More information on CPU usage measurements on the Darkglass Anagram

This page reports how CPU usage measurements for all products released on the Anagram marketplace were obtained.

Please take these numbers with a grain of salt, as it is virtually impossible to arrive at definitive figures for a number of technical reasons. CPU usage:

  • May depend on parameter settings, as well as current and past input, in certain algorithms.
  • May also change with firmware updates, not to mention potential new hardware revisions of the Anagram.
  • Can be affected by other blocks being used.

It is also important to pay particular attention to worst-case (peak) values in this context, which may deviate substantially from average figures.

Measurement process

At the time of writing (April 26th, 2026), there is no standard measurement protocol, and neither Darkglass nor any marketplace vendor has publicly released such figures.

Nevertheless, the firmware used by developers offers special settings to check DSP usage directly on the device. It can display an on-screen overlay with these statistics.

The measurements were obtained by loading each tested block into an empty preset, leaving all parameters at their default values, and with no input jack attached. The average and maximum CPU usage values were then observed for a few dozen seconds. Finally, an "eyeballed average" and the observed maximum values were recorded.

To obtain the final results, the baseline average CPU usage — that is, the average CPU usage measured with no blocks in the processing chain — was subtracted.

All measurements were performed using KosmOS firmware version 1.14.0.

Results

This is the measured baseline CPU usage: avg 12.8%, max 17.4%.

Below are the numbers obtained for a small set of standard blocks, for reference.

Blockavg (%)max (%)
Harmonic Booster0.55.2
Gentle4.611.8
Bass Cabinet1.46.1
Entropia4.39.3
Ignissor1.36.1
Pitch Shifter1.34.9
Neural Amp4.911.8
Neural Pedal4.710.0
Neural Loader15.424.7

The numbers for each Clack block are reported on the respective product pages.

Final considerations

Please note that the testing conditions rely on special developer firmware, which may affect the measurements. Regular firmware would likely result in similar or lower CPU usage.