Conclusion
Last
summer we published a similar series of articles, where we were analyzing gaming
performance on the basis of two different processors. Back in the days it became
very clear that performance differences between two CPU's at high resolutions
are close to zero. The reason for this can be found within the fact, that the
processor isn't the bottleneck of a system, when you're playing games at high
resolutions. In this case it's the graphics card, which has to work overtime. In
case of lower resolutions the influence of the processor becomes clearly
visible, since the graphics card isn't the limiting factor anymore.
Having a closer look at the results we gathered while testing eight different
games and two benchmarks with a low and a high quality preset, we see that the
Core i5-3570K, with our "low-preset" is on average 0.7 percent faster than the
Core i5-4670K. Switching to our "high-preset" shows that the Core i5-3570K's
can maintain it's lead with 1.0 percent advantage. It's actually quite
surprising for us to see that the old Ivy Bridge CPU is quicker at the same
frequency than Haswell. Overclocking the Core i5-3570K to 4.5 GHz makes the performance
with our "low-preset" go up by 11.4 percent but when it comes to the high-preset
the increase in performance is only 1 percent. Regarding the Core i5-4670K the
situation is similar (10.9 percent "low-preset", 1.0 percent
"high-preset"). What's quite interesting to see is how the wattage of our test
system increased while maintaining 4.5 GHz stably. The 3570K needed 27.3 percent more power and in case of the
4670K the increase was a whopping 47 percent.
Should you be thinking about upgrading your Core i5-3570K setup with a Core
i5-4670K, we can tell you that this step can be skipped. It was really
surprising to see that the old Ivy Bridge architecture was faster than new
Haswell in our selection of games. Next to that, the performance increase remains
consistent even if the clock speed is raised by 1 GHz, from 3.5 GHz to 4.5 GHz.