Comments on the following newsitem http://einstein.phys.uwm.edu/all_news.php#186

Which is a joke: It uses 1 full CPU core and partially the GPU. But since CUDA task can't share the GPU it means the GPU is blocked for the as long as the CPU needs.
It would be have been better to say: Our science is not portable to GPU yet so we still only use CPU. Instead they throw out a CUDA application, to participate in the GPU hype, which gives you 20% faster WUs while using 200% of the resources. So actually the GPU is a anti-optimized application.