System Builder Marathon, Q2 2013: 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 $650 Mini-ITX Gaming PC
Day 2: The $1300 Mini-ITX Enthusiast PC
Day 3: The $2500 Mini-Performance PC
Day 4: Performance And Value, Dissected
Day 5: The $400 "True Spirit of Mini-ITX" PC
Introduction
When it comes to compact PCs, the boutique companies that can afford to commission their own designs are often the ones pushing the performance envelope hardest. And yet, enthusiasts still believe that adopting a form factor like mini-ITX necessarily means making severe compromises. But we showed in Meet The Tiki: Core i7-3770K And GeForce GTX 680 In A Mini-ITX Box? why this doesn't have to be true.
Of course, the only way to get Falcon Northwest's enclosure is buying the company's PC. So, for this quarter's System Builder Marathon, we wanted to give fans of speed in small spaces a handful of do-it-yourself options. Although none of the boxes we built are as small as the Tiki, they represent a healthy cross-section of what can be constructed using today's most efficient components.
Formerly a love-it or hate-it form factor reserved for boring little office PCs, mini-ITX gained gaming cred thanks in part to AMD’s DTX efforts, perhaps almost ironically given the power consumption figures of modern components. All three of our builds employ ITX motherboards, yet all three cases have the DTX-mandated second slot required for double-wide graphics cards. Similarly, all three builds use full-sized power supplies to feed those hungry cards.

Having thoroughly deviated from VIA’s ITX specification, we’ve now moved into an era of semi-compact, open-architecture, full-performance computing.
We already proved that our smaller builds can perform like their full-sized predecessors, and now we’re ready to see how they compare to each other in terms of performance and value.
| Q2 2013 System Builder Marathon | |||
|---|---|---|---|
| $650 Gaming PC | $1300 Enthusiast PC | $2500 Performance PC | |
| Processor | Intel Core i3-3220: 3.3 GHz, Dual-Core, 3 MB Shared L3 Cache | Intel Core i5-3570K: 3.4 GHz Base, 3.8 GHz Max. Turbo Boost, Quad-Core, 6 MB Shared L3 Cache | Intel Core i7-3770K: 3.5 GHz Base, 3.9 GHz Max. Turbo Boost, Quad-Core, 8 MB Shared L3 Cache |
| Graphics | PowerColor PCS+ AX7870 2 GB 256-bit GDDR5 | Sparkle GeForce GTX 680 2 GB 256-bit GDDR5 | Asus GTX690-4GD5 GTX 690 4 GB |
| Motherboard | ASRock B75M-ITX: LGA 1155, Intel B75 Express | MSI Z77IA-E53: LGA 1155, Intel Z77 Express | Asus P8Z77-I Deluxe: LGA 1155, Intel Z77 Express |
| Memory | Crucial Ballistix Tactical Tracer BLT2KIT2G3D1608DT2TXRG: DDR3-1600 C8, 4 GB (2x 2GB) | G.Skill Ripjaws X F3-14900CL8D-8GBXM: DDR3-1866 8 GB (2 x 4 GB) | Crucial Ballistix Tactical BLT2K8G3D1608ET3LX0: DDR3-1600 C8, 16 GB (2 x 8 GB) |
| System Drive | Western Digital WD5000AAKX: 500 GB, SATA 6Gb/s Hard Drive | Adata XPG ASX900S3-64GM-C: 64 GB, SATA 6Gb/s SSD | Mushkin MKNSSDCR240GB-DX: 240 GB, SATA 6Gb/s SSD |
| Storage Drive | Uses System Drive | Western Digital WD1002FAEX: 1 TB, SATA 6Gb/s Hard Drive | Western Digital WD2002FAEX: 2 TB, SATA 6Gb/s Hard Drive |
| Optical | None | Lite-On iHAS124-04: 24x DVD±R, 48x CD-R | Asus DRW-24B1ST: 14x BD-R, 16x DVD±R |
| Case | Cooler Master Elite 120 Advanced | Lian Li PC-Q08B | BitFenix Prodigy BFC-PRO-300-RRXKR-RP |
| Prodigy Mesh Front Panel C-PRO-300-KRFXA-RP | |||
| BitFenix 140 mm Fan BFF-SCF-14025WW-RP | |||
| SilverStone FF143B 140 mm Dust Filter | |||
| Power | Corsair CX500: 500 W, ATX12V v2.3, 80 PLUS Bronze | Corsair CX750M: 750 W Modular, ATX12V v2.3 80 PLUS Bronze | Seasonic SS-660XP2: 660 W Modular, ATX12V v2.3, 80 PLUS Platinum |
| CPU Cooler | Intel boxed heatsink/fan | Antec Kuhler H2O 620 Liquid Cooling System | NZXT Kraken X40 RL-KRX40-01 |
| Build Cost | $653 | $1354 | $2451 |
The comment I made about paying $50 bucks extra for $50 fewer features to enable a mini-ITX configuration still stands. Paul's $650 gaming PC had to give up its optical drive to approach its now-theoretical budget limit. That’s alright for many people who have second machines to rip images that they can then drop onto USB thumb drives. But it's something you'll need to keep in mind before committing to such an approach.
At the opposite end of the pricing scale, a somewhat-costly Blu-ray burner gets my $2500 closer to its budget while adding convenience and expanding its capabilities. I simply couldn’t find a compelling performance upgrade on which to spend my left-over loot, and instead decided to focus on features.
In the middle, Don’s enthusiast build breaks the bank to add a 60 GB SSD to its 1 TB hard drive. He gets an artificially-inflated storage score, since the boot drive really can't hold all of the applications he's benchmarking. Even still, on a budget, we'd rather see a small SSD than no solid-state storage at all.
With Paul dropping his optical drive to stay within budget and Don almost ignoring his budget to keep his SSD, the fact that I was adding features to my machine simply to burn through the budget is going to come back and bite me on the rear, I just know it.

Test Settings And Overclocked Configurations
| Test Hardware Configurations | |||
|---|---|---|---|
| $650 Gaming PC | $1300 Enthusiast PC | $2500 Performance PC | |
| Processor (Overclock) | Intel Core i3-3220, 3.3 GHz, Two Physical Cores No O/C | Intel Core i5-3570K, 3.4 GHz, Four Physical Cores O/C to 4.3 GHz, 1.3 V | Intel Core i7-3770K, 3.5 GHz, Four Physical Cores O/C to 4.6 GHz, 1.3 V |
| Graphics (Overclock) | PowerColor PCS+ AX7870 2 GB: 925-975 MHz GPU, GDDR5-6000 O/C to 1100 MHz GPU | Sparkle GeForce GTX 680 2 GB: 1006-1059 MHz GPU, GDDR5-6008 O/C to 1153 MHz GDDR5-6408 | Asus GTX690-4GD5: 915-1019 MHz GPU, GDDR5-6008 O/C to 1200 MHz GDDR5-6400 |
| Memory (Overclock) | 4 GB Crucial Ballistix DDR3-1600 CAS 8-8-8-24, No O/C | 8 GB G.Skill DDR3-1866 CAS 11-11-11-28 1T, O/C at 1.5 V to DDR3-1866 CL 8-9-9-24 2T | 16 GB Crucial DDR3-1600 CAS 8-8-8-24, O/C at 1.5 V to DDR3-2133 CL 9-9-9-24 |
| Motherboard (Overclock) | ASRock B75M-ITX: LGA 1155, Intel B75 Express | MSI Z77IA-E53: LGA 1155, Intel Z77 Express | Asus P8Z77-I Deluxe: LGA 1155, Intel Z77 Express |
| Optical | None | Lite-On iHAS124 24x DVD±R | Asus BW-14D1XT: 14x BD-R |
| Case | Cooler Master Elite 120 Advanced | Lian Li PC-Q08B | BitFenix Prodigy w/Mesh Front |
| CPU Cooler | Intel Boxed Heat Sink And Fan | Antec Kuhler H2O 620 | NZXT Kraken X40 |
| Hard Drive | Western Digital WD5000AAKX: 500 GB, SATA 6Gb/s Hard Drive | Adata XPG ASX900S3-64GM-C: 64 GB, SATA 6Gb/s SSD | Mushkin Chronos Deluxe DX 240 GB, SATA 6Gb/s SSD |
| Power | Corsair CX500: 500 W, ATX12V v2.3, 80 PLUS Bronze | Corsair CX750M: 750 W Modular, ATX12V v2.3 80 PLUS Bronze | Seasonic SS-660XP2: 660 W Modular, ATX12V v2.3, 80 PLUS Platinum |
| Software | |||
| OS | Microsoft Windows 8 Professional x64 | ||
| Graphics | AMD Catalyst 13.4 | Nvidia GeForce 314.22 | Nvidia GeForce 314.22 |
| Chipset | Intel 7-series Inf v. 9.3.1025 | Intel INF 9.3.0.1026 | Intel INF 9.3.0.1026 |
Rather than manually overclock his memory, Don decided to run it at SPD defaults for baseline measurements and XMP defaults for overclocked tests. I managed to push my DDR3-1600 to DDR3-2133 by adding one cycle to its timings, while Paul’s non-overclockable platform remains locked into its defaults.
Don’s Core i5 system appears to have hit a barrier at 4.3 GHz, which is slightly lower than the 4.4 GHz I’ve grown to expect from Intel's lower-binned quad-core chips. Conversely, my Core i7 offered a better-than-expected 4.6 GHz at the same voltage setting.
GPU overclocking will likely make a far more consistent performance difference in all three systems, but only in games and perhaps the occasional OpenCL-boosted application.
| Benchmark Configuration | |
|---|---|
| 3D Games | |
| Battlefield 3 | Campaign Mode, "Going Hunting" 90-Second Fraps Test Set 1: Medium Quality Defaults (No AA, 4x AF) Test Set 2: Ultra Quality Defaults (4x AA, 16x AF) |
| F1 2012 | Steam Version, In-Game Test Test Set 1: High Quality Preset, No AA Test Set 2: Ultra Quality Preset, 8x AA |
| The Elder Scrolls V: Skyrim | Update 1.5.26, Celedon Aethirborn Level 6, 25-Second Fraps Test Set 1: DX11, High Details No AA, 8x AF, FXAA enabled Test Set 2: DX11, Ultra Details, 8x AA, 16x AF, FXAA enabled |
| Far Cry 3 | V. 1.04, 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 |
| Adobe Creative Suite | |
| Adobe After Effects CS6 | Version 11.0.0.378 x64: Create Video which includes three Streams, 210 Frames, Render Multiple Frames Simultaneosly |
| Adobe Photoshop CS6 | Version 13 x64: Filter 15.7 MB TIF Image: Radial Blur, Shape Blur, Median, Polar Coordinates |
| Adobe Premeire Pro CS6 | Version 6.0.0.0, 6.61 GB MXF Project to H.264 Blu-ray, Output 1920x1080, Maximum Quality |
| Audio/Video Encoding | |
| iTunes | Version 10.4.1.10 x64: Audio CD (Terminator II SE), 53 minutes, default AAC format |
| Lame MP3 | Version 3.98.3: Audio CD "Terminator II SE", 53 min, convert WAV to MP3 audio format, Command: -b 160 --nores (160 Kb/s) |
| HandBrake CLI | Version: 0.98: Video from Canon Eos 7D (1920x1080, 25 FPS) 1 Minutes 22 Seconds Audio: PCM-S16, 48,000 Hz, Two-Channel, to Video: AVC1 Audio: AAC (High Profile) |
| TotalCode Studio 2.5 | Version: 2.5.0.10677: MPEG-2 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 |
| Productivity | |
| ABBYY FineReader | Version 10.0.102.95: Read PDF save to Doc, Source: Political Economy (J. Broadhurst 1842) 111 Pages |
| Adobe Acrobat X | Version 10.0.0.396: Print PDF from 115 Page PowerPoint, 128-bit RC4 Encryption |
| Autodesk 3ds Max 2012 | Version 14.0 x64: Space Flyby Mentalray, 248 Frames, 1440x1080 |
| Blender | Version: 2.64a, Cycles Engine, Syntax blender -b thg.blend -f 1, 1920x1080, 8x Anti-Aliasing, Render THG.blend frame 1 |
| Visual Studio 2010 | Version 10.0, Compile Google Chrome, Scripted |
| File Compression | |
| WinZip | Version 17.0 Pro: THG-Workload (1.3 GB) to ZIP, command line switches "-a -ez -p -r" |
| WinRAR | Version 4.2: THG-Workload (1.3 GB) to RAR, command line switches "winrar a -r -m3" |
| 7-Zip | Version 9.28: THG-Workload (1.3 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=5" |
| Synthetic Benchmarks and Settings | |
| 3DMark 11 | Version: 1.0.3, Benchmark Only |
| PCMark 7 | Version: 1.0.4 x64, System, Productivity, Hard Disk Drive benchmarks |
| SiSoftware Sandra 2013 | Version Version 2013.01.19.11, CPU Test = CPU Arithmetic / Cryptography, Memory Test = Bandwidth Benchmark |
3DMark shows the expected performance-to-price scaling. It costs around twice as much to boost performance by about half.
Many of PCMark’s tests favor drive performance, giving the $1300 and $2500 PCs advantages that rarely occur in real-world use. With that said, we also know that the benefits of an SSD are often difficult to quantify, since they're largely response-driven. At least we have a mechanism here for crediting the solid-state-equipped systems with higher scores.



Again, PCMark’s storage tests do a good job of representing program startup times, in spite of the fact that most folks spend a small portion of their days waiting for apps to fire up. That’s why we use the most-pertinent access patterns we could think of to make up only 10% of each system’s performance profile.
The $2500 PC takes a surprisingly large win over its $1300 counterpart in Sandra's Arithmetic component. Both processors possess the same number of physical cores. However, the addition of Hyper-Threading technology allows the Core i7 to post much higher numbers in this synthetic measure of compute performance. In the real world, SMT won't have such a pronounced effect. Then again, that's why this is a synthetic, and why we don't factor most in synthetics to our overall scoring.


The $1300 machine’s CAS 11 memory defaults do a good job of keeping up with the $2500 machine’s CAS 8 XMP configuration, though overclocking makes the big difference.
Unable to operate any faster, the $650 machine’s dual-channel memory controller is stuck at a still-good 20 GB/s.

Battlefield 3 is capped at 200 FPS, cutting into the $2500 machine’s overall performance leadership and hurting its potential value score (in spite of the fact that hardly anyone cares about average frame rates that are this high).
More interesting is the $650 PC’s ability to push three monitors at the Medium quality preset and 4800x900 pixels.


Stepping up to Ultra quality pushes the $650 PC out of triple-display playability, but the $1300 system still gets there. The $2500 machine's extra potential is being wasted, unless you want to run an even higher resolution or connect 120 Hz displays.
F1 2012 is easy on mid-range builds, so we’re not surprised to see the $650 machine producing solid numbers. It generates a perfectly-smooth 61 FPS average at maximum details and 4800x900.
Surprisingly, the $1300 PC beats my $2500 machine at 1600x900, though we're presenting these numbers purely for the sake of comparison. Nobody buys either of those configurations to play at such a low resolution.


Skyrim is the second completely processor-bound game in a row, and the third to suggest that my $2500 machine's hardware is excessive for the titles and settings we're using for comparison. Even the $650 build can do its job, adequately, at 4800x900 and Ultra details. The $1300 machine trails only slightly behind the most expensive build at everything except 4800x900, where it barely edges out the cheap PC.


Far Cry 3 is the only game in our suite capable of proving the $2500 machine’s necessity, as only its dual-GPU GeForce GTX 690 can push playable frame rates at 4800x900 and the Ultra quality preset. Even then, the machine needs to be overclocked. This game's frame rate bounces around enough that its 42 FPS average drops to sub-30-FPS choppiness in the toughest portion of the test.


As we know well by know, Apple's iTunes is going to favor the platform running the fastest. After all, these three machines all employ Intel's Ivy Bridge architecture. More cores and cache don't help at all.
LAME is largely the same, its workload also single-threaded.


In contrast, HandBrake and TotalCode Studio extract maximum benefit from my pricy setup's Hyper-Threading technology, which employs logical cores to utilize physical resources as fully as possible.


When Adobe After Effects is presented with a large number of physical and logical cores, it needs lots of system memory to allocate to each one. Because my Core i7 gives the application eight threads, I used this benchmark as an excuse to drop two 8 GB modules into my little $2500 machine. The experiment appears to pay off, and the priciest machine achieves the best performance.

Photoshop appears to prefer my system’s dual-GPU graphics card in its OpenCL test.

Premiere’s preference for expensive processors reflects the small but noticeable boost we’ve seen in a few other applications able to benefit from Hyper-Threading. The $2500 machine’s CPU is also clocked the highest, accounting for part of its lead.

A productivity application that’s associated with Adobe’s creativity suite, our PowerPoint-to-PDF conversion workload using Acrobat X is decidedly single-threaded. Nevertheless, Don's overclocked Core i5 secures a win even in light of its less aggressive frequency.

Other than Adobe Acrobat, on the previous page, our productivity suite reacts well to the application of Intel's Hyper-Threading technology. In fact, I even retested my system with SMT turned off in order to verify the magnitude of these victories and the feature responsible for them. This is a great place for the $2500 machine to outshine the $1300 build, though performance remains disproportionate to price.



As with most of our productivity applications, 7-Zip gets a huge boost from the $2500 PC’s Hyper-Threading capability. WinRAR shows a slighter preference, and WinZip follows in its CPU-only compression setting.
Performance is still disproportionate to price for all three systems, making Paul's $650 machine appear a top value, even if it's also the slowest contender.



Except for the GPUs, which sport different architectures, our power consumption chart reflects the notion that better performance necessitates more draw from the wall. That is to say, the highest score is actually the worst score, even though it’s also tied to higher performance. In the case of the graphics cards, Nvidia's Kepler design tends to offer higher frame rates at lower power than AMD's GCN.

The $1300 PC also runs exceptionally cool in its stock form. Conversely, if you end up building the $2500 configuration, you'll want to keep your room temperature below 34° C to prevent thermal throttling.

For some reason, Don didn't use the data from his storage benchmarks in his write-up. There it is, though, a notable improvement compared to last quarter. And of course, we know from your feedback that SSDs are a must-have in his mid-range build and above.
Since load times aren’t reflected in most of our benchmarks and represent only a small portion of the time we spend in front of our computers, that component of the suite makes up just 10% of our combined performance metric.

Using our slowest system as the baseline (100% performance) and offsetting the efficiency chart to 0% (by subtracting 100% from calculated values), we see that the $1300 PC draws 2% less power and provides 74% more performance. Prior to overclocking, it starts out as the most efficient configuration in this System Builder Marathon.

We approached this quarter’s builds with a semi-compact theme and the idea that small systems can perform just as well as full ATX-based platforms. But mini-ITX motherboards do cost more, simultaneously leaving us with fewer features to brag about. That extra premium is most bothersome at the bottom end of our pricing scale. Paul had to drop the optical drive from his machine to get close to his budget, but you wouldn't know that just by looking at his benchmark results. After all, an optical drive is one of those features that doesn't affect the final analysis, even if it's an important convenience.
When our builders aren't compelled to drop capabilities, and they use the same number of components to hit their respective price points, Don's mid-range machines historically win about as many value comparisons as Paul's entry-level builds. Don didn’t need to cut any of his features to maintain a Core i5 and GeForce GTX 680. Those parts did force him to hit his budget ceiling with only a 60 GB SSD in his shopping cart, though. That's enough for a boot drive. Done today, we think he would have gone with a GeForce GTX 770 and tried to get a 120 GB SSD in there somehow.

Our drive benchmarks are supposed to represent the entire user experience, so I technically could have docked Don half of his drive score for using an SSD that only fits half of our test suite. But he would have lost less than 10% of his total, and still taken second place in our value chart. Equipped with a Tahiti-based Radeon HD 7870 and a fairly modest dual-core processor, Paul’s $650 machine wins no matter how I adjust our performance weighting.

This was also the first time Paul put his budget gamer through the rigors of high-resolution testing, and he takes top value there too. Unfortunately, the frame rates generated by his machine at this resolution require lower-quality settings in both Far Cry 3 and Battlefield 3.
Don’s $1300 machine slaughters our Battlefield 3 benchmark and copes through our lower Far Cry 3 settings at 4800x900, but chokes when Far Cry 3’s details are turned up. Only the $2500 PC is sufficient in our toughest gaming test, and that's the only one that requires performance beyond the $1300 build's capability.