Articles: Mainboards
 

Bookmark and Share

(4) 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 ]

Testbed Configuration

We carried out our tests on a testbed that included the following components:

  • Mainboards:
    • Gigabyte GA-Z68X-UD3P-B3 v1.0 (LGA1155, Intel Z68 Express, BIOS ver. F4);
    • Gigabyte GA-Z68X-UD4-B3 v1.0 (LGA1155, Intel Z68 Express, BIOS ver. F6);
    • Gigabyte GA-Z68X-UD5-B3 v1.0 (LGA1155, Intel Z68 Express, BIOS ver. F6);
    • Gigabyte GA-Z68X-UD7-B3 v1.0 (LGA1155, Intel Z68 Express, BIOS ver. F6);
  • Intel Core i5-2500K CPU (3.3 GHz, Sandy Bridge, LGA1155);
  • 2 x 2048 MB DDR3 SDRAM Patriot Extreme Performance Viper II Sector 5 Series PC3-16000, PVV34G2000LLKB (2000 MHz, 8-8-8-24 timings, 1.65 V voltage);
  • MSI N570GTX-M2D12D5/OC graphics card (Nvidia GeForce GTX 570, GF110, 40 nm, 786/4200 MHz, 320-bit GDDR5 1280 MB);
  • Kingston SSD Now V+ Series (SNVP325-S2, 128 GB);
  • Scythe Mugen 2 Revision B (SCMG-2100) CPU cooler and an additional 80x80 mm fan for cooling the area around the CPU socket during overclocking experiments;
  • ARCTIC MX-2 thermal interface;
  • CoolerMaster RealPower M850 PSU (RS-850-ESBA);
  • Open testbed built using Antec Skeleton system case.

We used Microsoft Windows 7 Ultimate SP1 64 bit (Microsoft Windows, Version 6.1, Build 7601: Service Pack 1) operating system, Intel Chipset Software Installation Utility version 9.2.0.1025, Nvidia GeForce/ION Driver 266.58 graphics card driver.

Operational and Overclocking Specifics

I experimented with the mainboards in the same way as they are described in this review: from the junior to the senior models. This is partially due to the fact that I had most problems with the first mainboard, Gigabyte GA-Z68X-UD3P-B3, although I must confess that some of the problems were due to my own mistakes. First off, Q-Flash, the BIOS update utility integrated into the mainboard’s own BIOS, refused to see my flash drive with a new version of firmware, so I had to update the BIOS with the @BIOS tool. Afterwards I realized that my flash drive had NTFS formatting whereas Gigabyte’s Q-Flash is still incompatible with that file system. The BIOS code was successfully updated using a firmware file from the hard disk, but later on I found out that the @BIOS utility didn't work very well. It could not find a BIOS update on any of the numerous Gigabyte websites. @BIOS has a very polite interface but I'd prefer the utility to be crude yet working.

Then I got down to my tests and wasn’t surprised to see the mainboard being unable to make the CPU stable at 4.8 GHz, like most other mainboards do. Well, I had had some mainboards which could only overclock the CPU to 4.7 GHz, so I went on experimenting. I sensed a problem when the mainboard and CPU could not pass a stability check even at a CPU clock rate of 4.5 and 4.4 GHz, and I have to confess the problem was in me. Whenever we give you our advice on CPU overclocking, we always say that each step must be completed successively, without changing more than one parameter at once. Otherwise, it may be unclear what exactly parameter is the cause of a problem. This time around, I failed to stick to my own recommendations. Trying to save some time on a stability test, I eventually lost half a day beating my brains out over the question why the mainboard could not overclock the CPU.

So, the fact is that even before trying to overclock the CPU I had found out that the mainboard would not allow my memory modules to work at 1866 MHz. Therefore I had set the memory frequency at 1600 MHz and timings at 6-6-6-18-1T. Every mainboard I had tested before could work with such settings, but the Gigabyte GA-Z68X-UD3P-B3 became the first exception. So it was the memory modules rather than the CPU that prevented the system from passing the stability test. To ensure stability, I had to increase the memory timings to 7-7-7-20-1T and the CPU could then be overclocked to 4.7 GHz. The clock rate of 4.8 GHz still could not be conquered.

I had to content myself with the same, rather unsatisfactory, results when overclocking my Gigabyte GA-Z68X-UD4-B3. The odd thing was that I had to increase the CPU voltage somewhat more than on the previous mainboard, yet the resulting voltage turned out to be lower than expected. The CPU voltage was obviously set too high by the Gigabyte GA-Z68X-UD3P-B3.

It is the GA-Z68X-UD5-B3 that was most like Gigabyte’s earlier problem-free products. However, even this model could not overclock the CPU to 4.8 GHz and stopped at 4.7 GHz. The memory frequency was set at 1866 MHz, but, unlike the MSI P67A-GD80 (B3) mainboard which had been the first to prove the ability of my memory modules to work at 1866 MHz, the memory timings were set at 9-9-9-24-1T rather than 7-7-7-20-1T. You will see in the performance tests whether the increased memory clock rate can make up for the worse timings.

The Gigabyte GA-Z68X-UD7-B3 pleased me as it supported the memory frequency of 1866 MHz with 7-7-7-20-1T timings without any fine-tuning. But, like the rest of the new Gigabyte mainboards, it could only reach 4.7 GHz in terms of the CPU frequency.

Every mainboard was overclocked without disabling Intel’s power-saving technologies. When idle, the mainboards would lower the CPU voltage and reduce the CPU frequency by lowering the multiplier. To keep the number of screenshots to a minimum, I will show you one animated picture with four screenshots proving the functioning of those technologies:

Summing up the results of the Gigabyte mainboards in my overclocking tests, I should confess I am disappointed as none of them could clock the CPU at 4.8 GHz. And while the Gigabyte GA-Z68X-UD7-B3 should be given credit for clocking the memory modules at a high frequency with good timings, the GA-Z68X-UD3P-B3 and GA-Z68X-UD4-B3, unlike any other mainboard I’ve tested so far, are unable to work with normal timings even at a memory frequency of 1600 MHz.

By the way, each of these mainboards had a problem I hadn't seen before. The memory frequency would remain increased even if I restored the BIOS settings from a saved profile or by choosing the Load Optimized Defaults option. When this problem showed up occasionally, I had to manually set the memory frequency back to Auto. Unfortunately, this was not the single problem I had with these Gigabyte mainboards. Some more emerged during the performance tests.

It is a well-known fact that mainboards based on the same or related chipsets are usually similar in performance under identical conditions. However, when I launched SuperPi while benchmarking the Gigabyte mainboards in their default operation mode, I found them to deliver very low performance compared to same-class products. It was easy to find the cause of the problem because SuperPi is the only single-threaded application on my list. Intel’s Turbo Boost technology just didn’t work properly on at the default settings.

Let’s take a look at the Advanced CPU Core Features in the BIOS. The default clock rate of my Intel Core i5-2500K processor is 3.3 GHz. Easy to see, its clock rate will be increased to 3.4 GHz thanks to Intel’s Turbo Boost even when all of its four cores are under load. With three of the four cores loaded, the frequency multiplier will be increased to x35. With two loaded cores, it will be x36. The CPU will reach its maximum clock rate of 3.7 GHz when only one of its cores is under load.

However, the information in the BIOS of the Gigabyte mainboards is misleading. The CPU frequency multiplier is actually increased only to x34. Whatever the CPU load, the clock rate will be 3.4 GHz whereas other mainboards vary it from 3.4 to 3.7 GHz, which explains the lower performance of the Gigabyte ones. They can only rival others under high loads when all the four CPU cores are put to use. The problem is that when the C3/C6 State Support is set at Auto, the computer doesn’t switch into these deep power-saving modes and prevents Intel Turbo Boost from doing its best. Hence one more negative impression from the Gigabyte mainboards: they consume more power than others when the CPU is idle because they do not allow the CPU to use its power-saving features to the full extent.

As a matter of fact, this is a long-time problem that has plagued not only Gigabyte’s mainboards. Intel Turbo Boost was introduced with LGA1366 processors and was developed further in LGA1156 processors, but had to be enabled in the mainboard's BIOS at first. It would work only partially by default. This state of things has changed and modern LGA1366 mainboards enable Intel Turbo Boost by default without your having to tweak anything. The same goes for LGA1155 mainboards which support both power-saving features and Turbo Boost, but the new Gigabyte mainboards are an exception. Let’s now see how this affects their performance.

 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 ]

Discussion

Comments currently: 4
Discussion started: 07/06/11 10:59:41 AM
Latest comment: 09/14/11 01:59:15 AM

View comments

Add your Comment