System Builder Marathon Q1 2015: System Value Compared

Benchmark Results

I’d love to believe that my pair of GeForce GTX 970s might double the performance of a single card, but Don’s machine proves that’s not going to happen. Since his overclock is better than mine, the most notable gain in 3DMark comes from the Graphics module at just under 80% and with both machines at their stock clock rates.

Don also finds his machine unable to double the graphics score of Paul’s low-cost gaming powerhouse, with the balance of tests indicating that the processor is his most important upgrade.

My $1750 machine's PCMark storage result mysteriously dropped by more than 400 points after continuous testing. Rather than try to figure out why, I pushed through with the lower numbers. This is reflected in the difference between Don’s machine and my own, both of which have similarly-capable SSDs. Paul’s $700 system gets around half the score from its hard drive.

Other PCMark scores still favor the most expensive machine, though not by a large margin. These won’t matter in the final value analysis anyway, since we only use the storage test in our overall tabulation.

Surprisingly, my $1750 machine produces higher Sandra Arithmetic scores than the $1300 PC, even though both use the same CPU. Paul’s dual-core processor shows poorly, producing less than half the performance of Don’s.

Don’s overclocked $1300 PC outperforms my $1750 build in a couple of Sandra Cryptography tests, while Paul’s $700 machine produces better than half the performance of Don’s.

The reason for Don’s performance advantage in Sandra Cryptography becomes clear upon viewing memory bandwidth results. My 16GB kit has twice the capacity, but requires poor timings for ultimate stability.

Two graphics cards are definitely better than one in Arma 3, where Don captures only a single 1600x900 win. Paul’s $700 PC easily pulls past its 1920x1080 target.

A 200 FPS cap on Battlefield 4 is sure to hurt my $1750 PC’s value score. But that ceiling only affects two out of eight tests. My 5760x1080 results top Don’s at 4800x900. Even the weaker GPU in Paul’s $700 build is adequate for his lower-resolution target.

Don’s super-high frame rates in Far Cry 3’s lowest test settings will never make sense to me, just as my bottleneck (on every system I’ve tested) have never made sense to him. Triple-screen resolutions look great on the $1750 machine, and the $700 PC produces adequate results through the toughest single-screen test.

Unlike Far Cry 3, Grid 2’s bottlenecks can be defined fairly easily. The memory timings of my $1750 machine are less than spectacular. Higher resolutions are where GPU performance matters, but don’t let the good scaling fool you. Every machine yields playable performance in even the toughest test.

Image 1 of 4

Paul’s $700 PC performs well in single-threaded iTunes, HandBrake and Acrobat workloads, but Don’s $1300 machine proves that the extra CPU expense is well-spent. SLI does little for my $1750 build, though the extra memory capacity enables a big boost in Adobe After Effects.