Articles: Mainboards

Bookmark and Share

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

Operational and Overclocking Specifics

If we disregard the horrible contact between the CPU and the cooler, which is not the mainboard’s fault at all, assembling a system on Asus P9X79 Deluxe was easy and intuitive. First of all we decided to check out the new “USB BIOS Flashback” tool. In fact, it isn’t that new, because we have already come across this function in our Asus Maximus IV Extreme review, but at that time it didn’t really impress us. The major reason for that was the fact that it could remain exclusive only for “Republic of Gamers” mainboard series. However, now we see that this function has also become available on the regular Asus mainboards. The second reason may strike you as a little odd, but only at first glance: at that time we haven’t yet started reviewing AMD Bulldozer processors.

The thing is that our tests of all Socket AM3+ mainboards with only one single exception started in the same manner. We assembled our testbed, confirmed that the board didn’t start with the new Bulldozer processor, took the system apart, assembled the same exact system with an old Socket AM3 processor, reflashed the BIOS with the latest available version, then took the system apart again and put it back together with a Bulldozer inside. Only after all these manipulations we could actually start the tests. We didn’t really stress this aspect in our reviews, because, obviously, all Socket AM3+ mainboards would eventually have the BIOS supporting new processors. However, the owners of the first Bulldozer processors have most likely faced the same issues. They had to look for the old CPUs or contact the technical support services immediately after purchasing the mainboard, so that they could receive a BIOS update. The not very well prepared launch didn’t add any charm to the new AMD processors, but on the contrary created unnecessary complications for the users. However, if “USB BIOS Flashback” technology had been available at that time, we wouldn’t have even noticed this problem.

Theoretically, there shouldn’t be any problems with updating the BIOS like that. First of all you need to copy the new BIOS version onto a USB flash drive and rename the file in accordance with the mainboard model name. In particular, it should be called P9X79D.ROM for Asus P9X79 Deluxe mainboard. Now we need to connect the drive to a white USB 2.0 port on the back panel, press and hold the button for three seconds, and after that the reflashing process should start, which will be indicated by the blinking of the button. The BIOS update should be completed when the button light goes out. In our case, when we pressed the button it started to glow, but didn’t blink and the BIOS wasn’t being updated.

I have to say that we received a mainboard with an early BIOS version 0604, while the official Asus web-site had a startup BIOS version 0709 and the latest version 0802 available. Assuming that the initial BIOS version doesn’t support “USB BIOS Flashback” technology, we updated the BIOS to version 0709 in regular manner and then tried to use this technology again, but without any luck. The manual says that if the button stops blinking and is lit up permanently, it indicates that the technology is not working. There are two possible explanations to that. First, the drive may be connected improperly. However, when we reflashed the BISO in a traditional way the board saw it just fine and everything worked. A second possible explanation is the incorrect name of the BIOS file. However, we tried P9X79D.ROM and p9x79d.rom, but the technology still didn’t work. In the end we reflashed the BIOS using a built-in “EZ Flash 2” utility, and proceeded to the actual performance tests.

The nominal frequency of the Intel Core i7-3960X Extreme Edition processor is 3.3 GHz, but we never saw this particular setting. Due to Intel Turbo Boost technology, the CPU frequency was at 3.6 GHz most of the time, reaching 3.9 GHz under lower operational load, when only one out of six processor cores was utilized. In idle mode processor power-saving technologies dropped its clock speed to 1.2 GHz. The first issue surfaced upon mainboard boot-up: when we overclocked our processors by raising its clock frequency multiplier the board still displayed its nominal frequency instead of the actual frequency at the particular instant.

We uncovered the next issue attempting to set the memory parameters recorded in the modules XMP. In fact, there were no problems with the settings related directly to the memory, as the mainboard set the frequency, timings and voltages absolutely correctly. However, after that the processor frequency started growing to 3.9 GHz under any load, so we can’t really say that the board works flawlessly with the XMP profiles, because as we can see, Intel Turbo Boost technology gets messed up.

However, we do prevent the processor clock frequency multiplier from increasing beyond a certain set value during overclocking, so this issue will hardly become a serious problem for us. However, at first we decided to check out the automatic overclocking abilities of the new Asus mainboard. In this case we had to select “OC Tuner” in the BIOS or simply flip the TPU switch on the mainboard itself. In this case the mainboard raised its base clock from 100 to 126 MHz. Together with the x34 clock frequency multiplier it increased the processor frequency to 4.3 GHz. The memory frequency rose from 1333 to 1685 MHz and we used the 9-10-9-27-2T timings recorded in the modules XMP profile.

This is not a very impressive result, especially keeping in mind that power-saving technologies have become partially non-operational. In idle mode the clock multiplier kept dropping to x12, but the CPU Vcore remained excessively high.

There is no automatic overclocking technology out there that could allow the user to achieve the same success as with careful manual selection of the optimal settings. At first we were shooting for 4.6 GHz frequency, which could be achieved by increasing the CPU Vcore by 0.09 V with enabled Load-Line Calibration technology counteracting the processor voltage drop under heavy load. However, under heavy load created by LinX utility with AVX instructions support it turned out that the CPU temperature was approaching 90°C causing throttling to kick in and performance to drop dramatically. Therefore, we decided to focus on 4.5 GHz frequency and found the minimal core voltage increase of +0.05 V, which allowed us to successfully pass LinX tests with the temperature remaining below 80°C. However, all our hopes were shattered when we launched Prime95 test. In these conditions the utility could start normally, but then we would see the blue screen of death within the first few seconds into the test. Trying to achieve stability we pushed the core voltage even higher increasing it by up to +0.08 V. As we remember, higher voltage would cause the temperature to increase excessively. So, in the end we had to put up with our best results of 4.4 GHz achieved by raising the Vcore by +0.07 V.

Despite our initial expectations, even after we increased the core voltage from 1.5 V to 1.65 V, the memory refused to work at any frequency above the nominal 1866 MHz. we also failed to lower the pretty high timings of 9-10-9-27. We could only change the 9-10-9-27-2T timings set when we selected the XMP profile to 9-10-9-27-1T and the memory voltage didn’t have to be increased in this case. Power-saving technologies remained up and running lowering the processor Vcore and clock multiplier in idle mode.

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


Comments currently: 2
Discussion started: 01/03/12 06:21:07 PM
Latest comment: 03/08/16 03:54:55 AM

View comments

Add your Comment