Test System

Initial Setup

The initial setup proved simple and easy, with the BIOS automatically setting a default speed of 3.5 GHz under Turbo mode, and defaulting the memory to JEDEC standard 1333 MHz CAS 9. We ran several rounds of Memtest x86+ to verify stability, and then installed Windows 7 64-bit. We then enabled the 2133 MHz XMP profile for our Corsair memory, and ran Memtest x86+ again. While Memtest x86+ proved to pass with flying colors, we found that when in Windows, we encountered some stability issues, which were remedied by increasing memory controller voltage from a default 1.05 V to 1.124 V. We did have to manually set secondary timings as well, using a tool provided on the driver disc, which we will look at in some detail in the "Overclocking" section. After several hours of stability testing with various programs over the course of two days, we found sweet spots for both CPU and memory speeds when overclocked, and then reset the BIOS to default speeds and JEDEC timings, in preparation to begin our usage and performance testing.

PWM Power Consumption

Since one of our first tasks was to truly verify system stability, while doing so, we measure CPU power consumption. We isolate the power coming through the 8-pin ATX connector using an in-line meter that provides voltage and current readings, as well as total wattage passed through it. While this may not prove to isolate the CPU power draw in all instances, it does serve as a good indicator of board efficiency and effective VRM design.


The ECS P67H2-A2 really surprised us on the power consumption front. Our "B2" stepping board proved to have least efficient VRM design at both idle and load, but the new "B3" product managed to lower idle numbers, while load numbers did increase a little bit. With idle numbers near twice that of the other tested products, we were left unimpressed, however load numbers do fall in line on the outside edge of efficiency. This difference is quite expected though, as there are not many options to disable phases at idle, nor is there a power-saving mode available for the VRM, so while unimpressive, they are clearly understood. Our Gigabyte test board did show the same power usage numbers, while using quite a few more phases in the VRM, so while the numbers were unexpected, they are definitely within acceptable values.