Strata At Scale

Introduction

I been contacted by a few people involved in movie scoring and soundtracks about the use of Strata in very large DAW projects. Movie DAW projects become complicated very quickly because they have to incorporate so many different kinds of tracks:

• Full orchestral sections (strings, brass, woodwinds, percussion), often layered, divided, and doubled.
• Orchestral sound design such as braams, clusters, metal hits, and low-frequency impacts.
• Electronic and hybrid effects including risers, transitions, pulses, and rhythmic textures.
• Synthesizers, pads, drones, basses, and harmonic beds layered across multiple registers.

There are so many moving parts in these DAW projects that the track count can run as high as 2000 tracks. To see if Strata could be used in such large projects, I performed an experiment to test Strata at scale.

Experimental Design

• Date: 14 January 2026.
• Computer: Mac Studio with 96GB RAM and 28 cores (20 performance and 8 efficiency).
• Operating System: macOS Sequoia 15.6.
• DAW: Bitwig 6.0 Beta 12 in "by Vendor" sandboxing mode.
• Number Of Tracks: 600
• On Each Track: Sylenth1, Tool, Strata Track (VST3).
• On One Other Track: Strata Vision (VST3).

Bitwig in "by Vendor" mode creates a separate macOS process for each vendor's plugins. so, as expected, looking in the mac app Activity Monitor showed that there were three "Bitwig Plug-in Host ARM" processes, one for each of Sylenth1, Strata Track, and Strata Vision. The amount of CPU and memory used by each of these processes can be used to establish an upper bound on the amount of CPU memory each plugin instance was using.

Sylenth1 was chosen because it is an old efficient synthesizer that could make a sound without consuming significant of CPU or memory. During the test, MIDI was fed into Sylenth so that during the test, all tracks were carrying sound.

Experimental Results: Memory

Conclusion: Strata consumed a total of about 6MB per track at scale.

Experimental Results: CPU

At 600 tracks, the project was stuttering, so I deleted 100 tracks. At 500 tracks, the Bitwig CPU meter showed that the project was under load, but there was no longer any stuttering. Here's what Activity Monitor looked like:

Conclusion: Strata consumed a total of about 1% CPU per track at scale.

Stability

Strata did not crash during this test.

Strata Vision View Note

When I added the 500 tracks, the default Strata Vision view looked dysfunctional because the track list could only list the first 50 or so tracks and, because I had set every track playing. the graph display was completely cluttered and functionality useless. However, if you turned off the track list and had a movie score where (say) only 50 to 100 tracks were typically playing at a time, the Strata Vision graph could be very useful.

Concluding Remarks

This experiment showed that Strata is perfectly capable of operating efficiently in a 500-track project on a high-end Macintosh computer.

For a 500-track project, Strata will consume approximately 4GB RAM and 5 CPUs.

While this is a good result, it does suggest that, at higher track counts, Strata could become problematic in relation to CPU. For example, at 1000 tracks, Strata would be consuming 8GB RAM (which is completely acceptable), but would be consuming about 10 CPUs of power. On a modern high end Macintosh with 20 performance cores, this would likely be unacceptable.

Strata is a new product, and there is plenty of scope to make it more CPU efficient. For example, in many of these large projects, most tracks are silent. If Strata were to detect that silence and stop processing those tracks, this would probably make Strata CPU scalable to many thousands of tracks.

Mmmmmmm! Cookies! This site uses cookies.