Testbed and Methods

All practical tests and experiments were performed on the following test platform:

• Mainboard: Gigabyte MA790XT-UD4P (Socket AM3, AMD 790X/SB750, rev. 1.0; BIOS F2b, F3);
• CPU: AMD Phenom II X4 810 (Socket AM3, 2.6GHz, 200MHz base frequency, 4MB L3 cache, Deneb, rev. C2, 1.3V Vcore);
• Memory: 2 x 2048MB DDR3 Wintec AMPX PC3-12800, 3AXH1600C9-4096K, (1600MHz, CL9, 1.5-1.9V voltage);
• Graphics card: ATI Radeon HD 4870 512MB (RV770, 750/750/3600 MHz, 800 SP, 40 TMU, 16 ROP, 256-bit 512MB GDDR5);
• HDD: Samsung SP2504C (250GB, SATA II, 7200RPM, 8MB, rev. A);
• CPU Cooler: Cooler Master GeminII (120-mm Protechnic Electric MGA12012HB-O25 fan, 1500-2500RPM);
• Thermal interface: Noctua;
• PSU: Sunbeamtech NUUO Series SUNNU550-EUAP (550W);
• Case: Antec Skeleton.

We used Microsoft Windows Vista Ultimate SP1 x86 operating system and ATI Catalyst 9.2 graphics card driver.

Performance in Nominal Mode

The system worked in nominal mode perfectly fine. We can compare the performance of our platform with a similar one, featuring the recently tested Asus M4A78T-E. The mainboards are based on different chipsets, which are, nevertheless, very similar in functionality: Gigabyte MA790XT-UD4P is based on AMD 790X, while Asus M4A78T-E is based on AMD 790GX. Other than that both testbeds were configured absolutely identically. All BIOS settings were at defaults and hence were set by the boards themselves. The only exception was the AMD Cool’n’Quiet power-saving technology that for some reason was disabled on the Asus board, so we had to turn it on manually.

As we have expected, both systems show very similar results: one board is a little faster in some tests, another board – in other tests. Small differences are absolutely natural. The only more or less noticeable difference is in copy speed measured by Everest program, where Gigabyte mainboard falls far behind. Both mainboards set DDR3 memory at 1333MHz frequency with 7-7-7-20-1T timings by default. Asus board is probably setting more aggressive secondary timings, which delivers such an impressive advantage. However, this advantage doesn’t really show in the benchmarks we use for our test session, so you shouldn’t be too concerned about the significance of Gigabyte’s loss. It is just another proof of a well-known fact that the results of synthetic benchmarks do not always represent the actual state of things.