System Builder Marathon, August 2012: System Value Compared

Power And Efficiency

Economy and efficiency are often two different things. Efficiency is a measurement of work per watt. Intel’s marketing department doesn’t always appear to realize the difference; we often find the term efficiency thrown around in discussions of its weakest products.

In today’s example, Paul's $500 PC consumes a lot more than half of the power of Don's $1000 effort. And yet, we already know from the benchmark results that it's not a lot more than 50% as fast. Really diving down is going to require some math in Excel.

The $2000 machine’s stock power consumption doesn’t look terrible, but the amount of extra juice required for its overclock is simply scary. This huge increase in consumption occurred in spite of our efforts to keep its core voltage relatively moderate, at 1.32 V.

We used the $500 machine’s performance as our baseline and instantly spotted a problem, as the $1000 PC produced far more than twice its performance. My system's numbers are even higher, though the delta between its performance and Don's $1000 system is much smaller.

Keeping the $500 PC as our baseline, we moved the chart scale to better reflect efficiency differences, rather than absolutes. After all, there is no such thing as 100% efficiency.

Paul's $500 PC boasts low power consumption. But it's power use isn't as low as its performance relative to the other machines, unfortunately. In fact, even my overclocked $2000 PC produces 19% more work per watt, in spite of its 200% higher average power consumption.

The $500 machine’s poor performance is almost completely responsible for the $1000 PC’s huge efficiency victory, though we have to give credit to Don's system's Ivy Bridge-based processor.