Is Our Three-Way SLI-Based Setup Efficient? Can We Even Talk Value?
The concept behind this quarter’s high-end build was to use a larger ATX form factor to improve upon the previous machine in as many ways as possible, for the same money. Recently-expired discounts have pushed its price nearly 4% higher than when we placed our order, though new and remaining discounts should still make the machine a great overall value. For those who pay their own electric bills, efficiency is probably a greater disappointment than the price difference. Fortunately, we see some hope in a newer Ivy Bridge-E-based Core i7-4930K, manufactured on Intel's 22 nm node.
Sandy Bridge-E has always been a power hog, and a trio of graphics cards typically consumes more power than one dual-GPU card. Yet, I’m still surprised by the enormity of power consumption differences. Maybe I shouldn’t have been?
Remembering that the new machine was tested with Intel's slower RSTe storage driver in stock trim, and the faster Microsoft driver when overclocked, I ran an experiment with the stock configuration. The 20% gain from overclocking drops to around 15% when the new machine is tested at stock frequencies, but complemented by Microsoft's driver. Moreover, its stock-to-stock value rating jumps from 99.4% to 104.6%. Since the previous build lacked support for RSTe altogether, this quarter's effort becomes a better value at stock settings than its predecessor.
Overclocking is a different story. The previous machine was a supreme overclocker in spite of its small size, thanks to an elaborately-planned cooling system that made it the biggest mini-ITX machine we’d ever seen. The new build didn't overclock well at all. Part of this was the fault of a surprisingly-hot CPU, which somewhat soured our opinion of Sandy Bridge-E. Moreover, ASRock’s X79 Extreme6 didn't fare as well as we were hoping it would. I almost wish I paid $15 more for the lower-end Extreme4 and used a cheaper case to make up the price difference. There are good reasons why the X79 Extreme4 won a Tom's Hardware award and the Extreme6 did not.
And then there’s gaming. Overclocking provided the previous build a huge boost in average frames per second, while the current build barely sped up at all. If you don't overclock, you're going to love the 15% performance advantage at stock clock rates, while overclockers will be disappointed to further that advantage by a mere 7%. The previous build’s excellent overclock wins in spite of a far smaller form factor.