cheerful hamster
Hamsterdamus
This subject has come up a few times on the forum regarding the Ocean Way plugin failing to load when it appears that there are plenty of DSP resources available.
In the past, I have pointed out that each instance of Ocean Way takes 75.3% of PGM resources per SHARC, which limits the number of instances to just one per SHARC. My tests have now shown this to be incorrect. Here is a single Quad running eight instances of OWS:
One instance of OWS on a SHARC does indeed use 75.3% of the PGM, but when a second instance is loaded, the PGM usage only rises by another 12.9% to 88.2%. So contrary to what I've stated in the past, the UAD Instance Chart on their website is not wrong. You can run two instances of OWS per SHARC.
So what is happening when OWS instances fail to load? It appears to be the load balancer. According to the UAD Instance Chart, PGM increases are not linear, so there is no easy way to predict when an OWS will fail. Case in point, an instance of the Manley MP (25.5% PGM) can share a SHARC with the OWS (75.3% PGM) and combined they will use 96.7% PGM, when you'd expect it to be 100.8% and so one of them would fail to load. Seemingly some PGM resources are being shared even between different types (EQ and reverb in this example), but compiling a list of plugins that do this and in what circumstances would be tedious, and probably fruitless.
So suffice it to say, if it loads, great! If it doesn't, well...consult the Ouija board...
In the past, I have pointed out that each instance of Ocean Way takes 75.3% of PGM resources per SHARC, which limits the number of instances to just one per SHARC. My tests have now shown this to be incorrect. Here is a single Quad running eight instances of OWS:

One instance of OWS on a SHARC does indeed use 75.3% of the PGM, but when a second instance is loaded, the PGM usage only rises by another 12.9% to 88.2%. So contrary to what I've stated in the past, the UAD Instance Chart on their website is not wrong. You can run two instances of OWS per SHARC.
So what is happening when OWS instances fail to load? It appears to be the load balancer. According to the UAD Instance Chart, PGM increases are not linear, so there is no easy way to predict when an OWS will fail. Case in point, an instance of the Manley MP (25.5% PGM) can share a SHARC with the OWS (75.3% PGM) and combined they will use 96.7% PGM, when you'd expect it to be 100.8% and so one of them would fail to load. Seemingly some PGM resources are being shared even between different types (EQ and reverb in this example), but compiling a list of plugins that do this and in what circumstances would be tedious, and probably fruitless.
So suffice it to say, if it loads, great! If it doesn't, well...consult the Ouija board...