OK, I've spent all day trying to optimize this thing. I've got a 7950 on a machine with 4GB of system RAM, and I HAVE read the manual.

I'm trying to figure out the best clock speeds as per the example in SCRYPT-README.

 

• 1. Every time I change the mem-clock, the hashrate will go up slightly (about 15Kh/s), and then drop back down to where it was (275Kh/s). According to GPU-Z, the speed is staying where I set it, so why is the hashrate going back down???
• 2. To try and fix the above problem, I stopped using --auto-gpu. Was this the right thing to do? It didn't work anyway.
• 3. The README says to keep increasing the speed incrementally until the GPU crashes. I've gone as high as 1600 and no crash (no performance increase either). Looking at the mining comparison charts, I can see that not many people have gone above 1500. Should I keep pushing it? Is there anything to be gained?
• 4. Without OC'ing, I was able to get ~550Kh/s. Should I just be happy with that?

Batch file is

I'm tired and confused, but if you want to abuse me for missing something obvious, then please feel free