I looked up some old 5850 vs 5870 clock for clock comps that people may find interesting ...
I think this gives a decent idea of what to expect with 670 vs 680, given that it's another case where the only difference between cards is the disabled cluster in the chip. In the case of the 5870, it has 11.1% more shaders active. In the case of the 680, it has 14.3% more shaders.
Note the results of the clock for clock tests between the 5850 and 5870 ... the green bar shows how far the 5850 is behind the 5870 at equal clocks. It's never more than 6.1%, and it's usually around 2-3% behind.
If the AMD example is illustrative of the Kepler scenario, some quick math suggests we're going to find that 680 is only going to be around 3%-4% faster CFC on average across a range of benchmarks ... an amount that could be easily overcome by a randomly chosen 670 card (that happens to have a very high boost clock ... like the OP's) on any one randomly chosen test.
I have to say ... although the 690 has really caught my fancy ... the idea of waiting for the inevitable 670 4GB model and throwing a couple of those in SLI for probably slightly < $$$ as the 690 is starting to sound really good
I think this gives a decent idea of what to expect with 670 vs 680, given that it's another case where the only difference between cards is the disabled cluster in the chip. In the case of the 5870, it has 11.1% more shaders active. In the case of the 680, it has 14.3% more shaders.
Note the results of the clock for clock tests between the 5850 and 5870 ... the green bar shows how far the 5850 is behind the 5870 at equal clocks. It's never more than 6.1%, and it's usually around 2-3% behind.
Quote:
The default Radeon HD 5850 is an average 12.7% slower than the Radeon HD 5870 in the low-quality modes and 14.1% slower in the high-quality modes. When the frequencies of the Radeon HD 5850 are increased to the level of the HD 5870 (850/4800MHz), the gap is smaller at 2% and 2.6%.
If the AMD example is illustrative of the Kepler scenario, some quick math suggests we're going to find that 680 is only going to be around 3%-4% faster CFC on average across a range of benchmarks ... an amount that could be easily overcome by a randomly chosen 670 card (that happens to have a very high boost clock ... like the OP's) on any one randomly chosen test.
I have to say ... although the 690 has really caught my fancy ... the idea of waiting for the inevitable 670 4GB model and throwing a couple of those in SLI for probably slightly < $$$ as the 690 is starting to sound really good
quoth the raven ... nevermore
(13 items) |
|
quoth the raven ... nevermore
(13 items) |
