Looks about right to me. CPU's are very slow at hashing compared to GPU's. If it recieves a Long-poll on the network telling it a block has been solved before it has completed the current share it will cause it to drop that share and any that are in the queue and ask the pool for more work. Your stale submissions are around 1.9% which is acceptable. You should consider utilising your GPU if possible it will be a lot more efficient.

The -g flag set the number of gpu threads.

For just my i7 I would use.

cgminer -o http://127.0.0.1:8337 -u gpu1 -p x -a 4way -t 7 -g 0

edit - I see you already set -g. Try either -a 4way or -a c

This was the second pool I mined on. Was a very stable pool unfortunately PPLNS isnt for me as I am not able to mine 24/7. If I was I would definitely go back to them.