I don't have priviledges to post in the appropriate forum so I am posting here.
I know a lot of people have been wondering why Studio One doesn't more effectively use all of the cores/threads available. The short answer here is that parallel processing of plugins within a single track really can't be processed using multiple threads. 1 thread processes 1 track basically from what I have seen in testing. But tracks can be processed in parallel, assuming you have the threads/cores for summing, which again is a serial process. This is not a Studio One/Presonus issue, this is just the nature of the multitrack digital audio processing. When you factor in aux sends for fx and other things it gets really complicated really fast and the best audio output will be obtained using the method employed in Studio One currently.
My test system is an AMD Ryzen 9 9950X (16 cores/32 threads), running at a 200W power limit with Curve Optimizer -20. The plugin used was the Relab 169 compressor, which is relatively CPU-intensive.
I used the same audio material and measured offline rendering speed as a multiple of real time. For example, 7× means a seven-minute recording would render in one minute.
Increasing the number of Relab instances on one track from one to two to four reduced rendering speed from 7.0× to 3.6× to 1.8×.
That's almost perfect inverse scaling, indicating that the processing times of the individual plugins are additive.
2. Independent tracks are processed concurrently.
Distributing four Relab instances across four separate tracks increased rendering speed from 1.8× to 5.0×.
This is a 2.8-fold improvement without changing the total number of plugins.
Clearly, Studio One is taking advantage of multiple cores when independent processing paths are available.
3. Doubling the number of equally demanding tracks did not increase rendering time.
This was the most revealing experiment.
I rendered two independent tracks, each containing four Relab instances in series.
Despite doubling the total number of compressors from four to eight, rendering speed remained at 1.8×.
This strongly suggests that Studio One processes independent track chains concurrently and that the longest processing chain can determine overall rendering time.
First: Single-threaded performance is paramount.
Plugins arranged in series create processing dependencies. Each plugin generally needs the output of the preceding plugin before it can process the signal.
Consequently, the performance of an individual CPU core can become the limiting factor, regardless of how many additional cores are available.
Second: Core count matters, but only to the extent that the project provides independent processing workloads.
Additional cores allow Studio One to process more independent tracks concurrently. However, subgroups and master processing introduce additional dependencies, so the useful number of cores isn't simply the total number of tracks plus buses.
For large projects with numerous CPU-intensive tracks, additional cores can be extremely beneficial. For smaller projects dominated by a few demanding plugin chains, faster individual cores may provide greater benefits.
These results don't establish exactly how Studio One schedules its threads internally, but they strongly support a model in which independent track processing is parallelized while plugins within an individual signal chain are processed serially.
The practical takeaway:
Prioritize strong single-core performance first, then choose enough cores to accommodate the number and complexity of independent processing paths in your typical projects.
This is particularly relevant when comparing modern high-core-count CPUs. A processor with more cores isn't necessarily faster at rendering a particular Studio One project, especially if the critical processing path is limited by single-core performance.
I know a lot of people have been wondering why Studio One doesn't more effectively use all of the cores/threads available. The short answer here is that parallel processing of plugins within a single track really can't be processed using multiple threads. 1 thread processes 1 track basically from what I have seen in testing. But tracks can be processed in parallel, assuming you have the threads/cores for summing, which again is a serial process. This is not a Studio One/Presonus issue, this is just the nature of the multitrack digital audio processing. When you factor in aux sends for fx and other things it gets really complicated really fast and the best audio output will be obtained using the method employed in Studio One currently.
Studio One Multithreading: Some Interesting Benchmark Results
I've been investigating how Studio One utilizes multiple CPU cores during offline audio rendering, particularly how plugin placement and signal routing affect performance.My test system is an AMD Ryzen 9 9950X (16 cores/32 threads), running at a 200W power limit with Curve Optimizer -20. The plugin used was the Relab 169 compressor, which is relatively CPU-intensive.
I used the same audio material and measured offline rendering speed as a multiple of real time. For example, 7× means a seven-minute recording would render in one minute.
Test Results
| Configuration | Rendering Speed |
|---|---|
| 1 track, 1 Relab 169 | 7.0× |
| 1 track, 2 Relab 169 instances in series | 3.6× |
| 1 track, 4 Relab 169 instances in series | 1.8× |
| 4 independent tracks, 1 Relab 169 each | 5.0× |
| 2 independent tracks, 4 Relab 169 instances each | 1.8× |
Observations
1. Plugin processing within a single track appears essentially serial.Increasing the number of Relab instances on one track from one to two to four reduced rendering speed from 7.0× to 3.6× to 1.8×.
That's almost perfect inverse scaling, indicating that the processing times of the individual plugins are additive.
2. Independent tracks are processed concurrently.
Distributing four Relab instances across four separate tracks increased rendering speed from 1.8× to 5.0×.
This is a 2.8-fold improvement without changing the total number of plugins.
Clearly, Studio One is taking advantage of multiple cores when independent processing paths are available.
3. Doubling the number of equally demanding tracks did not increase rendering time.
This was the most revealing experiment.
I rendered two independent tracks, each containing four Relab instances in series.
Despite doubling the total number of compressors from four to eight, rendering speed remained at 1.8×.
This strongly suggests that Studio One processes independent track chains concurrently and that the longest processing chain can determine overall rendering time.
What Does This Mean for CPU Selection?
My conclusion is that there are two primary considerations when selecting a CPU for Studio One.First: Single-threaded performance is paramount.
Plugins arranged in series create processing dependencies. Each plugin generally needs the output of the preceding plugin before it can process the signal.
Consequently, the performance of an individual CPU core can become the limiting factor, regardless of how many additional cores are available.
Second: Core count matters, but only to the extent that the project provides independent processing workloads.
Additional cores allow Studio One to process more independent tracks concurrently. However, subgroups and master processing introduce additional dependencies, so the useful number of cores isn't simply the total number of tracks plus buses.
For large projects with numerous CPU-intensive tracks, additional cores can be extremely beneficial. For smaller projects dominated by a few demanding plugin chains, faster individual cores may provide greater benefits.
Final Thoughts
Studio One is clearly multithreaded, but the degree of parallelism is constrained by the signal-processing dependencies inherent in an audio mix.These results don't establish exactly how Studio One schedules its threads internally, but they strongly support a model in which independent track processing is parallelized while plugins within an individual signal chain are processed serially.
The practical takeaway:
Prioritize strong single-core performance first, then choose enough cores to accommodate the number and complexity of independent processing paths in your typical projects.
This is particularly relevant when comparing modern high-core-count CPUs. A processor with more cores isn't necessarily faster at rendering a particular Studio One project, especially if the critical processing path is limited by single-core performance.