System Builder Marathon Q1 2016: $662 Budget PC

Test Setup & Benchmarks

Test Bench Hardware

The Redbone will be going up against the Intel Munchkin from last quarter as well as my predecessor's Q1 machine from last year. Paul's old build used a Haswell i3 and came in about the same price as the current Redbone, so it should be a close completion in many regards.

Benchmark Suite

Similar to my motherboard reviews, I set this machine to stock clocks and enabled Intel Speed Step and other energy saving features. I use Windows default "Performance" power option preset for everything except idle power consumption where it's set to "Balanced." One difference is that the stock runs enable the memory XMP where available. Another is that the CPU fan is set to standard speed.

Synthetics

Image 1 of 2

The 2015 and Munchkin both have stronger graphic cards, and so they take the lead in graphical tests. However Skylake demonstrates superior processing capacity with higher physics scores.

Image 1 of 3

Gaming

Image 1 of 8

We see interesting trends in the gaming benchmarks. At lower resolutions and details settings, situations where the CPU faces the brunt of the work, the 2016 machine often takes a strong lead. Arma is very CPU dependent and Grid loves memory bandwidth, and with both Skylake does better than Haswell. Turn up the details, or switch to Far Cry or Battlefield, which emphasizes graphics hardware more, and the GTX 950's limitations are shown. Though despite costing considerably less money, the overclocked 950 delivers playable experiences across the board at 1080p.

Applications

Image 1 of 4

All three machines have 2C/4T processors at similar speeds so it's up to architecture to be the differentiator. Skylake asserts its efficiency over Haswell with small victories across the board.

Power & Temperature

Image 1 of 2

While the 6100 is draws slightly less power than the 4170 or 4150, the big difference in power consumption is the GPU. Temperature-wise, the 200T keeps the 6100 just above ambient at idle. Moving the case fan over the GPU also lets it stay cooler, even when overclocked.

Overall Performance & Efficiency

The 2016 Skylake leads in all categories but two: storage and gaming. This is perfectly expected considering the less powerful GPU and spindle hard drive. Once overclocked, it leads by double-digits.

The Skylake build doesn't always lead in pure performance, but uses far less power proportionally speaking to the other machines, yielding it a big efficiency win.

System Value

Image 1 of 3

The 2016 Skylake comes in second for total performance when we calculate all the results together. However it did so while costing less than either of its competitors. It doesn't matter if you count the whole system, just the hardware, or the platform components alone, the 2016 build comes away with the value crown. While the 2 percent CPU overclock isn't a lot, combined with tuned RAM and graphics, means a value bonus of 20 percent over the other entries.

Final Word

Last quarter I pledged that I'd do a simple build this time after killing myself on two machines. And like so many enthusiasts I let myself get sucked back into the game. Why do we do this to ourselves? Are we gluttons for punishment? Is the siren song of low-cost / high performance machines truly irresistible? Or have we somehow convinced ourselves there is a free lunch out there?

Overall, this build did exactly what it was expected to, just not in the way it was expected. By paring back unnecessary costs and parts, we have a system that performs well in every facet yet is still reasonably priced. I must of course ask myself if I could do better.

First, this machine needs flash storage to be truly considered general purpose. I explained my reasoning why I skipped a SSD in this quarter's build. But anyone using this as a guide to build their own machine, I implore you to spend a little extra to get at least a 120GB SSD as a boot drive. It simply deserves it. If you need to, cut back the GPU to a 750 Ti. An SSD simply makes a computer feel faster and more responsive in nearly every aspect. Unless you have an extremely tight and inflexible budget, I recommend a SSD in every new computer.

Second, the overclock wasn't a complete failure, but it wasn't nearly as significant as I intended. I simply hit too many questions and obstacles to sort through in the brief time I had to tune this machine. But even with more time, I doubt I could improve much upon this machine as presently configured. The unstable memory at high BCLK makes this only half a machine and losing AVX defangs the i3 with its limited cores. Fixing this build in this regard likely means a more expensive motherboard at the very least. And of course that leads us to a familiar topic of whether it's worth spending the extra money for extra performance.

As I talked about last quarter, whether it's paying extra money for parts or spending your time and effort on tuning them (usually a combination of both), you will always spend more on an overclocked machine than one at stock clocks. Some of my colleagues talk about "leaving performance on the table." Truth be told, I've never liked that phrase. I don't find it to be true. It only takes one piece of the puzzle to bring the whole thing to a halt. In my experience that piece tends to be the motherboard.

Now the Pro4S seems like a fine board for the price to me. I didn't have time to give it a thorough review, but initial examination marks it as one that would do well in a budget overclocking build with an unlocked CPU. Remember that options available in a motherboard's UEFI are almost always dependent on the CPU used. Not all chips allow full control of every multiplier, every clock and every voltage. Pricier boards also tend to offer more settings and more control when overclocking.

My experiments with overclocking on a tight budget have consistently shown the very settings that would save a failed overclock are those that were hidden due to not spending more on parts. That SuperMicro board people are talking about goes for at least $130, over 50 percent more than I spent on mine. Many others setting records with i3 chips are doing so on premium Z170 boards. Had I purchased the MSI Z170-A in time, perhaps this machine would have done better. Had I purchased any manufacturer's top-end Z board, I'm certain of it.

Now don't get me wrong. I'm not trying to discourage anyone from trying. By all means, go out and experiment yourselves. Just understand you may fail a few times. I'd like to think the less than spectacular results I've had in the SBMs are mostly the result of very experimental builds (I'm sure some cynics will say they're proof I'm unqualified to write these). But every now and then you also hit significant pay dirt (I'll be touting the Intel Munchkin 2.0 for a while). Just remember you'll need to purchase your own lunch.

And on that note, if you'll excuse me, I think I'll go buy myself a sandwich.

Eric Vander Linden is a Contributing Writer for Tom's Hardware. Follow him on Twitter.

Follow us on Facebook, Google+, RSS, Twitter and YouTube.