: Adjusted multicore settings to 16 core and restarted S1. Load Time: 2:12 min

Third Load Time: Adjusted "preload buffer size" to 6kb, ensuring box was checked. multicore (16 cores) enabled.
Load time: 2:05 min

Note: Considering background processes/variables, not much of a difference in load times.

I then followed the steps in the following video to ensure windows defender excluded kontakt/nki files and locations I store Kontakt Libraries, all on SSDs:
How to SPEED Up Kontakt Load Times (Praise Tracks Simean Amburgey)

Then I restarted the computer.

On reboot:
First Load w/ multicore and lowered preload buffer size of 6.0kb: 2:14 min
Second Load w/ multicore and preload buffer size of 6.0kb: 2:06 min

Note: Again, no real difference in template load time. Fine -- not a big deal to me.

Good News:

I am experiencing MUCH faster K7 load times once in the DAW. Track Presets w/ K7 are loading very quickly. In some cases, immediately, no more than a couple seconds. For example, dragging over several fully loaded CSS instances (VL1, VL2, VLA, DB) from the browser loaded in under 10 seconds. This is a significant improvement on my end. A particularly long loading instrument I all but stopped using because it would take sometimes well over a minute or 2 (8dio studio suitcase piano), loaded in seconds.

I hope this can help someone. I'm cautiously optimistic, but so far so good.

 

[Surf.Whammy]

After considerable experimentation, my current strategy is to limit Studio One to no more than 25 VSTi virtual instruments at a time; and even then it can take a minute or longer to load a ".song", where for reference at present I am using an Apple iMac Retina 5K, 27-inch, 2019) because (a) it has an Intel processor, (b) it runs current versions of everything, and (c) it's inexpensive, costing approximately $1,000 at Other World Computing (Macsales) when a machine becomes available, observing that they refurbish the machines and provide in-house extended two-year warranty, which they honor if something stops working, although so far I have no problems and have been using older Mac computers for two or so decades.

My strategy is based on a few rules, and the goal is to keep everything simple and practical:

(1) limit it to 25 VSTi virtual instruments at a time

(2) when I am happy with an Instrument Track and its effects plug-ins, I generate the audio to an Audio Track; do a "Save as. . . " to make an archive copy, and start working on the next iteration where the Instrument Track is retired or removed, which makes it possible to start working a new instrument

(3) as the ".song" moves forward, there are more Audio Tracks than Instrument Tracks, and Studio One handles Audio Tracks very efficiently if only because there are fewer VSTi virtual instrument engines to consume resources compared to Audio Tracks.

(4) you cannot see many tracks clearly at once (Instrument Tracks, Audio Tracks, Bus Tracks) so limiting them makes the visual aspect less cluttered, and technically nobody can focus clearly and distinctly on more than perhaps a 3" by 5" notecard, evidenced by observing you can read perhaps three or so of these words without having to look somewhere else

(5) as the ".song" moves forward in development, I group Audio Tracks to Bus Tracks and then hide the Audio Tracks, where for example I might group all the drumkit tracks to a Drumkit Bus Track, so when if I am happy with the drumkit, all I need to see is the Drumkit Bus Track and then can use it for mixing

(6) since I save each step along the way as a separate ".song" using a naming scheme like {Song-PT1.song, Song-PT2.song, . . . , Song-PTn.song}, if I need to revisit something, I can import it along with its effects and make whatever changes I think are necessary, where for reference this is something Studio One does efficiently and elegantly

(7) I keep a few instruments as Instrument Tracks so I can fine-tune them as the song progresses, which usually are {kick drum, snare drum, bass, and an electric guitar or two}, primarily because I like snare drum rimshots and consider them to be vastly important for songs, as are kick drums and bass, which is evident by the most iconic Intro of a hit song ever, "Billie Jean" (Michael Jackson)

(8) here in the sound isolation studio there are 12 notes and 10 or so octaves, at least two of which are provided solely to entertain bats, birds, cats, dogs, dolphins, porpoises, sea turtles, and whales because humans cannot hear them, include babies and children whose parents have taken them to a KISS concert without providing OSHA-approved hearing protection

(9) because I only learned soprano treble staff as a child in a liturgical boys choir, I do everything with soprano treble staves, which I configure for transposition based on the range of the instrument, where bass is played two-octaves lower than notated and guitar is played one-octave lower than notated, with this part of the overall strategy mapping to needing to know only 12 notes and that they can be low, middle, or high, which is vastly easier than having to remember 120 notes in an overwhelming immediately conscious way

(10) when doing digital music production, providing articulations, dynamics, styles, and all that stuff in music notation (a) is useless nonsense, (b) wastes time, (c) accomplishes nothing because most VSTI virtual instrument engines cannot make useful sense it, and (d) is visually cluttering, hence is best avoided because (a) you can do most of it by using specific sampled-sound libraries where the musician is playing in the desired articulation and style, (b) you can control dynamics with gracious compressor-limiters and ducking, and (c) for practical purposes songs are played through car sound systems, computers with less than ideal loudspeakers, and headphones or ear buds at full volume, so all the subtle and "artistic" nuances are squashed and just map to annoying variances in the overall volume level--"annoying" based on the rule that if you cannot hear it, then it's useless noise and is not necessary, where dynamics might work in a concert hall but otherwise don't work

There are a few more rules, but the summary is that if you have 100+ instances of Kontakt (Native Instruments), SampleTank (Ik Multimedia), and so forth, then you can open the Studio One song and then play 18 holes of golf while it's loading.

No matter what you do (a) loading Instrument Tracks in Studio One takes a while, but (b) loading Audio Tracks in Studio One is efficient and fast; so managing the number of Instrument Tracks at any given step in the development of a song is a good strategy for optimizing your time and the use of computer resources toward the worthy goals (a) of keeping everything peppy and (b) of being able to focus on the music rather than on messing with computer stuff.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[THW]

After considerable experimentation, my current strategy is to limit Studio One to no more than 25 VSTi virtual instruments at a time; and even then it can take a minute or longer to load a ".song", where for reference at present I am using an Apple iMac Retina 5K, 27-inch, 2019) because (a) it has an Intel processor, (b) it runs current versions of everything, and (c) it's inexpensive, costing approximately $1,000 at Other World Computing (Macsales) when a machine becomes available, observing that they refurbish the machines and provide in-house extended two-year warranty, which they honor if something stops working, although so far I have no problems and have been using older Mac computers for two or so decades.

My strategy is based on a few rules, and the goal is to keep everything simple and practical:

(1) limit it to 25 VSTi virtual instruments at a time

(2) when I am happy with an Instrument Track and its effects plug-ins, I generate the audio to an Audio Track; do a "Save as. . . " to make an archive copy, and start working on the next iteration where the Instrument Track is retired or removed, which makes it possible to start working a new instrument

(3) as the ".song" moves forward, there are more Audio Tracks than Instrument Tracks, and Studio One handles Audio Tracks very efficiently if only because there are fewer VSTi virtual instrument engines to consume resources compared to Audio Tracks.

(4) you cannot see many tracks clearly at once (Instrument Tracks, Audio Tracks, Bus Tracks) so limiting them makes the visual aspect less cluttered, and technically nobody can focus clearly and distinctly on more than perhaps a 3" by 5" notecard, evidenced by observing you can read perhaps three or so of these words without having to look somewhere else

(5) as the ".song" moves forward in development, I group Audio Tracks to Bus Tracks and then hide the Audio Tracks, where for example I might group all the drumkit tracks to a Drumkit Bus Track, so when if I am happy with the drumkit, all I need to see is the Drumkit Bus Track and then can use it for mixing

(6) since I save each step along the way as a separate ".song" using a naming scheme like {Song-PT1.song, Song-PT2.song, . . . , Song-PTn.song}, if I need to revisit something, I can import it along with its effects and make whatever changes I think are necessary, where for reference this is something Studio One does efficiently and elegantly

(7) I keep a few instruments as Instrument Tracks so I can fine-tune them as the song progresses, which usually are {kick drum, snare drum, bass, and an electric guitar or two}, primarily because I like snare drum rimshots and consider them to be vastly important for songs, as are kick drums and bass, which is evident by the most iconic Intro of a hit song ever, "Billie Jean" (Michael Jackson)

(8) here in the sound isolation studio there are 12 notes and 10 or so octaves, at least two of which are provided solely to entertain bats, birds, cats, dogs, dolphins, porpoises, sea turtles, and whales because humans cannot hear them, include babies and children whose parents have taken them to a KISS concert without providing OSHA-approved hearing protection

(9) because I only learned soprano treble staff as a child in a liturgical boys choir, I do everything with soprano treble staves, which I configure for transposition based on the range of the instrument, where bass is played two-octaves lower than notated and guitar is played one-octave lower than notated, with this part of the overall strategy mapping to needing to know only 12 notes and that they can be low, middle, or high, which is vastly easier than having to remember 120 notes in an overwhelming immediately conscious way

(10) when doing digital music production, providing articulations, dynamics, styles, and all that stuff in music notation (a) is useless nonsense, (b) wastes time, (c) accomplishes nothing because most VSTI virtual instrument engines cannot make useful sense it, and (d) is visually cluttering, hence is best avoided because (a) you can do most of it by using specific sampled-sound libraries where the musician is playing in the desired articulation and style, (b) you can control dynamics with gracious compressor-limiters and ducking, and (c) for practical purposes songs are played through car sound systems, computers with less than ideal loudspeakers, and headphones or ear buds at full volume, so all the subtle and "artistic" nuances are squashed and just map to annoying variances in the overall volume level--"annoying" based on the rule that if you cannot hear it, then it's useless noise andis not necessary, where dynamics might work in a concert hall but otherwise don't work

There are a few more rules, but the summary is that if you have 100+ instances of Kontakt (Native Instruments), SampleTank (Ik Multimedia), and so forth, then you can open the Studio One song and then play 18 holes of golf while it's loading.

No matter what you do (a) loading Instrument Tracks in Studio One takes a while, but (b) loading Audio Tracks in Studio One is efficient and fast; so managing the number of Instrument Tracks at any given step in the development of a song is a good strategy for optimizing your time and the use of computer resources toward the worthy goals (a) of keeping everything peppy and (b) of being able to focus on the music rather than on messing with computer stuff.

Hey SurfWhammy, thanks for breaking down your process in detail. I find it so interesting to learn about different workflows. I can appreciate yours -- especially having a clean, uncluttered workspace. I haven't tried working like that -- saving each iteration as a separate .song. I see the merits and how S1 can handle that efficiently and elegantly indeed. Usually I'll save multiple versions once I enter the mixing phase, but until then (and upto now), I keep it all in one save.

I've find the S1 toolbox and filter features excellent to help keep clutter down and stay organized (like viewing only tracks with events, or filtering by instrument group).

In my mind, the added benefit of "committing" to audio as the song progresses also keeps the song actually progressing rather than getting sucked into vertical composition...layers on layers without moving it along. That is one area I sometimes get stuck in (though I find the arrange tool very helpful for that).

I don't typically use the music notation feature in S1, unless I'm bringing something I've written in notion (rare), or want to double check a musical phrase I'm referencing...for example if I'm doing a midi-mockup of an orchestral piece.

I have been exploring the browser features more, track presets, and revisiting tracks/instruments from other projects and dragging them over to what i'm working on. Kontakt load time is what was really turning me away from that. BUT -- I am optimistic about the setting changes I made to Kontakt tonight! With the faster load times, that is a game changer for me. Time will tell!

 

[OutrageProductions]

I have orchestration templates with over 120 VSTi instruments loading in one session. Most are KK (with Kontakt), VSL, UVI, etc.
I found that the thing that works for me is to set up the template and then <Disable> all of the Instrument tracks except for Scaler3 and a generic piano. This allows the basic template to load Very quickly, and when I need a particular instrument, I just RMB and <Enable> it. It then just takes the default load time for whatever that particular instrument may be. Hollywood piano loads immediately, Noire takes a little longer. Generic solo violin is short, Stradivarius Violin takes a little longer. Ad nauseum.
Works for me.

 

[Surf.Whammy]

I have orchestration templates with over 120 VSTi instruments loading in one session. Most are KK (with Kontakt), VSL, UVI, etc.
I found that the thing that works for me is to set up the template and then <Disable> all of the Instrument tracks except for Scaler3 and a generic piano. This allows the basic template to load Very quickly, and when I need a particular instrument, I just RMB and <Enable> it. It then just takes the default load time for whatever that particular instrument may be. Hollywood piano loads immediately, Noire takes a little longer. Generic solo violin is short, Stradivarius Violin takes a little longer. Ad nauseum.
Works for me.

That's interesting!

If I understand the strategy correctly, then you have an Audio Track for each Instrument Track and the "template" is just a list and a bit of control information for what the corresponding Instrument Track is and how to load it, should you want to have it loaded.

This way, you hear the song, since there are already generated and recorded Audio Tracks; but you only need to enable an Instrument Track if you want to change something and then render a new version of its Audio Track

If this is the way it works, then (a) I like the idea and (b) it's faster than importing Instrument Tracks from earlier versions or even different songs, although importing an Instrument Track from a different song would map to that specific Instrument Track not being part of the "template" until you imported it and then added it to the "template".

In other words, you load 120 Audio Tracks but no Instrument Tracks, although the information to load one or more of the Instrument Tracks exists in the "template".

I did a quick experiment and the effects plug-ins for each Instrument Track also are kept but are not loaded in the "template".

So long as there is an already generated and recorded Audio Track for each Instrument Track, this strategy works nicely!

The key to this making sense to me, after perhaps a decade of thinking people who suggested they load songs with hundreds of Instrument Tracks either had super fast Windows machines or were not being truthful, is your including "<Disable>" and "<Enable>", which was the clue I needed to understand this.

Previously, I thought those folks were running Cray Supercomputers or had their own Computing Cloud Servers.

I should have asked how they did it, but asking did not occur to me at the time.

At the risk of prideful boasting, I am very smart in some respects; but in other respects I am a bit profoundly stupid until there is an obvious clue.

Thanks for being clear and not hiding the key information and important clues!

 

[OutrageProductions]

120+ Instrument tracks (I rarely if ever use multis in Kontakt) each is a single instance of Komplete Kontrol with Kontakt or something similar with an individual instrument loaded (example: Session Strings). There is no audio at all when I start a project.
When I have a part written in MIDI driving (again, example: Session Horns Tenor Sax) I then transform it to audio with the ability to 'un-freeze' it later to re-edit if necessary. Move on to other tracks/instruments and repeat.
In orchestration, I rarely need EFX plugins on individual tracks, but sends to FX Busses for reverb/delay, etc.
I always track first, add any EFX (eq, comp, etc) when closer to mixing.

IIRC this is a pretty good video on setting up large templates:

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[Jemusic]

Kontakt 8 loads faster than 7 Quite a lot apparently Maybe it’s time to upgrade

 

[THW]

On my end unfortunately (at least with k8 player), I didnt have a difference between on load times between k7 and k8. I am happy to share that the tweaks I made above have helped quite a bit once the template is loaded. So I will make same setting updates and see what happens!

 

[OutrageProductions]

The key to this making sense to me, after perhaps a decade of thinking people who suggested they load songs with hundreds of Instrument Tracks either had super fast Windows machines or were not being truthful, is your including "<Disable>" and "<Enable>", which was the clue I needed to understand .

Logic, Cubase, Nuendo & PT all have some way to quickly enable/disable [activate/deactivate or On/Off] instrument tracks to reduce CPU & RAM buffer, Sonar calls it 'Archive'.
I've been doing it that way for longer than I can remember.

 

[TimC340]

My studio system is very similar to THW’s (I assume it’s a 5950X not ‘5050’ as you said in the OP!). I’m on Win 11 Pro 25H2; I use K8 Full, along with EW Opus and a few others. I do occasionally play with fairly large templates - quite often when attempting some of Guy Michelmore’s ThinkSpace courses!

I haven’t really experienced long load times, but I am using a technique similar to that described by OutrageProductions where I disable tracks I’m not actively working on having first bounced them to audio if I need to hear them during the process. The Kontakt instruments I’m using include the various orchestral libraries in Komplete Ultimate 15, Sonuscore’s The Score as well as free libraries such as Berlin, Spitfire BBC Discover, VSL and similar. With Opus, I have the full range of EW’s stuff from Composer Cloud+.

I’ve also started a trial of Cubase 15 to see how it’s changed since I last used it many years ago (I go back to Steinberg Pro24 and the earliest iterations of Cubase) and so that I can directly use some of the ThinkSpace course material.

 

[THW]

My studio system is very similar to THW’s (I assume it’s a 5950X not ‘5050’ as you said in the OP!). I’m on Win 11 Pro 25H2; I use K8 Full, along with EW Opus and a few others. I do occasionally play with fairly large templates - quite often when attempting some of Guy Michelmore’s ThinkSpace courses!

I haven’t really experienced long load times, but I am using a technique similar to that described by OutrageProductions where I disable tracks I’m not actively working on having first bounced them to audio if I need to hear them during the process. The Kontakt instruments I’m using include the various orchestral libraries in Komplete Ultimate 15, Sonuscore’s The Score as well as free libraries such as Berlin, Spitfire BBC Discover, VSL and similar. With Opus, I have the full range of EW’s stuff from Composer Cloud+.

I’ve also started a trial of Cubase 15 to see how it’s changed since I last used it many years ago (I go back to Steinberg Pro24 and the earliest iterations of Cubase) and so that I can directly use some of the ThinkSpace course material.

Yes, 5950x! Thanks for catching, made the edit.

Disabling tracks is definitely the way to go! Unfortunately for me it didnt impact initial load time in my casual testing, though many tracks were disabled aside from 20 or so.

Once writing especially, I will bounce certain events, like a synth patch or drone for both easier audio editing and to save on cpu. So nice how easy it is in S1 to go back and edit if needed. Transform in place is also a nice feature.

We use some of the same vsti as well. All the best on Cubase, it really is impressive and featureloaded!