Nevertheless, we all know from our previous reviews that maximum processor overclocking doesn’t always produce maximum results in performance tests. Namely, when we increase the base frequency to 188-190MHz we have to lower the memory frequency quite significantly, which negatively affects performance. That is why, overclocking to 181MHz base frequency seems to be the most optimal for majority of mainboards out there. We do lose in maximum CPU frequency, but we don’t have to raise the core voltage and have all Intel power-saving technologies up and running. Besides, 181MHz base clock suits perfectly for our Kingston HyperX KHX14900D3T1K3/3GX memory modules. In this case system memory can work at 1810MHz, which is exactly the value when we could lower CAS Latency to 8. At higher memory frequencies CAS Latency needs to be increased to 9.

Unfortunately, MSI Eclipse Plus mainboard cannot compete against Gigabyte GA-EX58-UD4P during system overclocking to 181MHz base clock. I won’t even provide the results, because I stopped testing half way through. Made simply no sense. It was a total defeat, because MSI Eclipse Plus would only remain stable at 176MHz base frequency without any core voltage adjustment. It meant that in this case the CPU worked at a lower frequency under any workload, especially under multi-threaded one, with lower memory frequency and worse timing settings. No advantage anymore, even no parity, only a pretty serious lag…

However, we could once again make the peculiarities of MSI Eclipse Plus mainboard work to our advantage, or at least eliminate the negative effects. Who said that we have to run the tests at 176MHz base frequency? While other mainboards partially disable Intel processor power-saving technologies when we increase the CPU Vcore, MSI Eclipse Plus doesn’t have this problem and these technologies stay up and running at all times. So, why don’t we push it up as far as we need, within reasonable limits, of course. As for the fact that the memory doesn’t work with CAS Latency 8 at higher frequencies, there is nothing we can do about it. It means we have to find the most optimal mode for our testing conditions. We need to find the base frequency, at which we won’t need to lower the memory frequency – just like during maximum CPU overclocking. Namely, we need to reach the maximum memory frequency at CL2, which we achieved at 185MHz base clock.

Of course, we will only be able to hit 3.9GHz CPU frequency under relatively low workload, because in all other cases the clock frequency multiplier will remain at 20x and the frequency will be only 3.7GHz.

And here is another proof of how the voltage in idle mode reduces proportionally to the initial increase. Without any initial processor core voltage adjustment, it lowered to 1.056V in idle mode. When we added 0.09V during overclocking to 190MHz base clock, it lowered to 1.152V. This time we increased the CPU Vcore by 0.05V and got an intermediate value of 1.104V. Everything matched perfectly.

Now that we have the most optimal overclocking settings for each mainboard, the comparison becomes more or less fair again. The CPU on Gigabyte GA-EX58-UD4P mainboard always works at 3.8GHz frequency, while on MSI Eclipse Plus – 100MHz higher or lower depending on the load. The memory on Gigabyte mainboard works at 1810MHz with 8-8-8-20-1T timings. On MSI mainboard the memory frequency is higher – 1850MHz, but so are the timings - 9-9-9-22-1T. So how will the two compare?

And the results turned out almost the same as before, when we ran the tests with maximum CPU overclocking. None of the boards can claim a convincing victory. In some tests MSI Eclipse Plus is faster due to higher processor and memory frequency, in some tests it yields to the competitor because of higher memory timings. There is no real winner, but most importantly, there is also no loser in this race. And I have to confess that at first I thought that MSI Eclipse Plus would stand no chances against Gigabyte solution, but luckily, I managed to find optimal settings, and take advantage of the board’s strengths. The power consumption tests also turned out very exciting, so let’s move on to them now.