Fastest GPU miner

[Mushoz]

What is currently the fastest Miner/Kernel combo for 5850/5870 hardware?
And what is currently the fastest Miner/Kernel combo for 6950/6970 hardware?

I'm currently using Phoenix with the Phatk2 kernel, and I was wondering if maybe Diablo, CGMiner or maybe something entirely else would perform better. I'm currently not running into any problems, so the only reason for me to switch would be speed. I would test it myself if I could, but I'm not running my small mining cluster at home, but at a different location, so I can't currently run any tests myself. Nor do I know the best switches to use for miners other than Phoenix. I would be really grateful if someone has any information on this subject. Thank you very much in advance!

[superfastkyle]

Its hard to say.... I have on this system a 5850 and 5570. Using latest guiminer and this setting "-k phatk2 VECTORS BFI_INT FASTLOOP=false AGGRESSION=12 WORKSIZE=256" guiminer shows 499-507 mhash total. Using cgminer it shows 499-501 mhash. Its so close honestly it could just be a margin of error <1%. Honestly I would use cgminer if I had all the same cards in the system as it can setup overclocks, fan speed etc. Its very difficult to setup with several different model cards so I still use sapphire for overclocks, afterburner for temps, and guiminer for mining

edit: cgminer also seems to use less resources, have 100% cpu use on both, but still somehow is less laggy also uses less ram

[bcforum]

I don't know about the 6950, but I get ~450MH/s on each of my factory overclocked 6970's using phoenix with a custom kernel.

[ssateneth]

I've become quite accustomed to the user-friendly GUIMiner. Coupled with the latest phoenix and phatk2, I'm getting insane hash rates on my cards I have dedicated mining on. All cards @ 330 memory, 14 aggression, default worksize, vectors, and bfi_int unless specified.

5830 @ 970 core, 315.2 mhash, 40-70C (Outside computer, high variance based on weather)

5830 @ 1035 core, 336.7 mhash, 56C (Basement)
5830 @ 1015 core, 330.5 mhash, 53C (Basement)
5870 @ 1055 core, 489.0 mhash, 62C (Basement)

5830 @ 900 core 1000 memory, AGGRESSION=4 WORKSIZE=128 FASTLOOP=true, 243mhash, 70-75C (Living room PC. used sparingly for movies/browsing)

5830 @ 925 core 1075 memory, AGGRESSION=4 WORKSIZE=128 FASTLOOP=true, 270-283 mhash while browsing, 285 peak, ~200 while gaming, 70-75C (Gaming PC, outputs to display)
5870 @ 1040 core, 482.5 mhash, 61C (Gaming PC, dedicated mining card)

Not entirely sure why the major slowdown on the 5830 in living room; can crank up AGGRESSION but high responsiveness is needed in order to decode movies via DXVA since celeron 420 isn't strong enough to decode 1080p movies with just CPU power.

[iopq]

try setting the mem clock at 33-40% of the core clock and using worksize 256 and two-way vectors