The Price/Performance Comparison and Final Thoughts
The following graph illustrates the variance in build cost and observed performance, and it's pretty significant! Only the most and least expensive builds had the expected performance average (best, worst), but the remaining 14 were all over the map.
A very interesting thing happens when you look to the right, where the system prices drop significanly below $400. Down into the $350 range the average FPS numbers jump back up to 80% of the performance of the most expensive system build! This is the price/performance sweetspot of the 16 systems tested, and it's actually two similar builds with nearly identical performance.
- Full System Carts from Amazon
- GTX 750 Ti + X4 760K
- GTX 750 Ti + Pentium G3220
- GTX 750 + Pentium G3220
- R7 260X + Pentium G3220
The builds based on an Intel Pentium G3220 with either a Radeon R7 260X or GeForce GTX 750 graphics card were able to provide an average of over 60 FPS in the game benchmarks, and did so with a build price $100 less than the top-ranked system. And these builds feature 90% of the performance of that $452 top system, an ideal example of price/performance for gaming!
Of the final two best-value graphics card selections for the budget system, the Radeon R7 260X is my pick for overall low-cost GPU for 1080p gaming. Though the GTX 750 was nearly identical across the game benchmarks, the 260X put up a heck of a showing in the synthetic benchmarks often matching or even beating an overclocked GTX 750 Ti at times (which made me kind of question the game benchmarks I chose). Plus, the R7 260X supports Mantle. You know, that graphics API I didn’t use here?
The Intel G3220 has proven itself to be a great performer and an excellent value at around $60, and while an H81 motherboard it was paired with isn’t glamourous it’s definitely easy to find one cheap. For basic computing and single GPU gaming the H81 chipset gets the job done, but even if you decided to spend more on the motherboard to “future-proof” a bit (like Z87 or Z97) this dual-core Haswell part would still be a pretty good place to start, particularly if you’re using a 1080p monitor.
I will again say the A88X motherboard I chose was overkill for the CPUs I ended up with, and a combination of a slightly less expensive chipset/board and the X4 760K would provide all of the benefits of the extra cores for multithread aware games and applications. It would still be debatable whether this would be better from a pure price/performance standpoint against a CPU like Intel’s Pentium G3220 (unless you have a preference, which I have heard some of you may have).
One More Thing…
While either of the two highlighted builds I mentioned above can play modern DX11 games at 1080p with good – better than just "playable" – frame rates for about $355 total (before adding an enclosure or OS if needed), there's another interesting option here. Normally in a review we'd be giving most of the attention to the best performance, but here of course the emphasis has been on price. The other interesting combo is more expensive, and one that we've looked at before for budget gaming.
For just under $400, the GTX 750 Ti / Pentium G3220 combo provided the second best overall performance for $50 less than the system based with the same video card with the Athlon X4. And while a system with the R7 260X did at times beat the 750 Ti depending on the benchmark, this alternate build might be worth the extra money depending on your needs. Besides providing better average performance across all tests, the $399 build was more efficient and nearly silent. Though I didn't cover power consumption and noise in this article, the GTX 750 Ti draws very little juice (even this overclocked version doesn't require an external connector) and was very quiet throughout testing. (My single-fan R7 260X was one of the loudest cards I've ever used!)
I hope I’ve answered some questions about budget gaming here, but even with 16 combinations this article is no where near comprehensive. I could speculate endlessly about products I haven't personally tested. In the end it's really up to the reader to draw their own conclusions, and hopefully something in the benchmark data helps with a buying decision!