We carried out our tests on a testbed that included the following components:
- Mainboard: Gigabyte GA-990FXA-UD5 rev.1.0 (BIOS version F7e);
- AMD FX-8150 CPU (3.6 GHz, Zambezi (Bulldozer), AM3+);
- 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 of 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, Nvidia GeForce Driver 280.26 graphics card driver.
Operational and Overclocking Specifics
We didn’t experience any difficulties with the assembly of the Gigabyte GA-990FXA-UD5 based system. Unlike some other mainboards, for example, the same flagship GA-990FXA-UD7, there won’t be any problems during the system disassembly as well: the first PCI Express 2.0 x16 slot is far enough from the processor socket and the graphics card can be put in and taken out easily. The system worked fine in nominal mode, all power-saving technologies were up and running and lowered the processor clock frequency multiplier and core voltage in idle mode reducing the system power consumption. Turbo technology (CPB – Core Performance Boost) also worked fine and increased the CPU frequency in case of partial and even full core utilization. However, just like the top model this board set the memory timings to 7-7-7-20-2T instead of 7-7-7-20-1T, as it was supposed to. You can see from the BIOS screenshots that by default the mainboard has the “Native IDE” mode on. The same happens on Gigabyte mainboards for Intel processors, although they offer you a convenient way to automatically switch the drives into a more advanced AHCI mode during the first system boot-up. We didn’t see anything like that on the Socket AM3+ mainboard. It was frustrating that Clearing CMOS also reset the date and time. Moreover, the mainboard occasionally refused to let us access the BIOS ignoring my vigorous attacks on the Del key during startup. Although I have to admit that it occurred much rarer than on the flagship mainboard.
I couldn’t locate the “HPC Mode” parameter in the mainboard BIOS, which is supposed to prevent the CPU frequency from dropping to 3.3 GHz under heavy load. However, the beta versions of the F7 BIOS have an “AMD APM Master Mode” parameter in the “Advanced BIOS Features” section, which can serve the same purpose. If you disable it, Turbo technology will stop working, so it doesn’t suit too well for prevention of the frequency drop in the nominal mode, unlike the “HPC Mode” option. But it will do just fine during overclocking, because in this case we disable the Turbo technology anyway. Unfortunately, we couldn’t overclock our processor to its maximum of 4.5 GHz. We could maintain system stability even at 4.4 GHz, so we had to stop at 4.3 GHz. However, there is also good news: it is for the first time that a Gigabyte mainboard let our memory modules work at 1867 MHz frequency instead of lowering it to 1600 MHz as we would normally have to do. The board raises the back clock quite noticeably, so all other frequencies connected with it also were a little higher than normal, such as the CPU clock, HyperTransport and memory frequencies.
We always overclock mainboards in such a way that they could be used permanently 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 features up and running normally. And this time all power-saving technologies remained up and running even in overclocked mode lowering the CPU voltage and frequency multiplier in idle state.
CPUID window of the AIDA64 suite doesn’t display one more system parameter, which also affects the performance level – the North Bridge frequency. Until today all mainboards allowed us to increase this frequency during overclocking from the nominal 2200 MHz to 2400 MHz without any special effort on our part, even without any voltage increase. Unfortunately, Gigabyte GA-990FXA-UD5 mainboard is the first one to have failed this task, so we had to leave the North Bridge frequency at its default value.
I have to say that we were paying more attention to Asus mainboards not only because they overclocked Bulldozer processors best of all, but also because most of the other manufacturers’ products were simply not ready for overclocking these new processor at all. BY the time new Socket AM3+ processors launched, only the BIOS of Asus and Micro-Star mainboards contained the “HPC Mode” parameter preventing the processor clock frequency from dropping under heavy load. A month later, there were still no BIOS updates from other mainboard makers. That is why we had to specifically request an update from Biostar for our TA990FXE review, we also had to request an update from Gigabyte in order to complete our GA-990FXA-UD5 testing. As a result, we got the opportunity to testdrive an F7e BIOS version and we performed all tests with it. However at the time we were writing this review, which was about a month and a half after the official CPU launch, Gigabyte updated their Socket AM3+ mainboard BIOSes and released F7d version, which was a few days younger than the F7e version we tested with. Of course, we couldn’t resist the temptation to check it out, too.
We didn’t uncover any serious differences between the version F7e and F7d. The board was still unable to overclock our processor to 4.5 GHz or at least to 4.4 GHz; it still set the default memory timings to 7-7-7-20-2T. The only noticeable difference was the persistent denial of the BIOS access, which drove me nuts during the earlier tests of the flagship mainboard. I had to reboot multiple times, because the board ignored my banging on the Del key and wouldn’t let me access the BIOS. Since the obtained results were practically identical with both versions, we didn’t rerun the performance and power consumption tests.