GPU vs CPU performance

I`m using AMD HD 7970 with old CPU (2600K), if I`m trying to crack WPA one of CPU cores is at 100%... and I`m getting 85.000 H/s but after OC 2600K to 4GHz GPU speedup to 92.000 H/s still using only one core at 100%. I cant OC more without better colling... in fact I dont want to OC my CPU. How can I use all core ?
hashcat uses all cores by default, whats your command line?

let me rephrase that:

hashcat uses 8 threads by default which will use all cores unless you have more than 8.
oclhashcat+ -m 2500 --outfile-format=3 -a 0 --status --status-timer 5 --gpu-temp-abort 85 --gpu-temp-retain=69 --outfile /root/output.txt /root/test.hccap /root/dictionary.txt

I tryed --cpu-affinity=1,2,3,4 but always it uses one core.

This is at stock frequency...

Attached Files Thumbnail(s)
Seems it's always some weird results from The-Distribution-Which-Does-Not-Handle-OpenCL-Well (Kali) users. I can go on full speed @2xHD6990 on weak AMD E350 APU. But it's on ubuntu with properly installed drivers. And also you are not using any rules, just single dictionary which should be not a problem for your cpu.
I`m using catalyst 13.4 from Debian-sid repo, (The-Distribution-Which-Does-Not-Handle-OpenCL-Well (Kali) have 12.10 or older) but on Debian sid I have same problem. Now I`m trying to install Ubuntu 13.10 and catalyst 13.10beta... mayby it will help.

I tryed to install binary driver from amd on The-Distribution-Which-Does-Not-Handle-OpenCL-Well (Kali) and Debian but --buildpkg always crash, same as binary installer directly.