UAD-Plugins can be teleported without any problems, simply install the wrappers just like you always do. The only problem with using it in Cubase SX3 is: The UAD-1-extra-bus-latency is not added to the Wrapper-Plugin-Latency, so the network-latency is correctly compensated, but the signal is still 2048 samples (or whatever is your UAD-buffer-setting AND depending on the plugin) too late.
Solution for this: Get the FREE \"Latency Delay\" Plugin on
http://www.voxengo.com, insert it right after any UAD-Plugin and adjust the value as desired. Phase Inversion on a second parallel signal-stream helps you to identify the correct samples-value. Of course this works for every plugin which reports a wrong latency-value or simply doesn´t report any latency, not only for UAD-Plugins.
It works flawlessly everywhere in Cubase (Inserts, Groups, FX-Channels).
Its also possible to run native plugins simultaneously on the FXT-Server. I tested it with some Hydratones and a few 1176SEs on the slave machine. The network packaging for the UAD-Plugins will consume quite a bit of the Slave´s CPU (as with any other plugin) as opposed to running it directly in your master-machine.
Problems:
- I could only teleport 12 1176SE-Instances in my 100Mbit-Network (Marvell-GBit-Adapter on Master 3,2 Ghz Prescott, NE2000 noname pci on slave 800 Mhz), regardless if mono or stereo. The network-load was about 80%, the UAD had still free ressources and the CPU-meter on the slave was at about 70% but the network-dropouts started and the server became unstable.
- Creating any new teleported plugin lasts 12 seconds. This is totally annoying. Whenever opening a Project with teleported plugins, EACH plugin needs 12 seconds. This can be VERY long. As soon as it´s been created, its fast and totally transparent, as if running on the master machine.
- Dont know what it´s like on 96 or 192 KHz. I only tested it with 44,1.
- Though it works in ProTools via the FXPansion Wrapper, I dont how PT TDM´s Plugin Delay Compensation behaves with the FXPansion-Wrapper-Buffer/FXT-Latency/UAD-Latency-construction. However, it´s working fine in LE with manual delay compensation (*sigh*).
It´s not a full replacement for a Magma-Chassis yet, but if you´re mostly using heavy plugins like the neve1073, the network-bottleneck will not be an issue for you, especially if you´re working with a few slaves.
I will repeat the test with a pair of Intel-GBit-Adapters and I hope desperately that the quite big teleporting-overhead on master and slave will be at least half as much - that would be awesome. And I hope that instantiating a plugin will also last half as long.