Background: I’ve been a daily Ableton user for the last five years. It has been my DAW of choice from the beginning and until recently it was the only one I’d ever used. Since the release of Live 10 I began to notice a growing number of complaints online regarding Ableton’s performance, but I never understood why. That changed after I tried Bitwig. But before we go any further, let’s define “slow”.
When I say Ableton is slow, I’m referring to project load times and the time it takes to load certain actions within the DAW. These actions include adding and removing stock effects, stock instruments, plugins, racks, tracks, and groups, as well changing channel routings and moving tracks and effects around a project. In my experience, these load times are directly related to the total number of effects being used within the project. For example, it takes Ableton a few seconds to load a completely blank project and the above actions, such as adding a new track, occur instantly. Add the tracks and effects contained within my default template and it now takes 19 seconds to load the project and 1.15 seconds to load actions. And, as expected, these numbers only grow as the project gets bigger.
In comparison, it takes Bitwig 3 seconds to open an identical template and actions are loaded without any delay whatsoever, just like the empty Ableton project. It would be easy to conclude that Bitwig is fundamentally faster than Ableton, but after digging through hundreds of threads, I'm not convinced. For every post seeking help with performance it seemed like there was another relaying the opposite experience. For every person asking why their sets take minutes to load, there was another saying it never takes more than fifteen seconds, even for their largest projects. So I wanted to know whether this sluggishness was in fact a part of Ableton or if something else was responsible and if anything could be done to optimize Ableton's performance.
----------
My Specs:
i7-9750H
GTX 1660 Ti
32gb RAM
2x 1TB WD SN750 NVME SSD's
Motu M4 Interface
Windows 10
Ableton 11
60 Channel Test:
To test the impact of effects on responsiveness, I created a blank project and added my default track. In the beginning, this track was set up as follows:
With 60 instances of this track, it took 2.7 seconds to load a new audio channel.
My first thought was, is it the track count or is it the effects? Running the same sixty channel test with blank tracks resulted in a near instant new track load. So it wasn’t the track count, it was the processing. With this in mind I would make one change to the default track, create 60 instances, record the time it took to add a new track, then revert the change and repeat the process.
Removing Pro-Q3 dropped the time by 0.3s.
Removing BC (Blue Cat's Gain plugin) dropped 0.3s.
Removing M/S (Utility mid/side split rack) dropped 0.65s.
Removing Gain (just a Utility) dropped 0.45s.
Removing M/S and Gain together dropped 0.9s.
Removing StandardClip dropped 0.2s.
Removing Spectrum x2 + Utility dropped 0.15s.
So far the results made sense. Fewer effects = less delay. But it was the more unexpected outcomes I found to be the most useful.
Ungrouping the Level rack dropped 0.45s.
Combining every effect, including the Level rack, into a new rack added 1.1s.
Takeaway: Racks are some of the largest contributors to sluggishness (but not always).
Grouping M/S and Gain together while inside the Level rack dropped 0.25s, while grouping them together outside the Level rack added 0.15s.
Takeaway: Racking within a rack can sometimes decrease delay.
Ungrouping Blue Cat increases the time by 0.15s.
Takeaway: Racks containing a single plugin can potentially decrease delay.
Replacing Pro-Q 3 with EQ8 made no difference.
Takeaway: Stock effects may contribute the same amount of delay as their equivalent VST’s, though more testing would need to be done to be certain this isn’t an isolated case.
Unmapping the Gain macro - no difference.
Takeaway: Macros do not seem to contribute to sluggishness (more testing needed to be sure).
Freezing all 60 channels - no difference.
Takeaway: Freezing tracks makes no difference.
Some more miscellaneous testing:
Disabling mono inputs and outputs - no difference.
Removing the auto-updating channel numbers - no difference.
Unplugging the interface dropped 0.4s but running off of the built-in card had no effect so the interface itself likely does not contribute, or at the very least it performs the same as the stock sound card.
Summary: The key to preventing slow down is using as little processing and as few racks as possible during the writing process. This also means cutting the effects on your default audio and midi tracks to the bare minimum, as well as deleting any unused processing as you write. I find the effects on my default tracks important to my workflow and so ultimately I chose to ungroup the Level rack (since it served no functional purpose) and leave everything else, while making a conscious effort to delete any unused processing while writing.
I know this is often the advice given to those seeking to reduce CPU load, but I never experience CPU problems. This is about workflow. Nothing kills momentum more than having nearly every action be preceded by a five to ten second delay and that’s exactly what would happen every time I began making real progress on a project. So from here I wanted to know what sort of impact minimizing the use of processing and racks could have on a real world example.
----------
Project Optimization:
I took a recently finished tune and cut away all the unnecessary processing, as well as some unused channels. I was left with a 65 track project. The bass design was done in a separate session, but there were still a handful of synths and there was processing taking place on every channel.
Original Project Load Time: 100 seconds
Optimized Project Load Time: 50 seconds
Original Action Response Time: 5 seconds
Optimized Action Response Time: 2.85 seconds
By cutting away only what wasn’t being used and by ungrouping unnecessary racks, I was able to cut the project load time in half and shave off over two seconds from the time it takes to perform an action like adding an effect. But these times didn’t stay consistent.
After some Ableton restarts the load times became 55 seconds and 3.3 seconds.
Following another restart, the load times became 75 seconds and 3.3 seconds.
Another restart, 65 seconds and 3.3 seconds.
I put the computer to sleep for a couple hours and tried again.
It dropped back down to 50 and 3 seconds. What’s with the variation? Could it be my PC?
Here’s a list of everything else I tried in an attempt to stabilize / further decrease load times:
-killing the index process
-updating all plugins
-Collect All and Save
-removing my music library folder from the sidebar
-closing an 8 tab Firefox session
-windows Disk Cleanup
-CCleaner
-deleting unused MIDI mappings
-setting the minimum processor speed to 100% (only succeeded in making the fans go crazy)
None of the above made any difference. The only thing that helped was disabling Windows Security. This cut about 5 seconds of project load time. But even with a load time of 45 seconds and a response time of 3 seconds, Ableton is still nowhere near Bitwig in terms of speed. So let’s circle back to the original question: Is Ableton slow? I don’t know. It’s possible my experience is an isolated one. Maybe there’s something wrong with my computer. Maybe some people see the same responsiveness in Ableton as I see in Bitwig. Although even professional producers with the most powerful computers seem to experience delay, at least to some degree. Nasko recently streamed a 90 track project with a response time of 1.65 seconds and during one of Culprate’s album streams he had a 130 track project with a 4.5 second delay.
This whole thing started as a reaction to Bitwig’s speed. Prior to downloading the trial I had never even thought about how slow things were happening in Ableton. It was just the way music software operated. But even now I have no desire to make a change. I love Ableton, I just wish it were faster. I hope this thread can serve as the spark for a discussion surrounding Ableton’s performance and that some of the things I’ve come across might be useful to anyone battling a sluggish DAW. So let’s hear it:
What is your experience with Ableton in terms of responsiveness? Is it at all similar to mine?
How long does it take to load a typical finished project?
How long does it take to do something like adding a new track in a finished project?
If you’ve tried both, how does Ableton stack up to Bitwig in regard to load times?
Do you have any recommendations for decreasing load times?
Looking forward to talking Ableton!