Sign in with
Sign up | Sign in
System Builder Marathon, Q1 2013: $1,000 Performance PC
By ,
1. Can $1,000 Buy A High-End PC?

System Builder Marathon, Q1 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 $600 Gaming PC
Day 2: The $800 Enthusiast PC
Day 3: The $1,000 Performance PC
Day 4: Performance And Value, Dissected
Day 5: The $1,600 Alternative PC

Introduction

I typically approach our highest-priced performance build as an attempt to squeeze value out of top-shelf parts by prioritizing performance. Versatility takes a somewhat-distant secondary position; really, I want to make the most expensive configuration feel like it's worth what I'm paying for the parts. After all, when you spend thousands of dollars on a do-everything PC, you expect it to actually do everything.

Generally, though, when we analyze performance-per-dollar for each System Builder Marathon machine on Day 4 (the story coming tomorrow), our top setup tends to suffer compared to more affordable builds, since I often use nicer parts that don't help performance. 

As you know, though, this quarter's angle involves a tight price spread. I'm most affected by this, and my distinctly high-end $2,000 budget gets cut to $1,000, putting it in the mainstream category (or, at least the mid-priced market enthusiasts are used to). Gone are high-capacity SSD and hard disk combinations, as well as most CrossFire, SLI, or even Core i7 configurations. We even need to skip out on a quiet case, a huge CPU cooler, extra RAM, and a Blu-ray drive.

We have the easiest time pushing additional performance in games, so this quarter's more versatility-oriented creation looks uncomfortably similar to last year's mid-end machine (Ed.: Little did Thomas know when he ordered these parts, his system is really, really similar to what Don published yesterday).

Q1 2013 $1000 PC Components
ProcessorIntel Core i5-3570K: 3.4 GHz Base Clock Rate, 3.8 GHz Turbo Boost, 6 MB Shared L3 Cache$230
GraphicsPowerColor PCS+ AX7870 Myst Edition 2GBD5-2DHPPV3E $240
MotherboardASRock Z77 Extreme4: LGA 1155, Intel Z77 Express$135
MemoryCrucial Ballistix Tactical BLT2K4G3D1608ET3LX0: DDR3-1600 C8, 8 GB (2 x 4 GB)$48
System DriveMushkin MKNSSDCR240GB-DX: 240 GB, SATA 6Gb/s SSD$180
Storage DriveUses System Drive-
OpticalLite-On iHAS124: 24x DVD±R, 12x DVD±R DL$17
CaseRosewill Redbone U3: USB 3.0, eSATA, 3 x 120mm$45
PowerAntec Neo Eco 520C 520 W, ATX12V v2.3, 80 PLUS-Certified$55
CPU CoolerCooler Master Hyper 212 Plus RR-B10-212P-G1$30
  Total Cost $980


On a per-gigabyte basis, hard drives are really cheap. But adding secondary storage would have cut deeply enough into our budget that we would have needed to drop from 240 to 120 GB of solid-state storage. I simply couldn’t find a 120 GB SSD I liked enough to make that compromise, and the lack of tiered storage makes this system even more gamely.

When in gaming territory, I do what gamers do: focus on graphics performance. Thankfully, AMD quietly started shipping its Tahiti LE-equipped Radeon HD 7870 not too long ago, and we reviewed it in Tahiti LE, Tested: PowerColor's HD7870 PCS+ Myst Edition. We had enough time before ordering our parts to check out its performance and decide it'd be appropriate here.

2. Graphics, CPU, And Motherboard

Video Card: PowerColor PCS+ AX7870 Myst Edition

AMD’s Tahiti LE GPU hit the PC gaming scene hard with shocking value from three launch partners, all of which were safe bets in my experience. Somehow, the only model in stock on the day we placed our order from Newegg was the same one we tested for that launch, PowerColor’s PCS+ AX7870 Myst Edition. Clocked at 925 MHz core (975 MHz Boost) and with 2 GB of GDDR5-6000, my only reservation about the recommendation Igor gave the card was PowerColor's relatively brief two-year warranty coverage.

Read Customer Reviews of PowerColor's PCS+ AX7870 Myst Edition


I was also forced to set aside my reservations about axial-flow fans that recirculate exhaust air inside a chassis, since no Tahiti LE-based cards were available with centrifugal coolers. That's probably just as well, though, since I couldn't afford a quiet-enough case to stifle the added noise most of those blower-style designs generate.

CPU: Intel Core i5-3570K

Intel’s Core i5-3570K gives up around 3% of the Core i7-3770K’s clock rate, a little shared L3 cache, and Hyper-Threading in return for around 30% cost savings, making it a clear value play.

Read Customer Reviews of Intel's Core i5-3570K


Lacking the budget for a higher-end Core i7 processor, I'm hoping that most of our benchmarks don't punish me for giving up the ability to schedule to eight threads simultaneously.

Motherboard: ASRock Z77 Extreme4

Weary of the weak voltage regulator found in our previous build, we decided to take a chance with an less expensive board that advertises a few extra voltage phases. Phase count isn’t nearly as important as total amperage capacity, but at this price, we were forced to stick with a known part that'd be good enough or gamble on something that might have worked better (but maybe not).

Read Customer Reviews of ASRock's Z77 Extreme4


We gave ASRock’s Z77 Extreme4 a similar rating as Gigabyte’s Z77X-D3H, though our round-up’s overclocking test was performed at a mere 1.25 V processor voltage setting. The Z77X-D3H later proved itself incapable of supplying stable current at voltage levels beyond 1.25 V, and we were shooting for around 1.28 V when we discovered that problem. We’ll “go there” again today.

3. DRAM, Storage, And Optical Drive

Memory: Crucial Ballistix Tactical 8 GB (2 x 4 GB) DDR3-1600 Memory Kit

Over the past couple of years, several readers have shot over suggestions to try out 1.35 V DIMMs for overclocking, citing their scalability. Our own experiences often contradicted those recommendations, though. Many older modules simply couldn’t tolerate the extra voltage needed to put them over the performance curve of mid-range parts running at standard voltages. Like any other step forward in technology, however, the low-voltage stuff is significantly improved today.

Read Customer Reviews of Crucial's BLT2K4G3D1608ET3LXO 8 GB Kit


Crucial’s Ballistix Tactical LP modules are among the recent generation of low-voltage RAM that survives the 1.5-1.6 V needed to overtake similarly-priced 1.65 V champions. Furthermore, the slightly lower overclocked voltage is also a little easier on our CPU's memory controller.

Speaking of its CPU-oriented advantages, the reduced height of these low-profile modules makes extra room for oversized heat sinks, too.

SSD: Mushkin Chronos Deluxe MKNSSDCR240GB-DX

We like the performance and price-per-gigabyte of Mushkin’s 240 GB DX-series SSD, but dropping to 120 GB would have been necessary to make room for a conventional disk. After considering all of my somewhat-limited hard drive options, I gave in and kept the big SSD.

Read Customer Reviews of Mushkin's 240 GB MKNSSDCR240GB-DX


The 240 GB DX remains storage editor Andrew Ku’s top performance-value pick thanks to its SandForce controller, Toggle-mode NAND, and competitive price.

Optical Drive: Lite-On iHAS124

Dependability and price drove our optical drive selection, with the iHAS124’s 24x burn performance considered nothing more than an added feature. We might have even picked a read-only drive to load our software, if not for the fact that read-only drives cost nearly as much.

Read Customer Reviews of Lite-On's iHAS124


At this price, the question of whether people still burn DVDs is little more than a diversion. Even though I can occasionally find 8 GB thumb drives for only a few dollars, I’m still more likely to pass around the cheaper DVD media to friends and family.

4. Case, Power, And CPU Cooling

Cases that include power supplies are my first choice when building on a budget. For instance, Antec's NSK 4480B comes with the company's high-efficiency EarthWatts 380, giving us a solid PSU and a 0.8 mm-thick steel chassis for only $100. But I wasn’t confident that a 380 W power supply would be enough in thise situation.

Case: Rosewill Redbone U3

Slightly thinner (and consequently more prone to flex) than Antec’s solution, Rosewill’s Redbone U3 saves us a bit of cash that we plan to spend on a beefier power supply. And, unlike competitively-priced solutions, the U3’s front-panel USB 3.0 connectors give me the I/O I've been requiring from all case review submissions lately.

Read Customer Reviews of Rosewill's Redbone U3


With three included fans, the Redbone U3 also offers potential cooling advantages over its competition. That makes us feel a little better, particularly in light of the CPU and GPU cooling limitations we're facing at the $1,000 price point (remember, I'm used to competing with two grand, at least).

Power Supply: Antec Neo Eco 520C 520 W

An 85% efficiency rating qualifies Antec’s Neo Eco for an 80 PLUS Bronze certification, though the company doesn’t apply the Bronze award to this part. It still gives us the added wattage we need to satisfy our maximum theoretical load requirements.

Read Customer Reviews of Antec's Neo Eco 520C


According to 80 PLUS, it takes around 630 W of draw from the wall to generate this unit's 520 W output rating, due to energy lost as heat during conversion. That’s really important to remember, since input power is also used in our efficiency calculations.

CPU Cooling: Cooler Master Hyper 212 Plus

Read Customer Reviews of Cooler Master's Hyper 212 Plus


Cooler Master’s familiar Hyper 212 Plus doesn't necessarily qualify as a high-end performer, but instead provides adequate cooling performance at a great price. Its capabilities make it a great match to a moderately-efficient CPU like the Ivy Bridge-based Core i5 in today’s build.

5. Hardware Installation

Cooler Master’s Hyper 212 Plus uses standoffs atop the motherboard to secure a support plate beneath the board. An included socket allows those who lack sufficient tools to tighten its nuts with a screwdriver. The cooler’s top bracket is then attached to those standoffs using spring-loaded screws.

Note that the screws have springs on both sides, with shafts that narrow below the bracket. These must be pulled outward in order to change their alignment within the bracket’s three mounting positions (LGA 755, 1155, 1366).

Rosewill’s Redbone cases (including the tested U3 version) do not support 2.5” drives. Fortunately, Mushkin’s Chronos Deluxe includes a 3.5” adapter tray.

Rosewill’s Redbone cases also lack any cable management features. Front-panel cables were too short to even tuck between the motherboard and its tray. We had to settle for a messier installation than I'd tolerate in a $2,000 build, using cable ties to secure them out of the airflow path.

Our CPU fan prevents the side fan from fitting in its upper mount, and the lower fan mount is blocked by any full-sized power supply. Fortunately, the vent holes are spaced close enough to the fan’s screw holes to mount it in the middle, just below the CPU cooler, where it feeds cold air to both the CPU and graphics card.

6. Overclocking

ASRock includes several handy integrated overclocking profiles, including 4.6 and 4.4 GHz. The lower of those two profiles worked most of the time, switching core voltage between 1.05 and 1.28 V under various loads. That would have been perfect, except that the system did occasionally (and rarely) crash. We decided to use manual settings to achieve similar results, without any occasional instability.

Our memory willingly complied with an increase from its standard 1,600 MT/s data rate to DDR3-2133. Further down the BIOS page, we set a fixed CPU core of 1.28 V.

When paired with the Core i5-3570K CPU, the Z77 Extreme4’s “Level 1” Load-Line Calibration provided an extremely stable voltage range between the set 1.28 and a maximum of 1.296 V. Keeping the processor below 1.30 V provides the added insurance of longevity.

We were actually able to hold 4.5 GHz for extended benchmarking sessions, without overheating, but again faced occasional and almost mysterious instability. Unwilling to push past 1.30 V, we settled for 4.40 GHz.

We would have expected low-profile 1.35 V DIMMS to get hot at 1.60 V, but Crucial’s Ballistix Tactical LP had no such issues. That extra voltage helped us retain stock 9-9-9-24 timings at this 33% overclock, edging out the CAS 10 timings we achieved last quarter using standard-voltage modules at the same DDR3-2133 data rate.

PowerColor’s Tahiti-LE-equipped Radeon HD 7870 reached 1,200 MHz GPU and GDDR5-6400 fairly easily, though we did need to change the maximum fan speed temperature from 90° to 80° Celcius. Running short of time for fine-tuning, we tried adding 50 MHz, only to have one of our games crash.

7. Test Settings And Benchmarks

Test Hardware Configurations
 Q1 2013 $1000 PCQ4 2012 $2000 PC
Processor
(Overclock)
Intel Core i5-3570K
3.40 GHz, Four Physical Cores
O/C to 4.40 GHz, 1.28 V
Intel Core i7-3770K
3.50 GHz, Four Physical Cores
O/C to 4.40 GHz, 1.26 V
Graphics
(Overclock)
PowerColor 2GBD5-2DHPPV3E: 975 MHz GPU,  GDDR5-6000
O/C to 1,200 MHz GDDR5-6400
2 x MSI R7970-2PMD3GD5/OC: 1,010 MHz GPU,  GDDR5-5500
O/C to 1,125 MHz GDDR5-6300
Memory
(Overclock)
8 GB Crucial DDR3-1600
CAS 8-8-8-24, O/C at 1.50 V
to DDR3-2133 CL 9-9-9-24
16 GB G.Skill DDR3-1600
CAS 8-8-8-24, O/C at 1.60 V
to DDR3-2133 CL 10-11-10-24
Motherboard
(Overclock)
ASRock Z77 Extreme4:
LGA 1155, Intel Z77 Express
Stock 100 MHz BCLK
Gigabyte GA-Z77X-D3H:
LGA 1155, Intel Z77 Express
Stock 100 MHz BCLK
OpticalLite-On iHAS124 24x DVD±RAsus BW-12B1ST 16x BD-R
CaseRosewill Redbone U3Cooler Master Storm Enforcer
CPU CoolerCooler Master Hyper 212 PlusCooler Master Hyper 212 Evo
Hard DriveMushkin Chronos Deluxe DX 240 GB, SATA 6Gb/s SSD
PowerAntec Neo Eco 520C: ATX12V v2.3, 80 PLUSCorsair HX750: ATX12V V2.3 80 PLUS Gold
Software
OSMicrosoft Windows 8 Pro x64
GraphicsAMD Catalyst 13.1AMD Catalyst 12.10
ChipsetIntel INF 9.3.0.1026Intel INF 9.3.0.1020


The overclocks between this month’s Core i5 and our previous Core i7 builds are surprisingly similar, even though the previous build’s CPU was good enough to run 4.40 GHz at a lower voltage. The overclocking limit in our previous build appeared to be a hot voltage regulator, while the overclocking limit of today’s build appears to be nothing more than a builder too stubborn to use even slightly more than 1.30 V CPU core over the long term. We've seen too many chips die in the lab, unfortunately.

We dropped StarCraft II from the rest of this Marathon. But, lacking the previous build on which to run its replacement, I’ve given it an encore in today’s tests.

Benchmark Configuration
3D Games
Battlefield 3Campaign 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
Version 1.2, Direct X 11, Built-in Benchmark
Test Set 1: High Quality, No AA
Test Set 2: Ultra Quality, 8x MSAA
The Elder Scrolls V: SkyrimUpdate 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
StarCraft IIV1.5.1, "Tom's Hardware Guide V2" custom map, 60-Second Fraps
Test Set 1: High Details, High Quality
Test Set 2: Ultra Details, Extreme Quality
Adobe Creative Suite
Adobe After Effects CS6Version 11.0.0.378 x64: Create Video which includes three Streams, 210 Frames, Render Multiple Frames Simultaneosly
Adobe Photoshop CS6Version 13 x64: Filter 15.7 MB TIF Image: Radial Blur, Shape Blur, Median, Polar Coordinates
Adobe Premeire Pro CS6Version 6.0.0.0, 6.61 GB MXF Project to H.264 to H.264 Blu-ray, Output 1920x1080, Maximum Quality
Audio/Video Encoding
iTunesVersion 10.4.1.10 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)
HandBrake CLIVersion: 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 2.5Version: 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, 2 Channel, 16-Bit, 224 Kb/s), Codec: H.264 Pro, Mode: PAL 50i (25 FPS), Profile: H.264 BD HDMV
Productivity
ABBYY FineReaderVersion 10.0.102.95: Read PDF save to Doc, Source: Political Economy (J. Broadhurst 1842) 111 Pages
Adobe Acrobat XVersion 10.0.0.396: Print PDF from 115 Page PowerPoint, 128-bit RC4 Encryption
Autodesk 3ds Max 2012Version 14.0 x64: Space Flyby Mentalray, 248 Frames, 1440x1080
BlenderVersion: 2.64a, Cycles Engine, Syntax blender -b thg.blend -f 1, 1920x1080, 8x Anti-Aliasing, Render THG.blend frame 1
Visual Studio 2010Version 10.0, Compile Google Chrome, Scripted
File Compression
WinZipVersion 17.0 Pro: THG-Workload (1.3 GB) to ZIP, command line switches "-a -ez -p -r"
WinRARVersion 4.2: THG-Workload (1.3 GB) to RAR, command line switches "winrar a -r -m3"
7-ZipVersion 9.28: THG-Workload (1.3 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=5"
Synthetic Benchmarks and Settings
3DMark 11Version: 1.0.1.0, Benchmark Only
PCMark 7Version: 1.0.4 x64, System, Productivity, Hard Disk Drive benchmarks
SiSoftware Sandra 2011Version Version 2013.01.19.11, CPU Test = CPU Arithmetic / Cryptography, Memory Test = Bandwidth Benchmark
8. Results: 3DMark And PCMark

Remembering that our current build costs half as much as its predecessor, it only really needs to perform half as well to achieve equal value. We’re hoping for something a lot better than that though, and some of what we're looking for is demonstrated right out of the gate in 3DMark.

In our first synthetic test, the old build costs nearly 100% more, yet performs only 60% better.

PCMark leans heavily on storage testing. Because we're using the same SSD from Mushkin that we selected last quarter, both systems fall into the same performance range. The former $2,000 build is the fancy house next door to the $1,000 configuration in this test.

Ten percent of our value score is storage-based, and those calculations reference the three PCMark tests that we believe are most appropriate. Using the same drive for both systems once again gives our little PC a big value boost.

9. Results: SiSoftware Sandra

Something appears horribly wrong with our current build in Sandra’s Arithmetic module, but editor-in-chief Chris Angelini explains that the benchmark gets a big boost from the previous build’s Hyper-Threading capability.

Sandra attempts to capture theoretical performance, rather than real-world differences. I never would have imagined, though, that a few extra registers in a CPU core could make such a huge difference, even in a synthetic.

Sandra's Cryptography sub-test doesn't benefit from Hyper-Threading. Instead, it's bottlenecked by our memory subsystem, which feeds data as fast as it can to both AES-NI-equipped processors.

Our new build’s memory provides similar overclocking capability to its predecessor, but with slightly tighter timings that give it a small advantage in Sandra's Memory Bandwidth benchmark.

10. Results: Battlefield 3 And F1 2012

Battlefield 3 is capped at 200 FPS, and its 2560x1600 results appear somewhat CPU-limited at the Medium quality preset. Both of those things inflate our cheaper system’s apparent value in spite of the performance inferiority of its single graphics card.

Though our previous build reaches amazing performance heights at moderate resolutions, the cheap new system remains competent throughout Battlefield 3’s Ultra Quality preset.

A recent investigation that demonstrated F1 2012’s memory bottleneck using hard data also explains how the new single-GPU system beats the former dual-GPU champion at five of its eight test settings. Though the old configuration begins to pull away as we dive deeper into Ultra Quality testing, this quarter's cheaper build plays through those tougher settings with aplomb.

11. Results: The Elder Scrolls V: Skyrim And StarCraft II

Our previous dual-GPU build appeared CPU-limited through Skyrim’s High Quality preset, while the new system’s single GPU bumps up against the same bottleneck only at our lowest test resolution. Because stereoscopic displays encourage you to pursue 120 Hz, anyone interested in those will also favor the more expensive build.

The frame rate difference between Skyrim’s High and Ultra quality presets is of little consequence to this class of hardware, allowing the cheap system to remain playable through 2560x1600.

Our previous build uses big StarCraft II numbers to bully the current cost-cutter. However, the cheaper system maintains its composure with ultra-smooth frame rates at 2560x1600.

12. Results: Audio And Video Encoding

Apple's iTunes and LAME MP3 are still single-threaded workloads. So, they pretty accurately reflect the architectural similarities, and close stock and overclocked frequencies, shared between our $1,000 and $2,000 builds. At the end of the day, if you're using an Ivy Bridge-based processor, you won't see any better single-threaded performance unless you can overclock more aggressively.

On the other hand, HandBrake and TotalCode Studio both appear to benefit from the previous build’s logical cores. I haven’t seen Hyper-Threading provide this much benefit since Intel shifted away from its NetBurst architecture.

13. Results: Adobe Creative Suite

Adobe After Effects and Photoshop both appear to gain little from the Core i7 machine's Hyper-Threading technology.

Unfortunately, our OpenCL-accelerated Photoshop numbers aren't quite right due to a check-box issue that kept our last config from properly enabling this API. The CPU-based tests are correct, though, and they show just how close both platforms really

Premiere Pro is a well-threaded test, so we'd expect it to benefit from last quarter's Hyper-Threading technology. On the other hand, our Acrobat workload is not, so the fact that it ran better last quarter as well suggests gains from elsewhere (a larger L3 cache, perhaps). In both cases, the older machine leads at similar clock rates.

Because it’s not a content creation-oriented application, Acrobat X will be added to the “Productivity” results of our overall performance scores.

14. Results: Productivity

Like most of our other professional applications, 3ds Max, Blender, and ABBYY FineReader all benefit to some degree from the Core i7's Hyper-Threading technology, which more completely utilizes physical cores in workloads able to benefit from parallelism.

The advantage quantified in our Google Chrome compile workload is particularly pronounced.

15. Results: File Compression

All three of these archival apps are optimized for processors with multiple cores, but to different degrees.

7-Zip, for example, is known to be really well-threaded. It's certainly nice to overclock and see substantial performance gains, but the ability to schedule to eight threads concurrently via Hyper-Threading does far more for our benchmark.

WinRAR, on the other hand, cares little about the Core i7's extra features. Instead, it appears to respond to clock rate. And the fact that our two contenders run at similar frequencies means they show similarly.

Finally, WinZip's CPU and OpenCL-accelerated benchmarks get notable gains from CPU and GPU overclocking.

16. Power, Heat, And Efficiency

The presence of only one graphics card gives our current build a big advantage in power consumption. It also proves that we were wrong to assume we'd need more than a 380 W power supply. Our measurements apply to the complete build, including energy lost as heat within the PSU. The 400 W maximum input reflected for our tuned configuration corresponds to a 340 W output at 85% efficiency.

Our current build’s CPU also runs much cooler at 4.40 GHz compared to last quarter's overclocked build. This is in spite of the previous build’s use of a nearly identical CPU cooler, the same core architecture, and a lower overclocked voltage. We could credit part of the current system’s relatively cool operation to our own side-fan placement optimization, but we also know that Hyper-Threading facilitates better utilization. So, doubling up on Prime95 threads could very easily tax that platform more acutely, hurting thermal performance.

Since our new system is the competitor, we used the previous build’s stock performance as the baseline for our calculations. The new build starts out 25% slower, and its overclocked configuration finishes 22% slower than Q4's tuned machine.

The purpose of calculating average performance here is that it allows us to compare average power in our efficiency chart. Those calculations result in a baseline of 100%. But we subtract 100% from all of those results, since nothing can be more than 100% efficient. The resulting chart shows how far each alternative configuration deviates from that baseline.

Lower energy consumption allows today's build, which is marginally slower, to establish big gains in energy efficiency, gaining 49.5% in stock trim and 35.6% when overclocked. The reason the overclocked configuration picks up less efficiency is because its power consumption goes up faster than performance.

17. Could We Have A Value Winner At $1,000?

This month, I was forced to give up $1,000 worth of my budget to create a tighter comparison when we get the chance to break down value. But last quarter's $2,000 setup won't be a part of that story. So, I'll draw my comparisons now. We know that it cost twice as much and came nowhere near delivering two times as much performance. This is our chance to quantify the difference.

Serving up 83% of the previous build's performance, today’s half-priced machine gets a 61% value boost from its lower price. Both systems gain value from free overclocking, which is technically free. However, this quarter's overclocked value leads our previous effort by around 66%.

Gaming value becomes even more important when a PC is built expressly for that purpose. We’d like to say that today’s build turned into a capable gaming platform by accident, but we always knew that leaving out enhancements that don't affect performance would limit the utility of this machine to the most performance-sensitive workloads.

With that in mind, Q4 2012's advantage in games is larger than its overall performance lead (when we factor in the other benchmarks). But that advantage is still not large enough to overcome its astronomical increase in price.