Build the Best Bang for the Buck PC
We highlight the hardware that gets you the most performance per dollar spent
We all know that, generally speaking, buying the newest top-end part gets you the most performance. But in most cases, the premium you pay for that part covers a whole lot of other stuff as well that has no bearing on frame rates or video encoding times. We’re talking about the added cost of covering research and development, product marketing, lower production yields, etc. That high price also includes a vanity tax, if you will—the extra charge incurred by folks who simply want to have the latest hardware, hot off the fab, for bragging rights.
Splashing out on the very latest gear is all well and good, if you’re rolling in cabbage. But for most of us, the job of choosing what part to buy is much more nuanced. The object is to find the hardware that offers the greatest value—the best price-to-performance ratio.
That’s our objective here—to examine each of the major PC component categories, run the relevant benchmarks where warranted, and zero in on the coveted sweet spot. What we’ve come up with is a list of parts that are capable of meeting a typical enthusiast user’s needs for the best price possible.
CPU
Pay for the threads you need, not the threads you want
Make no mistake, we love cores and threads. But when you start talking about bang for the buck and sweet spots, eight-core and six-core CPUs don’t fit into the equation (well, unless you consider AMD’s Bulldozer, but that’s another matter). As power users who like to regularly drop in new CPUs, we have a personal fondness for the LGA2011 platform, but frankly, LGA2011 carries a price premium that the vast majority of users don’t need to pay.

Really, when you talk about bang for the buck, it very much gets into the complicated question of just what kind of bang you’re after. Are you a 100 percent gamer (a rare duck, in our book) or someone who spends most of his or her time pushing pixels professionally?
Since you pay for every thread in your system, you want to make sure they’re actually working for you. Take a gander at our benchmark chart, which compares a six-core Core i7-3930K to the quad-core Core i7-3820. To see what impact threads have on the benchmarks, we turned Hyper-Threading on and off for both processors. The value of the threads are there—but only if you have multithreaded apps that can use them. For the most part, today’s apps are optimized for quad-core or lower.

This quad part offers Ivy Bridge performance and is unlocked for overclocking.
Based on our thread experiment, we believe we can make a case for Intel’s $230 quad-core 3.4GHz Core i5-3570K chip sans Hyper-Threading. Ivy Bridge offers at least a 10 percent performance advantage over Sandy Bridge and the 3570K is unlocked. Keep in mind, if you run thread-heavy apps such as encoding or 3D rendering, you should pay for Hyper-Threading or move up to a six-core part, but the sweet spot today remains four cores.
Benchmarks
|
3.4GHz Core i7-3930K with Hyper-Threading
Enabled
|
3.4GHz Core i7-3930K with Hyper-Threading
Disabled
|
3.6GHz Core i7-3820 with Hyper-Threading
Enabled
|
3.6GHz Core i7-3820 with Hyper-Threading
Disabled
|
| Thread Count |
12 |
6 |
8 |
4 |
| Price |
$583 |
$583 |
$294 |
$294 |
| 3DMark 11 |
X5,794 |
X5,703 |
X5,730 |
X5,669 |
| 3DMark 11 Physics |
11,299 |
8,974 |
8,901 |
7,090 |
| x264 HD 5.0 Pass 1 (fps) |
85.8 |
51.9 |
62.8 |
51.6 |
| x264 HD 5.0 Pass 2 (fps) |
19 |
15.6 |
13.4 |
11.2 |
| ProShow Producer 5.0 (sec) |
1,553 |
1,500 |
1,565 |
1,546 |
| Stalker Day (fps) |
181 |
182 |
184.2 |
184.9 |
| Stalker Night (fps) |
206.5 |
211 |
206.5 |
206.5 |
| Stalker Rain (fps) |
240.7 |
238 |
240.7 |
236.4 |
| Stalker SunShafts (fps) |
159.2 |
158 |
161.2 |
160.4 |
| Stitch.EFx (sec) |
906 |
969 |
962 |
1,057 |
| Cinebench 11.5 (score) |
10.2 |
8.35 |
7.72 |
5.85 |
| Valve Particle Benchmark (fps) |
249 |
186 |
181 |
146 |
| 7Zip (MIPS) |
29,748 |
21,655 |
21,122 |
15,767 |
We used an Asus Sabertooth X79 motherboard, 8GB of DDR3/1333, a GeForce GTX 690, and OCZ Vertex 3 SSD, and 64-bit Windows 7 Professional for testing both CPU configurations.
Videocard
It’s a tight race but this relative newcomer takes the crown
When you begin to compare price/performance ratios of GPUs at the high end of the scale, you typically see diminishing returns once you go beyond about $300. That’s rarefied air up there, where 5–10 additional frames per second will cost you an extra $100 or so. The real battle for frames, frags, minds, and dollars exists below that price point, as the GPUs loitering in the mid-$200 region offer very good performance for literally half the price of the flagship videocards. This is the sweet spot—typically around $250 or so—where you can find 1080p gaming at 60fps for the least amount of money possible.

Don’t let its size and single 6-pin power connector fool you—the GTX 660 is a serious gaming weapon.
After examining the cards in this pricing segment, as well the cards in the pricing segment above it, we’re granting the brand-new Nvidia GTX 660 best-bang-for-the-buck status. At just $230, this GPU is faster than the slightly more expensive (or same-priced, depending on where you shop) AMD HD 7850 by a respectable margin, and even beats the pricier $280 Radeon HD 7870 in some of our tests, making it the fastest GPU under $250 by a wide margin.
So, why not spend $70 more and get a GTX 660 Ti? Looking at the benchmark chart, you can see that extra money nets you some performance gains, but in our judgment you don’t get enough of a return on that investment. The GTX 660 Ti costs roughly 25 percent more than a GTX 660, and yet offers an average of 15 percent performance improvement in most tests. When you tabulate price-per-fps, the GTX 660 also has the advantage over the GTX 660 Ti in the majority of tests we run, making it the value leader. As icing on the cake, you only need one 6-pin connector instead of the two that bigger cards require, so you’re saving money in the PSU department, as well.
Benchmarks
|
Gigabyte GTX 660 OC Version |
XFX Radeon HD 7850 |
EVGA GTX 560 Ti 448 |
XFX Radeon HD 7870 GHz |
MSI GTX 660 Ti Power Edition |
| Price |
$230 |
$245 (street) |
$250 |
$280 (street) |
$310 |
| 3DMark 2011 Perf |
6,935 |
6,075 |
6,295 |
7,001 |
9,118 |
| 3DMark Vantage Perf |
27,858 |
24,584 |
25,523 |
27,953 |
31,575 |
| Unigine Heaven 2.5 (fps) |
36 |
31 |
31.8 |
33.2 |
39.8 |
| Shogun 2 (1080p, fps) |
53.7 |
47 |
43.1 |
62.2 |
71.1 |
| Far Cry 2 / Long (fps) |
119 |
103 |
112.4 |
112.7 |
133.3 |
| Dirt 3 (fps) |
75.7 |
50 |
68.8 |
82.4 |
95.1 |
| STALKER: CoP DX11 (fps) |
42.8 |
34.7 |
38.4 |
42.6 |
49.9 |
| Just Cause 2 |
60.3 |
51 |
55.8 |
59.6 |
68 |
| Batman: Arkham City (fps) |
76 |
60 |
64 |
72 |
80 |
| Metro 2033 (fps) |
22 |
23 |
22.6 |
24.2 |
22.3 |
Best scores are bolded. Our test bed is a 3.3GHz Intel Core i7 3960X Extreme Edition in an Asus P797X Deluxe motherboard with 16GB Corsair DDR3/1600 RAM and a Corsair AX1200 PSU. The OS is 64-bit Windows 7 Ultimate. All games were run at 1900x1200 with 4x AA unless otherwise noted.
Click the next page to see the bang for your buck motherboard and SSD.