I've tested that patch too.
First of all, I normally don't use web UI, but now I've noticed that it suggests using patch number 0-40 and entering 0 gives the dreaded 0x60 error that people get here from time to time. As expected, the same happens when I specify SLOT=40 using CLI.
Next, the patch runs (on Magus) fine, but it can clip due to excessive buildup of reverb. This happens on output stage, so can be fixed with by attenuating outgoing data. Obviously it won't help on a device with line level inputs.
And the interesting part - rebuilding it using latest Faust with fast math enabled reduces CPU usage from 89% to 76%