Articles: Mainboards

Bookmark and Share

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 ]

Manual Overclocking

So far we have been using Intel Core i7-3960X Extreme Edition processor for our mainboard tests. It wasn’t the best CPU, at least because of its recommended price of over $1000. Moreover, it was an engineering sample based on pre-production C0 core revision, while the processors that hit the stores were already C1. Therefore, as soon as C2 processors came out, we got our hands on as many as three Intel Core i7-3930K CPUs, which differed from the exact same C1 processors by SPOKY marking. The only significant difference between this CPU and the top model is 12 MB L3 cache instead of 15 MB and 100 MHz lower nominal clock frequency, which, however, doesn’t really matter since its clock frequency multiplier is unlocked anyway. However, the MSRP of this processor is almost half of what the top model costs, which is still a lot, but definitely more acceptable.

The nominal clock frequency of the Intel Core i7-3930K processor is 3.2 GHz, but it is just a formality. In reality, Intel Turbo Boost technology makes the CPU work at 3.5 GHz even under heavy loads, and if only one processor core is utilized, the frequency may increase to 3.8 GHz. All three processors proved capable of working at 4.6 GHz frequency, which is more than the previous maximum of 4.4 GHz achieved on the old Intel Core i7-3960X Extreme Edition. We were hoping to hit 4.7 GHz, but we couldn’t get our system to work stably at this frequency. So, we ended up selecting the CPU with the lowest nominal core voltage, overclocked it to 4.6 GHz on our Asus Rampage IV Formula mainboard, and ran our performance and power consumption tests after that.

For comparison purposes we needed to obtain similar results on some other mainboard, but we have suddenly come across unexpected problems. Asus mainboards traditionally compete against those from Gigabyte, but we couldn’t overclock our processor to the same 4.6 GHz on the previously tested Gigabyte GA-X79-UD7. We didn’t test the stability at lower frequencies, because we were looking to compare mainboards against each other in identical operational conditions. So, we decided not to waste any time and try the Asus P9X79 Deluxe instead, which made a very good impression on us during the review process and received our highly positive remarks. However, this board also was unable to overclock the new processor to 4.6 GHz frequency. Having assumed that this processor could only overclock well on an Asus Rampage IV Formula mainboards, we took a different CPU sample, because all three Intel Core i7-3930K processors we had on hand were capable of reaching 4.6 GHz frequency with our test board. Nevertheless, nothing changed and we could only go as far as 4.4 GHz.

Quite unexpectedly, Intel DX79SI mainboard saved the day. In the past we weren’t particularly happy with its not very high performance. Moreover, at that time it was unable to overclock our processor to the same heights as Asus P9X79 Deluxe and Gigabyte GA-X79-UD7. However, a new BIOS version for this mainboard came out in the end of January, which helped this mainboard to overclock the CPU to 4.6 GHz and also almost completely eliminated the performance lag, as we will see in the next chapter of our review. It is still occasionally unable to load the operating system and throws in a BSOD every now and then during system reboot shut down, but these could be the issues associated with our specific mainboard sample. In any case, only Intel DX79SI mainboard could replicate the same overclocking results, as on Asus Rampage IV Formula, while all other mainboards failed to do so.

I have to say that during numerous test sessions and multiple CPU and mainboard replacements it turned out that the Zalman CNPS12X cooler wasn’t well suited for frequent mounting and dismounting. One of the hexagon retention screws came completely loose, and we ended up replacing the cooler with Noctua NH-D14, so that some of the tests on Asus P9X79 Deluxe and the final round of tests on Intel DX79SI were performed with the Noctua cooler. This cooler turned out to be not as gigantic as it seemed from the photographs and reviews and is quite comparable to Zalman CNPS12X in size. It is a little longer, mostly because of the external fan, but not as wide, so it won’t interfere with the graphics card installed into the slot closest to the processor socket as frequently anymore. The heatsink fin array starts higher above the PCB, so there won’t be any conflict with the memory modules featuring tall heat-spreaders. Noctua NH-D14 cooler is much easier to install, with just a regular screwdriver instead of a special wrench, which can be held vertically instead of being held at an angle, which was necessary with Zalman CNPA12X. Therefore, we have no concerns about the ability of this cooler to survive multiple installations.

Two fans of the Noctua NH-D14 cooler consume a little less energy than three fans of the Zalman CNPS12X, which required additional power consumption tests to be run on Asus Rampage IV Formula. Moreover, we also had to check the system stability once again, to ensure that with the new cooler the mainboard was still ca[able of overclocking our processor to 4.6 GHz. In our “Super 5” roundup of LGA 2011 coolers the winners were Phanteks PH-TC14PE and Zalman CNPS12X, while Noctua NH-D14 SE2011 stayed only in the fourth place. In our case the maximum CPU temperature during overclocking to 4.6 GHz immediately dropped 6 degrees lower than with Zalman CNPS12X cooler (75°C vs. 81°C)! Of course, we didn’t closely monitor the differences in ambient temperature, so there could be some room for error, but it could under no circumstances be as high as 6°C. Besides, all temperature readings are relative, and we decided to give it another shot and try conquering the notorious 4.7 GHz once again. Unlike Zalman CNPS12X cooler, which we used in the beginning of our test session, we managed to succeed with Noctua NH-D14!

It is hard to tell what caused such a big difference. First of all, the contact between the base of Noctua NH-D14 cooler and the CPU heat-spreader is excellent over the entire surface of the CPU. Moreover, the lower central fan cools processor voltage regulator circuitry components much better than Zalman CNPS12X, so the CPU doesn’t receive additional heat from the neighbors. We used a regular Noctua NH-D14 with a separately purchased retention kit for LGA 2011 so that we could also use it on other platforms when necessary, because the Noctua NH-D14 SE2011 modification is deg=signed exclusively for LGA 2011 processors.

We always overclock mainboards in such a way that they could be used for a prolonged period of time in this mode. We do not try to make our life easier by disabling any of the mainboard features, such as onboard controllers, for example. We also try to keep the CPU's power-saving technologies up and running normally to the best of our ability. And this time all power-saving technologies remained up and running even in overclocked mode lowering the CPU voltage and frequency multiplier in idle mode.

Of course, as soon as we achieved stability at 4.7 GHz CPU frequency under Noctua NH-D14 cooler, all our hopes for a fair comparison in identical operational conditions vanished. Well, looks like Asus Rampage IV Formula simply doesn’t have worthy competitors in overclocked mode at this point. To be more exact, they could be out there, but we haven’t met them yet.

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 ]


Comments currently: 6
Discussion started: 02/29/12 06:10:08 AM
Latest comment: 01/10/14 07:15:52 AM

View comments

Add your Comment