they are short-sized preamps. For them you need to wait the big cuda implementation is completed. This release is the porting of the old code to cuda 4.0 It works way better, but it's limited to reverb presets and long-tailed kernels (at least 150 ms, possibly the same length for all kernels).In such case you save a lot of cpu cycles. I tested it in our new benchmark pc (i7, quadro fx 5800), and CPU is close to 0% when you load for example EMT reverbs, while GPU is very low (around 5% for each instance!)
we are exactly in the middle. We are re-writing from the scratch the vector engine, adding the CUDA processing option. Each time you download a new release you help us in testing that everything works as expected (we are modifying little sections of code, releasing a new version along with other enhancements).
We provided a cuda 4.0 cudabridge for pc and we are still working on the mac osx side too (there were major updates in this 4.0 release, so a lot of things changed and we are fixing things).
What I should do for run Nebula under CUDA? 1. Have a CUDA GPU. 2. Install a CUDA enable GPU driver and CUDA Toolkit version 3.00 (please use C:\ not C:\Program files in PC/Windows OS). 3. Install Acustica Audio CUDA Bridge version 1.1.036. 4. Copy the "Nebula3 Reverb.dll" file and rename it "Nebula3 CUDA.dll" or choose a name that you like. 5. Copy the "Nebula3 Reverb.xml" and rename it as above (e.g. "Nebula3 CUDA.xml") 6. Load Nebula3 CUDA and go to the MAST page and set the FREQD to mode 15 (CUDA 2ch Mono), then click in SAVE.