Sign in with
Sign up | Sign in
System Builder Marathon Q2 2014: System Value Compared
By ,
1. A SBM Based On Reader Feedback

System Builder Marathon, Q2 2014: The Articles

Here are links to each of the four articles in this quarter’s System Builder Marathon (we’ll update them as each story is published). And remember, these systems are all being given away at the end of the marathon.

To enter the giveaway, please fill out this SurveyGizmo form, and be sure to read the complete rules before entering!

Day 1: The Budget Gaming PC
Day 2: Our Mainstream Enthusiast System
Day 3: The Balanced High-End Build
Day 4: Performance And Value, Dissected

Choosing This Quarter's Winning Build

Paul, Don, and I already talked about the various reasons for this quarter’s budget restructuring in each of our respective stories. But the overall rationale involved a handful of requests from the audience. First, we saw a bunch of messages that Paul should return his budget gaming machine to its origins, “The $500 Gaming PC”. Second, we received a lot of requests to narrow the price range between our top and bottom builds. And third, we wanted to address the critique that our systems can't be compared to off-the-shelf machines because they don't include an operating system.

So Paul jumped back to $500, and then added an operating system. Since Don’s PC always costs twice as much as Paul’s, his sacrifice would be twice as great. Even harsher, my high-end build wouldn’t cost four times as much as Paul’s. It'd get a 3x multiplier instead. If you think a 3x jump is still a huge difference, consider the real-world price differences between entry-level and high-end components. My challenge would be great indeed.

Fortunately, we were all fairly close to that sweet spot in budget-oriented performance hardware right around $800 (give or take a couple hundred dollars). Although that sounds like a fairly large margin for error, the market is constantly changing as new mainstream parts are introduced. Furthermore, not all purchase decisions are made equally. For example, Paul was able to pull a value rabbit out of AMD’s hat with the low-cost Athlon X4 750K. Findings like those are probably why Intel finally caved and released an unlocked dual-core Pentium, which wasn’t available yet when we placed our orders.

Then there was the happy coincidence that the PowerColor graphics cards Don and I bought rose in price a little, but came bundled with a 250 GB Samsung 840 EVO just as our stories were going live. This small miracle addresses the biggest omission from Don's budget-constrained machine and saves me a $150 line item I was already planning to buy. Of course, neither of us gave ourselves credit for the windfall. But hey, as of this writing, the sale is still going on if you want to capitalize.

System Builder Marathon Q2 2014: The Machines
 $600 Gaming PC$1200 Enthusiast PC$1600 Performance PC
ProcessorAMD Athlon X4 750K: 3.4 to 4 GHz, Quad-Core, 4 MB Shared L2 CacheIntel Core i5-4670K: 3.4 to 3.8 GHz, Quad-Core, 6 MB Shared L3 CacheIntel Core i7-4770K: 3.5 to 3.9 GHz, Quad-Core, 8 MB Shared L3 Cache
GraphicsMSI R7 265 2GD5 OC 2 GBPowerColor TurboDuo AXR9 290 4GBD5-TDHE/OC Radeon R9 290PowerColor PCS+ AXR9 290X 4GBD5-PPDHE Radeon R9 290X
MotherboardASRock FM2A75M: Socket FM2+, AMD A75ASRock Z97 Pro3: LGA 1150 Intel Z97 ExpressAsus Z97-A: LGA 1150, Intel Z97 Express
MemoryTeam Group Dark TDBD38G1600HC9DC01: DDR3-1600 C9, 8 GB (2 x 4 GB) Team Group Vulcan TLYD38G1600HC9DC01: DDR3-1600 C9, 8 GB (2 x 4 GB) G.Skill Ripjaws X F3-14900CL8D-8GBXM: DDR3-1866 C8, 8 GB (2 x 4 GB) 
System DriveWD Blue WD10EZEX: 1 TB, 7200 RPM, 64 MB CacheWD Blue WD10EZEX: 1 TB, 7200 RPM, 64 MB CacheSamsung 840 EVO MZ-7TE250BW: 250 GB, SATA 6Gb/s SSD
PowerEVGA 100-W1-0430-KR: 430 W, ATX12V v2.31, 80 PLUSCorsair CX750: 750 W, ATX12V v2.3, 80 PLUS BronzeRosewill HIVE-750: 750 W Semi-Modular, ATX12V v2.3, 80 PLUS Bronze
CPU CoolerAMD Athlon Boxed Fan (included)Cooler Master Hyper 212 PlusThermaltake NiC L32 (CL-P002-AL14RE-A) 140mm 
 Platform Cost $454 $1066 $1362
Storage Drive(Uses System Drive)(Uses System Drive)Western Digital Blue WD10EZEX: 1 TB, 7200 RPM, 64MB Cache
OpticalAsus DRW-24B1ST/BLK/B/AS: 24x DVD±R, 48x CD-RLG GH24NSB0: 24x DVD±R, 48x CD-RLite-On iHAS124-04: 24x DVD±R, 48x CD-R
CaseRosewill Redbone U3Apevia X-HermesCM Storm Scout 2 Advanced
 Total HW $519 $1146 $1532
OSWindows 8.1 x64 OEMWindows 8.1 x64 OEMWindows 8.1 x64 OEM
 Total Cost $619 $1246 $1632

It's unfortunate that Intel's Devil's Canyon-based CPUs weren't available when we placed our orders back in May. We knew they were being introduced in early June. But they didn't ship out until a couple of days ago. Good thing we didn't wait. On the bright side, though, you can expect at least a couple of us to experiment with Intel's improved Haswell-based, overclocking-friendly models when we place our orders for Q3 here shortly. 

I realize that availability now of Core i5-4690K and Core i7-4790K means some of our readers are going to want to go that route instead. But we take comfort in the knowledge that our own tests reveal a mere 100 MHz overclocking advantage compared to a few of the Core i5-4670K and Core i7-4770K CPUs we’ve already tested.

That was also before I discovered the Core i7-4770K in this quarter's high-end machine is the worst-overclocking -4770K I've ever used, causing me to ponder the possibility of Intel setting aside its best dies for -4790Ks well before those newer processors launched. A complete comparison of overclock settings is found on the next page, so onward we go!

2. How We Tested Our Q2 2014 SBM Builds

The following tables include the stock and overclocked settings for this quarter's three System Builder Marathon configurations.

At the very bottom, you’ll find the programs and games used for benchmarking.

Test Hardware Configurations
 $600 Gaming PC$1200 Enthusiast PC$1600 Performance PC
Processor
(Overclock)
AMD Athlon X4 750K: 3.4 to 4.0 GHz, Quad-Core
O/C to 4.20 GHz, 1.27 V
Intel Core i5-4670K: 3.4 to 3.8 GHz, Quad-Core
O/C to 4.3 GHz, 1.285 V
Intel Core i7-4770K: 3.5 GHz to 3.9 GHz, Quad-Core
O/C to 4.20 GHz, 1.29 V
Graphics
(Overclock)
Radeon R7 265: 955 MHz GPU,  GDDR5-5600
O/C to 1050 MHz, GDDR5-5800
Radeon R9 290: 975 MHz GPU, GDDR5-5000
O/C to 1100 MHz, GDDR5-5600
Radeon R9 290X: 1050 MHz GPU,  GDDR5-5400
O/C to 1100 MHz, GDDR5-6200
Memory
(Overclock)
8 GB DDR3-1600 CAS 9-9-9-24, O/C to DDR3-1866 CAS 9-9-9-24, 1.63 V8 GB DDR3-1600 CAS 9-9-9-24, no O/C8 GB DDR3-1866 CAS 8-9-9-24, O/C to DDR3-2133 CL 9-10-10-27, 1.60 V
Motherboard
(Overclock)
ASRock FM2A75M: Socket FM2+, AMD A75 (CPU internal O/C 2000 MHz HT)ASRock Z97 Pro3: LGA 1150, Intel Z87 Express
Stock 100 MHz BCLK
Asus Z97-A: LGA 1150, Intel Z97 Express
Stock 100 MHz BCLK
CaseRosewill Redbone U3Apevia X-HermesCM Storm Scout 2 Advanced
CPU CoolerAMD Athlon Boxed Fan (included)Cooler Master Hyper 212 PlusThermaltake NiC L32
Hard DriveWD Blue WD10EZEX 1 TB, 7200 RPM, 64 MB CacheWD Blue WD10EZEX 1 TB, 7200 RPM, 64 MB CacheSamsung MZ-7TE250BW 250 GB SATA 6Gb/s SSD
PowerEVGA 100-W1-0430-KR: 430 W, ATX12V v2.31, 80 PLUSCorsair CX750: 750 W, 80 PLUS BronzeRosewill HIVE-750: 750 W Semi-Modular, 80 PLUS Bronze
Software
OSMicrosoft Windows 8 Pro x64Microsoft Windows 8 Pro x64Microsoft Windows 8 Pro x64
GraphicsAMD Catalyst 14.4AMD Catalyst 14.4AMD Catalyst 14.4
Chipset Intel INF 9.4.0.1026Intel INF 9.4.0.1026

You’d expect at least 4.3 GHz from a Core i7-4770K, right? That average is what, around 4.4 GHz? And our own samples frequently reach 4.7 GHz. It's tempting to blame the previously-untested CPU cooler for my $1600 machine’s poor results, but that mid-priced part still managed to support the CPU to our previous limit voltage. And voltage is the biggest variable in heat, particularly when most of that heat is trapped under a layer of sub-par thermal interface material that Intel previously used between its die and heat spreader.

Overclocking issues are virtually non-existent for any of the other parts in this competition. Don usually reaches 4.3 GHz on his Core i5s, and usually leaves his memory underclocked for initial testing whenever standard clocks require XMP (since this was supposed to be an overclocking technology). Paul reached only 4.2 GHz where he might have hoped for 4.4 GHz. But he didn’t spend any money on CPU cooling, so he doesn't fare badly at all.

Benchmark Configuration
3D Games
Arma 3
V. 1.20 Current PC, V.1.08 Q1 PC
30-sec. Fraps "Infantry Showcase"
Test Set 1: Standard Preset, No AA, Standard AF
Test Set 2: Ultra Preset, 8x FSAA, Ultra AF
Battlefield 4Version 1.0.0.1, DirectX 11, 100-sec. Fraps "Tashgar"
Test Set 1: Medium Quality Preset, No AA, 4x AF, SSAO
Test Set 2: Ultra Quality Preset, 4x MSAA, 16x AF, HBAO
Far Cry 3V. 1.05, DirectX 11, 50-sec. Fraps "Amanaki Outpost"
Test Set 1: High Quality, No AA, Standard ATC, SSAO
Test Set 2: Ultra Quality, 4x MSAA, Enhanced ATC, HDAO AMD/ HBAO NVidia 
Grid 2Version 1.0.85.8679, Direct X 11, Built-in Benchmark
Test Set 1: High Quality, No AA
Test Set 2: Ultra Quality, 8x MSAA
Audio/Video Encoding
HandBrake CLIVersion: 0.99, Video: Video from Canon EOS 7D (1920x1080, 25 frames) 1 Minutes 22 Seconds, Audio: PCM-S16, 48,000 Hz, Two-channel, to Video: AVC1 Audio: AAC (High Profile)
iTunesVersion 11.0.4.4 x64: Audio CD (Terminator II SE), 53 minutes, default AAC format 
LAME MP3Version 3.98.3: Audio CD "Terminator II SE", 53 min, convert WAV to MP3 audio format, Command: -b 160 --nores (160 Kb/s)
TotalCode Studio 2.5Version: 2.5.0.10677, MPEG2 to H.264, MainConcept H.264/AVC Codec, 28 sec HDTV 1920x1080 (MPEG-2), Audio: MPEG-2 (44.1 kHz, Two-channel, 16-bit, 224 Kb/s) Codec: H.264 Pro, Mode: PAL 50i (25 FPS), Profile: H.264 BD HDMV
Adobe Creative Suite
Adobe After Effects CCVersion 12.0.0.404: Create Video, 3 Streams, 210 Frames, Render Multiple Frames Simultaneously
Adobe Photoshop CCVersion 14 x64: Filter 15.7 MB TIF Image: Radial Blur, Shape Blur, Median, Polar Coordinates
Adobe Premiere Pro CS6Version 6.0.0.0, 6.61 GB MXF Project to H.264 to H.264 Blu-ray, Output 1920x1080, Maximum Quality
Adobe Acrobat XIVersion 11.0.0: Print PDF from 115 Page PowerPoint, 128-bit RC4 Encyption
Productivity
ABBYY FineReaderVersion 10.0.102.95: Read PDF save to Doc, Source: Political Economy (J. Broadhurst 1842) 111 Pages
Autodesk 3ds Max 2013Version 15.0 x64: Space Flyby Mentalray, 248 Frames, 1440x1080
BlenderVersion 2.68a, Cycles Engine, Syntax blender -b thg.blend -f 1, 1920x1080, 8x Anti-Aliasing, Render THG.blend frame 1
Visual StudioVersion 10.0, Compile Google Chrome, Scripted
Compression
7-ZipVersion 9.30 alpha, LZMA2, Syntax "a -t7z -r -m0=LZMA2 -mx=5"
Benchmark: THG-Workload-2012 (1.3 GB)
WinRARVersion 5.0, RAR, Syntax "winrar a -r -m3"
Benchmark: THG-Workload-2012 (1.3 GB)
WinZipVersion 18.0 Pro, Syntax "-a -ez -p -r"
Benchmark: THG-Workload-2012 (1.3 GB)
Synthetic Benchmarks and Settings
3DMark ProfessionalVersion: 1.2.250.0, Fire Strike Standard and Extreme
PCMark 8Version: 1.0.0 x64 Full Test
SiSoftware Sandra 2014Version: 2014.02.20.10, Processor Arithmetic, Cryptography, Memory Bandwidth Benchmarks
3. Results: 3DMark And PCMark

3DMark shows that the Radeon R9 290X was a great choice for my $1600 PC, almost propping it up as the CPU trips over itself in the overclocking portion.

The $1200 machine’s Radeon R9 290 still looks strong, though perhaps weaker than expected in comparison to the expensive machine’s single-GPU flagship.

Paul's $600 PC necessarily includes a weaker graphics card, though that's still a boon to value since the half-priced platform achieves more than half of Don's performance.

Paul figured out the trick to Don’s improved PCMark scores, and I eventually got in on the secret too: an error in the batch process produced low (but consistent) Home and Creative scores, so we re-ran this benchmark manually.

Because PCMark’s storage scores are based on real-world load times, they’re the only synthetic metrics that make it into our final performance analysis. All three systems feature a WD Blue 1 TB disk. However, my $1600 machine uses it as a secondary drive, backing a 250 GB Samsung 840 EVO SSD.

4. Results: SiSoftware Sandra

The $1200 and $1600 machines employ different models of the same quad-core Haswell-based die. Don's Core i5 doesn't benefit from Hyper-Threading though, and 2 MB of its shared L3 cache is disabled compared to my Core i7. The i5 is also clocked 100 MHz slower by default, but Don was able to overclock his -4670K 100 MHz higher than my Core i7.

Still, the i7 wins this face-off, demonstrating the benefits of parallelization, even through a technology like Hyper-Threading, where two threads are scheduled to each physical core.

My more conservatively-overclocked high-end machine also enjoys an advantage in Sandra's Cryptography module, though we know that this is related to memory bandwidth available for shuttling instructions to the processor's cores.

Sure enough, there's the source of my advantage (at least in the AES-NI-accelerated Encoding metric). Memory bandwidth charts should never scale proportionally to system price. But this one does. That’s partly because Don doesn’t overclock his memory, and lets it run at underclocked defaults as a baseline. At the other end of the chart, Paul can blame AMD's integrated memory controller for the $600 machine’s poor results.

Fortunately for both of my fellow editors, real-world software rarely reflects memory bandwidth with a 1:1 relationship. It's more likely that we'd see slightly lower performance in certain tests attributable to these numbers.

5. Results: Battlefield 4

Battlefield 4’s limit of 200 FPS hurts the value proposition of high-end graphics systems, if only because the game caps performance when it's needed to offset cost. As a result, we have to be careful to consider the most relevant resolutions.

Today, our comparison involves three lower-cost configurations though. Only the most expensive machine's graphics card runs into Battlefield's ceiling, and even then, only once.

All three authors surpass their goals: my $1600 build plays smoothly through 5760x1080, Don's $1200 machine through 4800x900, and Paul's $600 PC through 1920x1080 at Battlefield 4’s Ultra quality preset.

6. Results: Grid 2

Grid 2 won’t embarrass the owner of a cheap graphics card, but it will cause enthusiasts who combine a higher-end GPU with mid-range host processing or system memory to chase bottlenecks. Overclocking solved both limitations in my $1600 PC, while the Radeon R9 290 in Don’s $1200 PC appears held back slightly by his PC's DRAM bandwidth limits.

Grid 2 isn’t a great game for identifying enthusiast-level graphics bottlenecks, but it is good for diagnostics. It’s also pretty fun to play.

The harder you lean on graphics, the more Grid 2 demands of the rest of your system. The $1600 and $1200 machines both appear bottlenecked using the Ultra graphics preset at 1920x1080 and below, at least until they're overclocked. Paul had bigger concerns with the prowess of his R7 265, yet still managed to push over 50 FPS at his target 1920x1080 resolution.

7. Results: Arma 3

Like Grid 2, Arma 3 appears to encounter some kind of non-graphics bottleneck at 1080p and lower resolutions.

I don't have enough experience testing this game with various hardware combinations to properly identify the limiting component. However, Paul put a fair effort into dissecting Arma's pain points and eventually determined that our Infantry Showcase test isn't hard enough on the CPU to represent some of the game's more demanding maps. However, the title is very platform-dependent. It's even possible to overclock your graphics card and see no improvement if you don't also tune your host processor. 

Given the scale of performance gains from overclocking, I’m guessing there's also a DRAM bottleneck affecting our systems in Arma 3. That's mostly because I wasn’t able to get much headroom out of my CPU, but still realized impressive gains.

As a result, this game looks like a prime candidate for my memory kit round-ups.

All three systems survive their targeted resolutions using the Ultra preset, though my $1600 configuration barely qualifies as smooth with a 33.8 FPS average in its 5760x1080 test.

8. Results: Far Cry 3

On more than one occasion, I've noticed that Don's less expensive machines outperform mine in Far Cry 3 at 1920x1080, regardless of differences in hardware, drivers, or game updates. Though he might get credit for a slightly better CPU overclock, I'm still not certain how he manages to trump me.

On the other hand, my higher-end graphics card does a wonderful job of outpacing his once we switch to triple-display resolutions.

My machine plays through 4800x900 smoothly, but gets a little choppy at our highest Far Cry 3 settings and a 5760x1080 resolution. Don’s $1200 machine appears to need overclocking for smooth game play at 4800x900. And Paul likewise gets choppy at his targeted 1920x1080. That’s because the delta between average and minimums can be over 40% on the map used for our test.

9. Results: Audio And Video Encoding

Our Apple iTunes and LAME MP3 tests are greatly affected by CPU architecture. But because the $1200 and $1600 machines both lean on Haswell-based parts, performance differences in both benchmarks are limited to clock rate variations. That means the $1600 build’s higher stock frequency jumps into an early lead, while the $1200 PC's higher overclock leapfrogs my Core i7-4770K.

The $1600 and $600 builds are both overclocked to the same 4.2 GHz. However, Paul's platform centers on AMD's Piledriver architecture, which struggles with IPC throughput; it simply cannot get as much work done at a given frequency. That manifests in single- and multi-threaded workloads as a significant performance deficit.

It’s nice to see that the extra expense of a Core i7 is useful for some things. Benefiting from a little extra shared L3 cache and Hyper-Threading technology (to improve core utilization), the $1600 PC appears a monster in both HandBrake and TotalCode Studio.

10. Results: Adobe Creative Suite

Adobe After Effects appears to show an affinity for the high-end configuration's extra cache, while Photoshop does not. Somewhat surprisingly, our processor-limited Photoshop workload doesn't get as much benefit from Hyper-Threading as other applications in our suite.

Where we do see better scaling is the OpenCL-accelerated Photoshop workload, which appears to reflect my higher-end machine's faster Radeon R9 290X (compared to Don's R9 290).

The $600 PC drags through the entire suite, but doesn’t drop to half of the $1200 machine’s performance until we get to Adobe Premiere Pro. Its successes elsewhere mean that Paul can probably look forward to a nice position on the value chart.

Acrobat is an office-oriented program wrapped within the Creative Cloud family. Because it’s a document-oriented application, its performance is reflected in our Productivity test suite.

11. Results: Productivity

My $1600 build only needs to enable 50% more performance than Don’s $1200 machine to reach value parity. 3ds Max nearly gets me there.

Unfortunately for both of us, Don’s $1200 PC doesn't quite maintain twice the performance of Paul’s $600 configuration.

Don’s machine lurches closer to mine in Blender, but at least he’s getting a little advance on Paul. With the $600 PC performing beyond its price in most of our previous benchmarks, Don needs this break.

Don’s $1200 build is nearly twice as fast as Paul’s $600 PC again in Visual Studio 2010. Still needing only a 50% performance advance over Don, my $1600 PC breaks that barrier in the same benchmark.

12. Results: File Compression

Don and I run down to the wire in 7-Zip, with the $1600 PC’s performance exceeding its premium over the $1200 machine. Unfortunately for us, Paul’s entry-level $600 rig isn’t slow enough to offset Don’s comparatively doubled price.

The $1200 PC surges ahead in WinRAR, while the $1600 and $600 builds hit roughly the same chart scale as they did in 7-Zip.

WinZip mixes things up. I’m happy to see my $1600 PC performing nearly twice as well as the $1200 machine when using the program’s demanding “EZ” switch, but sad to see the $600 machine tied for second place. This will definitely hurt the value scores of both stepped-up configurations.

13. Results: Power And Heat

Intel's Hyper-Threading simultaneous multi-threading technology doesn’t increase consumption much on its own, but when it increases core utilization through greater parallelism, those cores do use more power. We've seen this in a few benchmarks, and it also shows up in power consumption numbers. There’s no free lunch.

The $600 machine is the true miser, but not because of its 100 W CPU. Rather, the Pitcairn-based R7 265 graphics card appears to be Paul's saving grace based on our load power measurements.

There’s that 80-degree ceiling again. You know it as the temperature where Intel’s Haswell-based cores throttle in a 20° (Celsius) room. But again, I have to point out that this was measured when both the CPU and GPU were fully-loaded, and that I picked a graphics card that vents all of its waste heat (and there's a lot) internally and put it in a case with one exhaust fan.

Then again, the same can be said of Don’s no-so-hot graphics card. Combining a lower overclocked CPU voltage and GPU wattage are a huge benefit to his temperatures under load.

AMD's processor behave a little differently, traditionally running into stability issues when the CPU hits mid-60s. So, Paul's $600 machine probably falls between the other two builds in terms of environmental heat tolerance. This is where I ask Paul to stop his fan and see if the system crashes.

14. Results: Overall Performance And Efficiency

Efficiency is where work meets power. This is all theoretical of course, since no physical work is being done by a CPU. It’s not like you use all those numbers to spin a wheel.

Since we’re calculating theoretical (rather than actual) efficiency, we need a baseline by which other products can be compared (or related, as shown in the title “Relative Efficiency” in the second chart). Since the lowest-performance PC and the lowest-energy PC are the same $600 unit, it becomes our baseline.

A peek at the raw data tells us where the wind will blow in the “Value” charts on the next page, since the double-priced $1200 machine is rarely twice as fast as the $600 system. But before we get there, we still have that long-discussed efficiency chart to discover.

Since efficiency can’t exceed 100%, the 100% baseline is subtracted from the chart above to make it zero. The resulting numbers are all relative, as in the $1600 PC achieves 18% more efficiency compared to the $600 PC.

Increased frequencies and voltages drive down the efficiency of all three overclocked machines, whereas a lesser overclock at stock voltage might have caused efficiency to climb by allowing more work to be accomplished with little additional energy.

15. Our System Builder Marathon, By The Numbers

I'm so used to seeing the cheapest PC win our value comparison that I nearly asked “by how much” rather than “who” the winner would be. And the best place to be when trying to score a value win with cheap components is at an open-air bench, since a benchable platform doesn’t need a case or optical drive. That's how the crew chose to start our value analysis this quarter: performance-oriented parts-only.

The above chart is really only useful to people who want to upgrade old system with new internal components though, and this is a System Builder Marathon. The cheap machine’s margin of victory is reduced when we create a complete system of these platforms.

And then there’s the argument that some enthusiasts don't have a decent operating system laying around (or at least a key not already in-use on another machine), and you really need one to play games or run any of the tested applications. Most of us certainly wouldn’t want to recycle an old Windows XP license, and that’s not even within the limits of your contract if the old OS was an OEM version. A third chart shows the “worst value scenario” for the cheap PC, where it’s paired with the same $100 Windows 8.1 license as its more expensive rivals.

If you want to game on a single FHD monitor, the $600 PC is a great start. You’ll save a bunch of money and have hundreds of gigabytes of spare storage space on its single hard drive for your videos, photos, and downloads. Programs are going to be slower to load than if you had an SSD. But if you're already using an older PC, that's not going to be surprising. Congratulations once again, Paul!

Don's $1200 PC edges out my $1600 PC in value, though it's also stuck running programs from the same slow hard drive. And its 10% margin occurred only after Don got lucky and drew better than me in the overclocking lottery. Naturally, your mileage may vary. Still, there's a lot of performance to appreciate from the mid-range configuration, and it appears well-balanced.

Enthusiasts with a little more to spend can appreciate that my $1600 PC is even faster, despite a slight value penalty. Tom's Hardware readers already know this, but as you throw additional budget at high-end systems, it gets increasingly difficult to improve value. Poor overclock aside, I can be fairly proud of the hardware combination I put together.

What I just said about overall value also applies to high-resolution gaming. Sure, there are one or two game settings where the $1200 PC barely fails and the $1600 PC barely passes. But I’d rather adjust my game settings down than drop an extra $400 on equipment if I were on a tight budget.

And so I again concede Don's second-place value over my own third-place build, even though I wouldn’t personally build that machine. Then again, I'm the guy with the lab full of computer equipment.