Why uses the HCC GPU APP 1 ATI-GPU AND 1 CPU. I had other GPU APPs thats uses oinly 0.1 CPU so I could use my CPU for other APPs.

But with HCC I can use only 7 and not 8 CPU oriented APPs.

In fact you can change this and have 0.5 CPU or less if you wish. For that you have to use an app_info.xml file to define how many CPU and GPU threads you want to have.
You have to tailor it to the card type you have.
HCC GPU WUs use still a good share of CPU. It is then important to have at least one full CPU core (simple or HT) for one GPU thread.

This app starts exclusively on the CPU core for the first 15-20 seconds. The task is then handed off to the GPU to do its' part and then is handed back to the CPU to finish. During the CPU part of the computation a full core is used although with the faster CPUs you can configure the app to use less of a CPU core so you can run multiple GPU tasks on 1 card.

When the HCC1 beta started the work units specified 1 GPU and 0.16 CPU. On my 8 thread i7 BOINC ran 8 CPU work units plus 1 GPU work unit. In windows normal threads ran with priority 8, the GPU work unit with priority 6 and the CPU work units with lowest priority of 4. I cannot remember why the 1 GPU + 1 CPU specification was introduced.

That's how it started, a fraction of a CPU, but then too many were failing, so the techs decided on reserving 1 CPU core as default. Whilst, you have app_info.xml and soon with a WCG endorsed BOINC client [something probably numbered 7.0.4x or higher], app_config.xml, which forward on will replace app_info/anonymous platform identification, in fact at some point in time outlaw app_info use at WCG [you can still use it for other projects, me thinketh]