The high end 7xxx series won't be out until Christmas at least. Not enough confirmed details are known about it to make any judgement with regards to its performance and how it compares to what is available.
There's a bunch of mobile or low-power models coming out later this year built with the new process but they won't be performance oriented thus not as suitable for mining purposes.
I honestly think mining will have moved onto ASIC setups by the end of the year, GPUs becoming as obsolete as CPUs did. Especially with the way the difficulty is cranking up, like five or six levels in just the next two months.
Topic: What does the new Southern Isles architecture mean for mining? (Read 1469 times)
It's a die shrink from 40nm to 28nm using the same VLIW4 stream processors.
There is room for more transistors and the chip becomes smaller (less power consumption, runs cooler). Going solely by die size, a modest estimate would be at least 30% faster than current 6xxx series cards.
They could have been forced to make architecture changes as well due to competition; Nvidia is known to bring out the Kepler arch. later this year or early 2012. They boast 4x performance increase over Fermi which is massive (could make Vvidia cards viable for OpenCL mining).
Then again, Nvidia grossly overrepresented Fermi's capabilities as well. They said it was "revolutionary". GTX 480 was only 20-40% faster than their previous single-core flagship GTX 285.
There is room for more transistors and the chip becomes smaller (less power consumption, runs cooler). Going solely by die size, a modest estimate would be at least 30% faster than current 6xxx series cards.
They could have been forced to make architecture changes as well due to competition; Nvidia is known to bring out the Kepler arch. later this year or early 2012. They boast 4x performance increase over Fermi which is massive (could make Vvidia cards viable for OpenCL mining).
Then again, Nvidia grossly overrepresented Fermi's capabilities as well. They said it was "revolutionary". GTX 480 was only 20-40% faster than their previous single-core flagship GTX 285.
So I've read this article here about the new SI architecture, and to be honest I think I kinda get it, but I am not sure what this means as related to bitcoin mining. Could somebody who is familiar with the current architecture and opencl give a brief summary? It doesnt have to be a detailed analysis (though one is welcome), just something to let us know if this is paradigm changing or more 'meh'. I know it is impossible, for example, to say that a 7870 is going to be 80% faster without any hard numbers from AMD, but a general idea on whether the new architecture is an improvement for bitcoin or not would be good.
Jump to: