The $800 PC costs half as much as its $1600 predecessors, so I was hoping it would get half the graphics scores, yet it only produced 2/5 and 1/3 the 3DMark graphics scores of the general-purpose and gaming PCs, respectively. On the other hand, it produce 3/5 and 5/4’s the Physics scores of those same machines. And you thought there wouldn’t be any math.
The better news for the half-priced machine’s hard drive is that it produced slightly more than half the drive performance of its predecessors. It’s also nearly on par with the $1600 gaming machine in Sandra Arithmetic, falling only 45 percent behind the $1600 general-purpose PC that was designed to excel in CPU-bound metrics.
The $800 build starts off strong at our most basic Arma 3 settings, but quickly falls off as resolutions and details are increased. By the time we got to Ultra quality, I felt lucky to see the machine play smoothly at 1080p. Fortunately, my colleague’s inability to run 4800x900 forced me to remove that result from all averages.
The $800 PC actually plays through all the games and settings at 1920x1080, which means that it really is a low-end gaming machine in addition to my goal of making it a low-end workstation. But how well does it fill that primary role?
Audio encoding is the most consumer-oriented of our timed benchmarks, and the $800 machine’s Skylake processor beats competitors there. It falls behind the $1600 general-purpose PC in prosumer video encoding, since it has fewer cores, as well as professional applications from Adobe Creative Suite to 3ds Max.
I haven’t quite worked out why the new $800 build was slower in Blender and 3ds max than the $1600 Gaming PC, but I have a feeling it might have something to do with its slow hard drive. It also took a huge performance hit in 7-Zip file compression.
Power & Heat
Remembering that the power consumption measurements for every system comes from the wall and includes efficiency losses within the power supply, the $800 machine’s power unit would need to be 90 percent efficient for its processing components to have consumed a maximum 200 watts. It’s less than 90 percent efficienct though, so the output requirement of its power supply was actually less than 200W. Pushing less than 50% percent its capacity, the 400W part could have easily been replaced with a 300W unit, if significant costs savings could have been found there. On the other hand, $40 is a great price for a high-quality ATX power supply of any capacity.
Those power savings pay off big in the $800 machine’s efficiency rating. As much as 41 percent slower than the previous work-oriented machine, it still produced up to 21 percent more work-per-watt.