Sign in with
Sign up | Sign in
System Builder Marathon Q4 2013: System Value Compared
By ,
1. Let The (System Builder) Games Begin

System Builder Marathon, Q4 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 $800 Gaming PC
Day 2: The $1600 Enthusiast PC
Day 3: The $2400 Performance PC
Day 4: Performance And Value, Dissected

Introduction

Getting into gaming is rarely cheap, but Paul Henningsen thrives against the struggle to get top performance for bottom dollar. That stuggle was threatened this month by stagnation in the low-cost gaming hardware market. His choices were to rebuild the same machine he presented last quarter, to step down to an even cheaper machine that would give up more performance than cost, or spend extra money on a few mid-range parts. Paul is typically not interested in boring, so we were all on-board with his decision to cram a high-end Radeon R9 280X into a mainstream-priced $800 machine.

We like to use budgets that facilitate easy comparisons. So, we could have stretched Don in one of two directions: build a $1200 machine at 1.5 times the cost of Paul’s or really step things up and use $1600 to try building twice the machine. The $800 system's specs are so impressive that nothing short of a $600 graphics subsystem represents a higher market class. We made the call to give Don $1600 for his effort.

The decision was made easier by my own desire to piece together a configuration faster than last quarter's flagship System Builder Marathon box for less money. At 1.5 times the cost of a $1200 PC, an $1800 machine wouldn’t have packed in enough performance. At twice the cost of Don’s machine, a $3200 build would have forced me to waste money chasing miniscule gains. By splitting the price structure into three multiples of $800, a $2400 budget was just what I needed.

The final line-up shows two gaming machines and a configuration optimized for general-purpose computing, which should be capable of doing everything exceptionally well. Games are one of its strengths. Crypto-currency mining could be another, if you're willing to run your R9 290s at 100% load all of the time. And the six Ivy Bridge-based x86 cores should take care of everything else.

Q4 2013 SBM Components
 $800 Gaming PC$1600 Enthusiast PC$2400 Performance PC
ProcessorIntel Core i5-3570: 3.4 GHz, Quad Core, 6 MB Shared L3 CacheIntel Core i5-4670K: 3.4 GHz, Quad Core, 6 MB Shared L3 CacheIntel Core i7-4930K: 3.2 GHz, Six Cores, 12 MB Shared L3 Cache
GraphicsGigabyte GV-R928XOC-3GD: Radeon R9 280X 3 GB2 x MSI N770 TF 2GD5/OC: GeForce GTX 770 2 GB (SLI)2 x Asus R9290-4GD5: Radeon R9 290 4 GB (CrossFire)
MotherboardASRock Z75 Pro3: LGA 1155, Intel Z75 ExpressAsus Z87-PLUS: LGA 1150, Intel Z87 ExpressASRock X79 Extreme6: LGA 2011, Intel X79 Express
MemoryTeam Vulcan TLBD38G1600HC9DC01: DDR3-1600 C9, 8 GB (2 x 4 GB)Patriot PV38G186C9KR: DDR3-1866 C9, 8 GB (2 x 4 GB) G.SkillF3-14900CL9Q-16GBXL: DDR3-1600 C9, 16 GB (4 x 4 GB) 
System DriveWestern Digital WD10EZEX:1 TB SATA 6Gb/s HDDSamsung MZ-7PD128BW: 128 GB, SATA 6Gb/s SSDSanDisk SDSSDHP-256G-G25: 256 GB, SATA 6Gb/s SSD
Storage DriveUses System DriveSeagate ST2000DM001: 2 TB SATA 6Gb/s HDDWestern Digital WD20EZRX: 2 TB, SATA 6Gb/s HDD
OpticalLite-On iHAS124-04: 24x DVD±R, 48x CD-RLite-On iHAS124-04: 24x DVD±R, 48x CD-RPioneer BDR-208DBK: 15x BD-R, 16x DVD±R
CaseXigmatek CCC-AE37BS-U02NZXT GAMA-001BKFractal Design Define R4
PowerEVGA 100-B1-0500-KR: 500 W, ATX12V v2.3, 80 PLUS BronzeCORSAIR TX750 V2: 750 W, ATX12V v2.3, 80 PLUS BronzeSeasonic SS-850AM: 850 W, 80 PLUS Bronze
CPU CoolerIntel Boxed CoolerEnermax ETS-T40-TBThermaltake CLW0217
 $800 $1600 $2409

The prices above represent what we paid for these parts back in November. All of us were able to find our hardware within the budgets we defined for ourselves, though a $10 discount expired between the time I picked my parts and when the order went through. Short-term specials are the least of anyone's worries today, though. As a result of big price hikes, Paul’s machine jumped by $125 and mine went up $300 after the orders were placed. I’ll go over how this affects each builder in my conclusion.

2. Benchmark And Overclocking Configurations

Like it or not, Intel’s XMP technology provides overclocking modes for DRAM. Don left his system at its unconfigured defaults (SPD values) for baseline measurements and enabled XMP as his overclocked mode.

But unlike other overclocking options, XMP modes are part of the module’s rated performance. Paul and I chose to enable this feature for baseline measurements, and I was able to put my memory performance even higher using manual O/C methods.

Test Hardware Configurations
 $800 Gaming PC$1600 Enthusiast PC$2400 Performance PC
Processor
(Overclock)
Intel Core i5-3570: 3.4 - 3.6 GHz, Four Physical Cores, O/C to 3.8 - 4.0 GHz 1.26 VIntel Core i5-4670K: 3.4 - 3.8 GHz, Four Physical Cores, O/C to 4.30 GHz, 1.30 VIntel Core i7-4930K: 3.4 -
3.90 GHz, Six Physical Cores
O/C to 4.4 - 4.6 GHz, +140 mV
Graphics
(Overclock)
Gigabyte R9 280X: 1100 MHz GPU, GDDR5-6000, O/C to 1115 MHz GPU, GDDR5-64002 x MSI GTX 770: 1089 MHz GPU, GDDR5-6008, O/C to 1212 MHz GDDR5-75122 x Asus R9 290: 947 MHz GPU,  GDDR5-5000
O/C to 1100 MHz, GDDR5-5600
Memory
(Overclock)
8 GB Team Vulcan DDR3-1600 CAS 9-9-9-24, No O/C8 GB Patriot DDR3-1600 CAS 9-9-9-24 1T, O/C to DDR3-1866 CL 9-10-9-27, 1.50 V16 GB G.Skill DDR3-1866 CAS 9-10-9-28, O/C to DDR3-2133 CL 10-11-10, 1.585 V
Motherboard
(Overclock)
ASRock Z75 Pro3: LGA 1155, Intel Z75 Express
Stock 100 MHz BCLK
Asus Z87-PLUS: LGA 1150, Intel Z87 Express
Stock 100 MHz BCLK
ASRock X79 Extreme4:
LGA 2011, Intel X79 Express
Stock 100 MHz BCLK
OpticalLite-On iHAS124-04: 24x DVD±R, 48x CD-RLite-On iHAS124-04: 24x DVD±R, 48x CD-RPioneer BDR-208DBK: 15x BD-R
CaseXigmatek CCC-AE37BS-U02NZXT GAMA-001BKFractal Design Define R4 Black Pearl
CPU CoolerIntel Boxed CoolerEnermax ETS-T40-TBThermaltake CLW0217 Water 2.0 Extreme
Hard DriveWestern Digital WD10EZEX: 1 TB SATA 6Gb/s HDDSamsung MZ-7PD128BW: 128 GB, SATA 6Gb/s SSDSanDisk SDSSDHP-256G-G2 256 GB SATA 6Gb/s SSD
PowerEVGA 100-B1-0500-KR: 500 W, ATX12V v2.3, 80 PLUS BronzeCORSAIR TX750 V2: 750 W, ATX12V v2.3, 80 PLUS BronzeSeasonic SS-850AM: 850 W, 80 PLUS Bronze
Software
OSMicrosoft Windows 8 Pro x64
GraphicsAMD Catalyst 13.11 Beta  9.4Nvidia GeForce R331.93 BetaAMD Catalyst 13.12
ChipsetIntel INF 9.4.0.1017Intel INF 9.3.0.1026
Benchmark Configuration
3D Games
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
Grid 2Version 1.0.85.8679, DirectX 11, Built-in Benchmark
Test Set 1: High Quality, No AA
Test Set 2: Ultra Quality, 8x MSAA
Arma IIIVersion 1.08.113494, 30-Sec. Fraps "Infantry Showcase"
Test Set 1: Standard Preset, No AA, Standard AF
Test Set 2: Ultra Preset, 8x FSAA, Ultra AF
Far Cry 3V. 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 CS6Version 11.0.0.378 x64: Create Video which includes Three Streams, 210 Frames, Render Multiple Frames Simultaneously
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 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)
HandBrake CLIVersion: 0.99: 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, Two-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 11.0.0.379: Print PDF from 115 Page PowerPoint, 128-bit RC4 Encryption
Autodesk 3ds Max 2013Version 15.0 x64: Space Flyby Mentalray, 248 Frames, 1440x1080
BlenderVersion: 2.67b, 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: 1.0.3, Benchmark Only
PCMark 8Version: 1.0.0 x64, Full Test
SiSoftware Sandra 2013Version: 2013.01.19.50, Processor  Arithmetic,
Cryptography, Memory Bandwidth Benchmarks
3. Results: 3DMark And PCMark

Don and Paul call their systems gaming machines, yet 3DMark shines most brightly on my own 3D rendering wunderbox. That is, I wonder if I can use my dual Radeon R9 290s in place of a space heater and have P2Pool pay my utility bills?

PCMark likes solid-state storage, but not to the exclusion of Paul’s hard drive-based $800 PC. Its storage score is the only synthetic result used in our overall performance analysis.

4. Results: SiSoftware Sandra

Sandra's Arithmetic module exhibits a very interesting performance scale which, if we applied it to our overall value analysis, would show the $1600 and $2400 machines offering nice value. I’m actually hoping for even more of an advantage in our real-world benchmarks to put me over the top.

There's a correlation between memory bandwidth and performance in the AES-NI-accelerated encryption test. The quad-channel controller can feed the host processor even more instructions once I overclock.

Don gave up after enabling XMP as his $1600 PC’s memory overclock. Paul turned XMP on as a baseline for his $800 machine, but wasn’t able to overclock beyond that. I picked memory I knew I could overclock, and was rewarded by an offensive 30% price hike a month later.

5. Results: Battlefield 4

We were hoping that Battlefield 4 would tax all of our gaming configurations more intensively than Battlefield 3. However, the newer title still handicaps the performance-per-dollar outcome of the $1600 and $2400 PCs with a 200 FPS cap.

Fortunately, that problem goes away when we switch on the Ultra preset. The $1600 PC achieves 79% more performance than the $800 configuration (that is, for 100% more money).

6. Results: Arma III

The results in Arma III appear almost binary, depending on whether each system is using one or three monitors. I suspect that the game may be platform-bound, since the results using the Standard detail preset are tied almost perfectly to CPU clock rate. Once we step up to Ultra quality, more of the workload shifts towards the graphics limits of the two lower-end machines.

7. Results: Grid 2

We haven’t been benchmarking with Grid 2 long enough to know for certain whether the title is CPU- or memory-limited. But you probably don't care at this point; bottlenecks that happen up in the 150 FPS range are hardly detrimental to an enjoyable racing experience. But because these average frame rates show up in our performance-per-dollar calculation, they do matter.

It looks like we push past those bottlenecks using the Ultra preset at 4800x900 and higher.

8. Results: Far Cry 3

What happened to my $2400 machine to make it appear platform-bound in Far Cry 3? I searched for configuration problems and found none. Whatever caused my performance issues, the bottleneck only disappears when I apply the Ultra quality preset to triple-screen resolutions.

9. Results: Audio And Video Encoding

Our single-threaded iTunes and LAME workloads help illustrate the per-core benefits of overclocking without the influence of multiple cores.

In contrast, the HandBrake and TotalCode Studio benchmarks reflect architectural optimizations, frequency advantages, and core count. Unfortunately for me, two times the performance for three times the price isn't going to help my $2400 machine win any value comparisons.

10. Results: Adobe Creative Suite

Adobe After Effects shows the $2400 machine performing nearly twice as well as the $800 machine, boosting the value of the cheaper build. On the other hand, it punishes Don's $1600 PC for having the same core count as Paul's $800 PC.

Photoshop’s OpenCL-based filters prefer the $1600 machine’s Kepler architecture over the $800 PC’s GCN-based Radeon. It demonstrates even stronger disapproval for the $2400 build’s two Radeon R9 290s in CrossFire, though the same machine’s CPU does very well when switched to host-processing-heavy filters.

Premiere scales well with core count. Meanwhile, our PowerPoint-to-PDF test is single-threaded. In that benchmark, the most effective architecture operating at the highest clock rates is going to win.

11. Results: Productivity

3ds Max, Blender, FineReader, and Visual Studio all utilize multi-core configurations effectively. As a result, the $1600 machine is the one that appears most out of place. It costs twice as much as Paul's setup, but because it sports the same number of x86 cores, performance is only slightly better thanks to higher clock rates and an updated Haswell architecture.

These benchmarks do no favors for the $2400 PC's value story. Sure, my machine is nearly twice as fast as the $800 configuration. But I needed it to be three times faster to keep up with cost.

I catch a break in Visual Studio; the $2400 machine is a little more than twice as fast as Paul's effort. I’d like to credit a combination of six cores running at high clock rates and plenty of memory bandwidth for the advantage, though extra shared L3 cache is likely playing a role as well.

12. Results: File Compression

Dependent on both DRAM and cache performance, the results we get from 7-Zip are strikingly similar to what we saw in Visual Studio on the previous page. The $2400 PC’s extra shared L3 still can’t make up for its higher price though, when Paul's $800 PC sells for one-third as much and is half as fast.

WinRAR is notorious for scaling based on architecture and clock rate, giving the least-expensive setup a pronounced value advantage (even if it's the slowest contender). WinZip scaling is even tighter, crushing any hopes that my $2400 PC might achieve value parity.

13. Power And Heat

If the price tags on the $1600 and $2400 systems weren't already high enough, their power consumption figures might put your electricity bill over the edge. Of course, a straight value calculation is rarely the basis most enthusiasts use to make their purchase decisions, but I still get a kick out of showing how well sub-$1000 machines perform given their cost.

Paul and I are a little dubious of Don's 2°-above-ambient idle CPU reading. Perhaps he measured right after booting up?

The only two exceptional temperatures come from the GPUs in the overclocked $1600 and non-overclocked $2400 PCs. I used a custom fan setting to force my overclocked build's graphics temperature way down, although it resulted in a significant increase in noise.

14. Overall Performance And Efficiency

The $1600 PC enjoys a 34% lead over the $800 machine in gaming performance, which is particularly significant since this quarter's System Builder Marathon targets gaming.

My $2400 PC enjoys a similar lead, but productivity is where its pricey Ivy Bridge-E-based CPU makes the most difference. Also, average performance doesn't include the results at 5760x1080, since Paul didn't test his $800 machine at that resolution.

The $1600 PC overclocks far better than the $800 system. Once again, we see even larger overclocking gains from my $2400 submission, even if its price continuously overwhelms a strict comparison of performance value.

Power use also overwhelms performance on the $1600 and $2400 machines, but by less than expected given that shocking power consumption chart on the previous page. Low-voltage overclocking helps the $800 PC gain over its stock configuration.

The above chart is zeroed out by subtracting 100% from the baseline, so that it doesn't show any PC being more than 100% efficient. Remember that these efficiency ratings are relative to the stock $800 PC baseline, and that the actual efficiency of a machine doing no physical work is zero.

15. Who Wins The Value Comparison?

I already know that my $2400 machine is no more than twice as fast as Paul's $800 effort, and that Don's $1600 configuration falls in the middle.

I also know from reading Paul's Day 1 coverage and writing my Day 3 story that each machine achieves higher value than its predecessor.

Don's previous mid-priced build achieved 70% the value of Paul's low-priced counterpart in September. Don's current build drops to 64% the performance-value of Paul's system. Meanwhile, I climbed from 46% to a much-improved 51%. In other words, Paul wins by more compared to Don, but I lose by less compared to Paul. Considering the traditional low value of high-end parts, I’ll gladly accept that step up.

Those calculations assume you're paying our purchase price, though. What happens if we build the same machines today?

Don’s $1600 machine only went up $5, while my $2400 build increased $300. A smaller $125 price increase hurts Paul’s $800 PC disproportionately, since that's such a significant chunk of the box's original cost. Given the unexpected price hike on AMD's graphics cards, I can only guess that Don's Magic 8-Ball said “Nvidia”. The outcome is that Don gets an unexpected 10% value surge, even though Paul still wins.

Personally, I cut my losses by 2% literally. But that doesn't mean I’m figuratively ready to do the same thing. I like this machine, after all!

A look at the high-resolution gaming performance numbers reveals one reason why I like this machine so much. This work-friendly PC solidly stomps on Don’s game-oriented build, losing the value comparison only by about the difference in CPU prices. And my CPU also proved capable of doing about 70% more work than Don's. So yeah, I'll take that too.

Paul wins as usual, but by less than last time. And his PC is the best value for anyone who can live with it. Builders with the money to spend and the monitors to game on will instead want to choose my $2400 PC. And anyone who wants really great gaming performance at 5760x1080 without breaking the bank should take a closer look at Don’s $1600 machine. But most of you already drew the same conclusions, so I'll turn this Marathon over to the guys giving these three systems away.