Videocard: Acc7970 XL2420T @120Hz
Processor: firstname.lastname@example.orgGHz NH-D14
Mainboard: MSI Z68A-GD80[g3]
Memory: 4x4GB 1600MHz 9,9,9,24
Soundcard: Essence ST / AKG K-701
PSU: FSP Gold series 750W
| posts: 2,078 | Location: EU, CZ, Brno
No, but a lot of people are decreasing their gaming performance by overclocking memory as high as you without adjusting power target.
While GPU is capable to insert empty cycles to keep thermals down, memories are not.
So unless you have power target like +20% you are likely to reach this limit thanks to memory OC and GPU is idling even while keeping your desired clocks.
7950 does not actually need memory OC higher than 1500MHz (even 1375MHz is quite OK). For every 10% of memory OC you gain like 1~2% of performance but spend your TDP limit dearly.
Without Power target adjustment I got easily to point where increasing memory clocks by 25MHz (2% of base) was gradually decreasing performance and moving power target up got this performance back including those small benefits of memory OC.
HD7970 have basic power target between 205~217 what I've seen so multiply by 1.2 to see full limit.
You can read it and change it via VBE7.
As for timing, it helps mining because compute is about "Just in Time" operations with right kind of data. While rendering is "Pump it Up" about constant delivery of huge amount of data.
Now if you fine tune memory timing where mining gets most benefits and then you OC by 1MHz you may actually get to point where mining performance would be lowest possible on given GPU. Because OC itself changes time of delivery of each data request.
This is sole reason why not every HD7970 Card has same mining performance while being at similar/same clocks.