Overclocking Specifics

We checked out overclocking potential of ASRock X58 Extreme mainboard following our standard procedure, however, it did prepare a few surprises for us. Let’s start with the fact that when the processor clock frequency multiplier is lowered to 12, the board could boot the operating system at the base frequency increased to 215 MHz, but we couldn’t pass the stability tests at this frequency despite all measures. AT 210 MHz frequency the board passed all stability tests with flying colors: it is a moderately good result, because most of the mainboards we tested so far, including ASRock X58 SuperComputer could remain stable at 215 MHz base frequency.

Although the information strings in the mainboard BIOS display the memory parameters recorded in the X.M.P. profiles absolutely correctly now, ASRock X58 Extreme refused to boot with the first profile. However, if we manually set 1867 MHz memory frequency and increased the voltages accordingly, the board copes perfectly fine with the high memory frequencies. This is a significant improvement compared with what we saw during our experiments with ASRock X58 SuperComputer. However, there still were a few things that slightly spoiled the joy from this achievement.

It took us a while before we managed to get the board to pass memory stability tests at 1867 MHz frequency. Increasing the memory timings and voltages didn’t help. In fact, it was at that particular moment that I noticed the peculiarities of ASRock X58 Extreme operation in Optimized mode. We didn’t really overclock the CPU, but simply increased the memory frequency of our Kingston HyperX DDR3-1866 KHX14900D3T1K3/3GX memory modules to their nominal value of 1867 MHz. That is why we left “Overclock Mode” parameter set to Optimized. However, the formerly safe minor increase in the CPU base frequency from 133 to 136.3 MHz this time caused the memory frequency to rise beyond 1900 MHz, and that is unattainably high for our memory modules. So, the key to success was not in finding the optimal increased voltage setting, but if making sure that “Overclock Mode” parameter is set to Auto. If they had included the description of the Optimized mode into the manual, I could have easily avoided all these complications.

The second problem revealed itself practically during the first ASRock X58 Extreme power-up, however, its consequences showed only at this point. I doubt that you have noticed it on your own, so I would like to draw your attention to the fact that we have slightly changed the configuration of our testbed. Instead of the Protechnic Electric MGA12012HB-O25 fan with 2500 RPM rotation speed that we have used on our Cooler Master GeminII CPU cooler so far, we started using Crown AGE12025F12J fan with maximum rotation speed of 2200 RPM.

This Protechnic Electric fan is a truly horrible device. It was initially installed into OCZ GameXStream OCZGXS700 power supply unit that we have in our system right now. However, many years ago we replaced it with a different fan, because this ill-balanced noisy fan created acoustically unbearable working conditions. It has been sitting idle in our lab since then until hot Intel Core i7 processors came out. It was noisy, but it was also very powerful at its 2500 RPM speed and coped perfectly with cooling overclocked processors. To ensure more or less acceptable working conditions, we had to switch it manually to 7 V power source, and eventually we got sick and tired of constant switching and unbearable noise. All contemporary mainboards can control the rotation speed of four-pin fans and we obviously need to have one.

120 mm fans are relatively quiet when working at <1000 RPM or slightly higher, that is why most PWM controlled fans support up to 1300 RPM rotation speed. It is fairly easy to find a 2000 RPM fan, but we wish we could get an even faster one. We did find it, but not in retail. It was a fan installed onto Vantec AeroFlow FX 120 (VAF-1225) processor cooler called Crown AGE12025F12J with maximum 2200 RPM rotation speed. It looks very beautiful. Its white plastic blades reflect the glow from the LEDs on the Antec Skeleton top fan, however, we are primarily interested in its performance and technical specs.

Since at that time we also had Asus rampage II Gene mainboard installed into our testbed, we tested this fan on that particular board for the very first time. The results after 25 runs of LinX utility on the Intel Core i7-920 processor overclocked to 3.8 GHz turned out quite good. With the old fan working at a constant speed of 2500 RPM the maximum temperatures of all four CPU cores were 80, 79, 78 and 75 °C respectively, while with the new fan supporting automatic rotation speed control and the maximum speed never exceeding 2200 RPM the temperatures were 81, 79, 79 and 75 °C.

Under heavy CPU load created by LinX utility, the fan rotation speed almost immediately increased to its maximum. Looks like Asus Rampage Gene II mainboard considers the cores temperatures, and most likely also the CPU current, to adjust the rotation speed of the CPU cooler fan accordingly. However, at first we had the impression that this rotation speed control feature didn’t work at all on ASRock X58 Extreme. We could change the fan rotation speed depending on the “Target Fan Speed” parameter setting, but it remained constant and didn’t increase any further when the CPU got loaded heavier and its temperature went up.

Things cleared up in a short while. The mainboard could in fact control the rotation speed of the CPU fan, we just had to wait a little longer, because it uses the thermal readings off the thermal diode beneath the processor socket. Depending on the conditions, such as room temperature, type and length of the CPU load, the difference between the actual temperature and the readings taken off the diode may drop down to 8 or increase up to 20 °C, but is about 15 °C on average. Under heavy load the processor core temperature increases almost momentarily, but the thermal diode warms up little by little. This increased delay between the temperature surge and the fan reaction makes it impossible to automatically adjust fan rotation speed during overclocking. Moreover, we couldn’t pass the tests even when the CPU wasn’t overclocked at all, but the memory frequency was increased to 1867 MHz. in this case we had to increase the UnCore voltage of the North bridge part integrated into the CPU, which increased the processor temperature and caused errors. If the fan rotation speed was at its maximum, the tests completed successfully. Unfortunately, it is only possible to use automatic fan rotation speed control on ASRock X58 Extreme mainboard in the nominal operational mode.

Our Intel Core i7-920 processor sample reaches optimal overclocking at 3.8 GHz frequency that can be achieved by increasing its base speed to 181 MHz. We usually don’t even have to bump up the processor core voltage, just make sure that it is protected against voltage drops under load. ASRock X58 Extreme mainboard has a new parameter called “ASRock VDrop Control”, but when we enabled it the processor core voltage increased slightly. We have already come across a situation like that before, and it can be resolved by disabling VDrop and increasing the processor Vcore a little bit to make up for the voltage drop under load. It is exactly the way we overcome the problem during our Asus Rampage Gene II tests: we only had to add 0.0125 V to the CPU core voltage. However, this fix didn’t work for ASRock X58 Extreme. We had to put up with excessive increase in the processor core voltage under load.

The last issue that we discovered during our overclocking experiments is problems with the memory. Most mainboards can ensure stable memory operation at 1810 MHz with 8-8-8-22-1T timings when their base frequency is increased to 181 MHz. ASRock X58 Extreme also conquered this frequency, but only with 9-9-9-24-1T timings.

As a result, Intel processor power-saving technologies did lower the CPU Vcore and multiplier in idle mode:

When the processor workload increased the clock frequency multiplier increased to 21 due to Intel Turbo Boost technology. In this case the CPU frequency rose to 3.8 GHz and the memory always remained at 1810 MHz with 9-9-9-24-1T timings. As you can see from the CPU-Z screenshot, under load processor Vcore is increased to 1.248 V (1.225 V being the nominal) because of the enabled “ASRock VDrop Control” function.

It is interesting that ASRock X58 SuperComputer mainboard demonstrated dynamic implementation of the Intel Turbo Boost technology. Processor clock multiplier increased to 21 only when the load was relatively low, otherwise it increased only to 20. ASRock X58 Extreme mainboard offers static implementation of Intel Turbo Boost technology: the CPU clock multiplier increases to 21 at all times, independent of the load level. As a result, we can sum up the results of our CPU overclocking on ASRock X58 Extreme mainboard as follows: