System Builder Marathon, Q1 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 $2400 Reader's Choice PC
Day 2: Our New Enthusiast PC
Day 3: The $750 Gaming PC
Day 4: Performance And Value, Dissected
Introduction
When the System Builder Marathon team started talking about giving each of us our own choice for the theme of this quarter's competition, I knew exactly the direction I wanted to go. I’d resist the temptation to spend more money on enthusiast-oriented part, even forgoing an SSD for snappier boot times. Rather, I’d fine-tune my previous effort and build a formidable yet affordable pure gaming box.
The $650 Gaming PC from Q3 2013 was designed to do all things well, without compromising the machine’s gaming prowess. I used an AMD FX-6300 CPU and Nvidia GeForce GTX 760 to get there. For the money, both components continually earn our recommendations.
Last quarter, to avoid repeating the same basic build, I was allotted an additional $150, which bought me a Radeon R9 280X graphics for higher-resolution gaming, plus an Ivy Bridge-based Core i5 for better overall performance. The resulting $800 Gaming PC was not only successful for bolstering benchmark scores across the board, but it even trumped the previous quarter's stellar bang-for-the-buck value.

I faced a problem, though. Duplicating those efforts this quarter would have cost me nearly $940 as a result of steep price hikes on AMD's Radeon cards, among other increasing expenses. So, my top priority was honing in on gaming, trimming as much of the prior build's fat as possible. Could I maintain the same performance in games at a more reasonable bottom line?
My first order of business was axing the Radeon R9 280X. Fortunately, prices on that card are stabilizing a bit today. But my options when we placed our orders were more limited. Maintaining the graphics horsepower to match my previous effort would require nothing shy of a $330 GeForce GTX 770, though.
Although I recommend Intel's Core i5 if you can afford it, stepping down to a less-fancy processor then allowed me to shave an easy $60 from the system's cost without giving up alacrity in my favorite games. AMD's FX-6300 would have been a good fallback. Or, I could have pulled from Intel's similarly-priced Core i3 family. A three-module CPU based on the Piledriver architecture would have probably been a better overall value option. But unless I added the expense of a cooler and more overclocking-friendly motherboard, I'd get higher frame rates from a Core i3-4130. High efficiency and locked-out overclocking meant I wouldn't need to dump extra cash into a beefy motherboard, either. Any stable H81-based board would work. I'd sacrifice features and use the savings to help offset the 30% premium charged for 8 GB of memory.
Nailing down my ideal list of components required an almost comical amount of tuning to hit $750. Significant fluctuations in pricing and availability stymied my efforts throughout the day our orders were to be placed. In fact, every single component I selected, aside from the Core i3, was eventually swapped out (in some cases, multiple times), including no less than four different GeForce GTX 770s at $330. When it came time to submit, I lucked out and snagged Zotac's offering for $20 less than any of the competing cards.
As my build came together, the team discussed introducing a new twist. Based on reader input, plus our desire to build nicer boxes, we decided we wouldn't factor in the price of the chassis or optical drive into the final equation. Don and I were free to venture beyond cheap $40 enclosures without giving up our performance parts, while Thomas would no longer be penalized for holding his build to a higher standard of form and function. We could also avoid debating whether the expense of a DVD or Blu-ray burner should be spent elsewhere.
| Component | Model | Purchase Price |
|---|---|---|
| CPU | Intel Core i3-4130 (Haswell) | $130 |
| CPU Cooler | Intel Boxed Heat Sink and Fan | 0 |
| Motherboard | Asus H81M-K LGA 1150 Intel H81 Express | $58 |
| RAM | Adata XPG V2 8 GB (2 x 4 GB) DDR3-1600 AX3U1600W4G9-DGV | $70 |
| Graphics | Zotac ZT-70301-10P GeForce GTX 770 2 GB | $320 |
| Hard Drive | Western Digital Blue WD10EZEX 1 TB | $65 |
| Case | Rosewill Line-M MicroATX Mini Tower | $50 |
| Power | Rosewill Capstone-450-M 450 W ATX | $60 |
| Optical | Asus 24x DVD Burner DRW-24B1ST/BLK/B/AS | $20 |
| Total Price | $773 |
Although I was free of previous budgetary constraints, I didn't want to go crazy on the enclosure. After all, my goal was to build a more affordable gaming system. And $750 in hardware is no small chunk considering that you also need to add Windows, a 1080p display, and peripherals. At the last minute, however, I spent an extra $23 tweaking the case and power supply to match my own personal tastes. So let's have a look at the parts I picked.

Processor: Intel Core i3-4130
Time and again, we've seen Intel's Core i3 processors prove themselves in most budget-oriented gaming builds. Although the fourth-gen Core i3-4130 cannot be overclocked, high per-clock performance and a respectable 3.4 GHz operating frequency make this chip a worthy foundation on which to build an affordable PC. Its pair of Hyper-Threaded cores is capable of scheduling four threads at a time, putting it out in front of Intel's dual-core Pentium processors in our favorite AAA titles.

Read Customer Reviews of Intel's Core i3-4130
CPU Cooler: Intel Retail Boxed Heat Sink & Fan
Once we drop under Core i5-class CPUs, Intel stops outfitting its bundled heat sinks with copper slugs. Instead, this familiar-looking cooler consists of a low-profile aluminum orb-style sink, a PWM-controlled fan, and a push-pin mounting bracket.
It's sufficient for what we need it to do, and even under full-load, the fan remains fairly quiet. Best of all, it doesn't pull funds away from my performance-oriented parts.

Motherboard: Asus H81M-K
Trimming down platform costs was crucial to meeting my goals, and an inexpensive microATX H81 Express-based motherboard represents the entry point to gaming with a fourth-gen Core processor. Considering that my CPU's multiplier is fixed, I don't need high-end features or future-looking expansion from my motherboard. I only want something reliable, freeing up the cash for a beefy graphics card and 8 GB of memory.
Because entry-level parts aren't reviewed as often, I picked Asus' H81M-K due to its popularity on Newegg and positive feedback on community forums.

Read Customer Reviews of Asus' H81M-K
So what exactly does H81 force you to sacrifice? To begin, there is no overclocking or multi-card support. You also lose Intel’s Rapid Storage and Smart Response Technology. We retain dual-channel DDR3-1600 compatibility, but only across two modules. And although it natively supports USB 3.0 and SATA 6Gb/s, connectivity is limited.
Memory: 8 GB Adata XPG V2 DDR3-1600 AX3U1600W4G9-DGV
In order to get 8 GB into this build, I had to choose between the most affordable kits (after all, prices were up 30% since last quarter). At $70, Adata's kit gave me the best available 1.5 V XMP settings of DDR3-1600 with 9-9-9-24 timings.


Graphics Card: Zotac GeForce GTX 770 2 GB ZT-70301-10P
As Chris Angelini pointed out in his launch article, The GeForce GTX 770 Review: Calling In A Hit On Radeon HD 7970?, the 770's GK104 GPU isn't new. Rather, it's just running faster than it did on GeForce GTX 680. Even still, at its price, the combination is a big win for enthusiasts.

Read Customer Reviews of Zotac's ZT-70301 GeForce GTX 770
Zotac's take on the GeForce GTX 770 sports a guaranteed 1059 MHz base clock, a typical GPU Boost frequency of 1098 MHz, and 2 GB of GDDR5 on an aggregate 256-bit interface at 7010 MT/s.
It’s outfitted with a quiet dual-fan cooler, and requires six- and eight-pin power connectors. As a bonus, when I bought it, it came with a copy of Assassin’s Creed IV: Black Flag. Outputs include DVI-D, DVI-I, HDMI, and DisplayPort.
Hard Drive: Western Digital Blue WD10EZEX 1 TB
Western Digital’s Blue-series 1 TB hard drive is becoming a staple in my System Builder Marathon configurations, giving me plenty of capacity and performance at a price I can typically squeeze into a limited budget.

Read Customer Reviews of Western Digital's Blue 1 TB
This SATA 6Gb/s-capable mechanical drive has 64 MB of data cache, a 7200 RPM spindle, and a limited two-year warranty.

Case: Rosewill Line-M MicroATX Mini-Tower
I swapped my case choice out at the last minute because the mid-tower I picked originally was bigger than it needed to be for my microATX motherboard (and frankly, it clashed with my own sense of style).
I'd personally prefer the reserved lines of Fractal Design's Arc Mini or SilverStone's Temjin on my desk. But both options were a little too pricey. So, I decided that Rosewill's affordable Line-M would work well enough with my entry-level H81-based board.

Read Customer Reviews of Rosewill's Line-M Case
This simple little mini-tower is still a solid choice for gaming, though. First, it can house monstrous graphics cards up to 12.5”-long. There's even a fifth expansion slot, accommodating motherboards with two PCI Express x16 slots for CrossFire and SLI with space in between for ample airflow. Speaking of flow, Rosewill bundles a pair of 120 mm fans to push and pull air straight through the system. And two more can be mounted on the side panel, if you want.
Front-panel connectivity includes a pair of USB 2.0 ports, plus an additional pair of USB 3.0 ports and audio I/O. While space for water cooling and storage is limited, there are three external and three internal drive bays that are more than adequate for our modest Haswell-based gaming PC. We could even add an SSD, fan controller, and card reader, if we had the money.
Power Supply: Rosewill Capstone-450-M 450 W
I wanted to use the 500 W 80 PLUS Bronze-certified EVGA power supply from last quarter, which was still available at the same price. But when it sold out, I stepped back to the company's 80 PLUS-certified model instead. I'm sure it would have worked, but the truth is that I know very little about the PSU.
As an alternative, I went with Rosewill's Capstone 450 W unit, which I've purchased for personal projects on a couple of occasions. Priced similarly, it's a good fit in the small enclosure I picked.

Read Customer Reviews of Rosewill's Capstone-450-M Power Supply
This 80 PLUS Gold-certified PSU delivers between 87 and 92% efficiency between 20 and 100% loads. It is both CrossFire- and SLI-ready, sporting a single beefy 37 A, +12 V rail, and a pair of 6+2-pin auxiliary leads.
Rosewill backs this unit with an impressive seven-year warranty.
Optical Drive: Asus 24x DVD Burner DRW-24B1ST/BLK/B/AS
While an optical drive may no longer be necessary in most builds, they're affordable, and I still think they're useful occasionally. Asus' 24x DVD burner is quite popular with Newegg's customers, so I'm going to give it a shot.

Read Customer Reviews of Asus' DRW-24B1ST/BLK/B/AS DVD Burner

Opening up the Line-M requires removing a pair of black thumb screws from each side of the enclosure. Inside, we're greeted by an all-black interior parts and a surprisingly roomy layout. The front panel can be snapped off with a tug from the bottom lip, if you need to get under it.
Rosewill only taps eight of the nine mounting holes used by microATX-compliant boards. That's not a problem if you're using a full-width 9.6”x9.6” motherboard, but the case lacks the lower right-hand “S-positioned” standoff used by narrower platforms like ours. Although I could have gotten by with the five screws that line up, I grabbed a plastic standoff from my stockpile to support the board's corner.
The rear slots are basic knock-outs, found on many other budget-oriented enclosures. Rosewill adds a bit of value by including a pair of vented replacement covers in case you want to cover a slot back up later on. The three external drive bays include mounting clips, and extra screws are provided for those as well. Either method works for us; this build will be broken down after the Marathon and shipped to one of our contest winners anyway. I had the option of mounting my hard drive up above, keeping the interior a little cleaner-looking, or fastening it to the case's floor, which arrived outfitted with vibration-dampening grommets.

The front LED-equipped fan is outfitted with three- and four-pin plugs. But since Asus' motherboard only has one header, which I reserved to control the rear exhaust fan, I could either leave the cooler disconnected, get creative in dropping its voltage myself, or generate more noise by running the front intake at full-speed all of the time.
More problematic is that Asus' H81M-K lacks a front-panel USB 3.0 header. At least the case includes a pair of front-panel USB 2.0 ports compatible with our motherboard.
Lastly, Rosewill does include a few black tie straps with its case, but there's otherwise very little to help facilitate cable management in the Line-M. The chassis more than covers our basic needs in a modest build, though, and the modular power supply does help prevent clutter by eliminating unneeded power cables.

Intel's Core i3 can't be overclocked at all, and I didn't bother trying to eke out gains from the system memory. Rather, I focused all of my tuning efforts where they mattered most: extending the high-resolution gaming capabilities of Zotac's GeForce GTX 770 graphics card through aggressive overclocking.

Right out of the box, the graphics card ran coolly and quietly, with in-game GPU Boost frequencies pinned at 1150 MHz under heavy load. In CPU-limited workloads, clock rate dropped as low as 719 MHz at 0.9 V. That's normal behavior with GPU Boost-enabled hardware, as power-saving features kick in to save power in lighter tasks.
Nvidia's drivers are set to "Adaptive" Power Management Mode by default. While overclocking, however, I changed that to “Prefer Maximum Performance”, which prevents 3D clocks from dipping below the GPU's base value. Since neither the CPU nor the memory are overclocked, today's story gives us the bonus angle of benchmarking the effectiveness of Nvidia's efficiency concession, and see if it has any side effects. You can see slight variations in our charts using CPU-limited quality settings. In short, though, the driver option didn't make a bit of difference. At least in this configuration, the technology works as it should, paring back power consumption and heat without adversely affecting performance.
I had ambitious hopes to hit a 1300 MHz GPU Boost clock rate with this GeForce GTX 770, but ran into instability at 1267 MHz using 1.2 V. Unwilling to give up, I switched to the latest beta build of MSI's Afterburner software, which unlocked a 12 mV bump. The GPU then passed my initial stability testing at 1306 MHz.

The GDDR5 probably had more headroom, but I stopped at a stable 7806 MT/s data rate. To ensure extended stability, I backed both frequency settings down a bit, yielding a 1283 MHz GPU Boost clock rate and 7610 MT/s memory.
The following tables give you a more in-depth look at the specifications of this quarter's System Builder Marathon configuration, along with the modest overclock I achieved with its graphics card, along with the same information from my previous two SBM builds. At the very bottom, you'll find all of the programs and games used for benchmarking.
| Current $750 PC System Test Configuration | ||
|---|---|---|
| Component | Base Settings | Overclock Setting |
| CPU | Intel Core i3-4130 (Haswell): 3.4 GHz, 3 MB shared L3 cache, Power-saving features enabled | Unchanged |
| CPU Cooler | Intel boxed heat sink and fan | Unchanged |
| Motherboard | Asus H81M-K, LGA 1150, Intel H81 Express, BIOS: v.0304 (08-23-13) | Unchanged |
| RAM | 8 GB (2 x 4 GB) Adata DDR3-1600 kit, CL 9-9-9-24 XMP at 1.5 V | Unchanged |
| Graphics | Zotac ZT-70301-10P GeForce GTX 770 2 GB GDDR5, 1059 MHz (1111 MHz GPU Boost, 1150 MHz maximum) GPU, 1.2 V, 1753 MHz (7010 MT/s) memory | 1283 MHz (Maximum GPU Boost), 1903 MHz (7610 MT/s) GDDR5 Memory, (+12 mV, 106% power, Custom fan profile) |
| Hard Drive | Western Digital Blue WD10EZEX 1 TB, 7200 RPM, 64 MB Cache | Unchanged |
| Sound | Integrated eight-channel HD Audio | Unchanged |
| Network | Integrated GbE networking | Unchanged |
| Power | Rosewill Capstone-450-M 450 W ATX | Unchanged |
| Optical | Asus 24x DVD burner DRW-24B1ST/BLK/B/AS | Unchanged |
| Software and Drivers | ||
| Operating System | Windows 8 Professional x64 | Unchanged |
| Graphics Driver | Nvidia GeForce 332.21 WHQL | Unchanged |
| Platform Driver | Intel Inf. v. 9.4.0.1017 | Unchanged |
And here's last quarter's configuration.
| Q4 2013 $800 PC System Test Configuration | ||
|---|---|---|
| Component | Base Settings | Overclock Setting |
| CPU | Intel Core i5-3470 (Ivy Bridge): 3.2 GHz Base Clock Rate, 3.6 GHz Turbo Boost, 6 MB Shared L3 Cache, Power-saving features enabled | 3.8 - 4.0 GHz (20 * 200), Power-saving features enabled |
| CPU Cooler | Intel boxed heat sink and fan | Unchanged |
| Motherboard | ASRock Z75 Pro3, LGA 1155, Intel Z75 Express, BIOS: v.P1.90 (07-12-13) | Unchanged |
| RAM | 8 GB (2 x 4 GB)Team Vulcan DDR3-1600 kit, CL 9-9-9-24 XMP at 1.5 V | Unchanged |
| Graphics | Gigabyte GV-R928XOC-3GD Radeon R9 280X 3 GB GDDR5, 1100 MHz GPU, 1500 MHz (6000 MT/s) memory | 1115 MHz GPU, 1600 MHz (6400 MT/s) GDDR5 Memory |
| Hard Drive | Western Digital Blue WD10EZEX 1 TB, 7200 RPM, 64 MB Cache | Unchanged |
| Sound | Integrated eight-channel HD Audio | Unchanged |
| Network | Integrated GbE networking | Unchanged |
| Power | EVGA 500B 100-B1-0500-KR 500 W ATX12V v2.91 | Unchanged |
| Optical | Lite-On 24x DVD burner SATA IHAS124-04 | Unchanged |
| Software and Drivers | ||
| Operating System | Windows 8 Professional x64 | Unchanged |
| Graphics Driver | AMD Catalyst 13.11 Beta 9.4 | Unchanged |
| Platform Driver | Intel Inf. v. 9.4.0.1017 | Unchanged |
And finally, the $650 PC from six months ago.
| Q3 2013 $650 Gaming PC System Test Configuration | ||
|---|---|---|
| Component | Base Settings | Overclock Setting |
| CPU | AMD FX-6300 (Vishera), 3.5 GHz (4.1 GHz Turbo Core), Socket AM3+, 8 MB shared L3 cache, Turbo Core enabled, Power-saving features enabled | 4 GHz (20 * 200), stock 1.2625 V VID, Turbo Core disabled, Power-saving features disabled |
| CPU Cooler | AMD retail boxed heat sink and fan | Unchanged |
| Motherboard | MSI 970A-G43, AMD 970 / SB950, BIOS: v.10.2 (02-04-13) | Unchanged |
| RAM | 8 GB (2 x 4 GB)Kingston HyperX DDR3-1600 kit, CL 9-9-9-27 XMP at 1.650 V | DDR3-1866, CL 9-10-10-28 1T at 1.65 V |
| Graphics | EVGA GeForce GTX 760 02G-P4-2760-KR2 GB GDDR5, 980 MHz (1033 GPU Boost, 1110 MHz maximum) GPU, 1502 MHz (6008 MT/s) memory | 1254 MHz (Maximum GPU Boost), 1801 MHz (7204 MT/s) GDDR5 memory, (110% power, +150 MHz GPU, +600 MHz memory, Auto Fan) |
| Hard Drive | Western Digital Blue WD10EZEX 1 TB, 7200 RPM, 64 MB Cache | Unchanged |
| Sound | Integrated eight-channel HD Audio | Unchanged |
| Network | Integrated GbE networking | Unchanged |
| Power | Antec VP-450 450 W ATX12V v2.3 | Unchanged |
| Optical | Samsung 24x DVD burner SATA Model SH-224DB/RSBS | Unchanged |
| Software and Drivers | ||
| Operating System | Windows 8 Professional x64 | Unchanged |
| Graphics Driver | Nvidia ForceWare 320.49 WHQL | Unchanged |
| Platform Driver | Unchanged | |
These are the System Builder Marathon benchmarks.
| Benchmark Configuration | |
|---|---|
| 3D Games | |
| Battlefield 3 | Version 1.6.0.0, DirectX 11, 90-sec. Fraps "Going Hunting" Test Set 1: Medium Quality Preset, No AA, 4X AF, SSAO Test Set 2: Ultra Quality Preset, 4X MSAA, 16X AF, HBAO |
| The Elder Scrolls V: Skyrim | Version 1.9.32.08, 25-Sec. Fraps Test Set 1: High Preset, No AA, 8x AF, FXAA Enabled Test Set 2: Ultra Preset, 8x AA, 16x AF, FXAA Enabled |
| Far Cry 3 | V. 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 |
| F1 2012 | Version 1.3.3.0, Direct X 11, Built-in Benchmark Test Set 1: High Quality, No AA Test Set 2: Ultra Quality, 8x MSAA |
| Arma 3 | Version 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 |
| Battlefield 4 | Version 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 2 | Version 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 CLI | Version: 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) |
| iTunes | Version 11.0.4.4 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) |
| 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 |
| Abobe Creative Suite | |
| Adobe After Effects CS6 | Version 11.0.0.378 x64:Create Video, Three Streams, 210 Frames, Render Multiple Frames Simultaneously |
| Adobe Photoshop CS6 | Version 13 x64: Filter 15.7 MB TIF Image: Radial Blur, Shape Blur, Median, Polar Coordinates |
| Adobe Premiere Pro CS6 | Version 6.0.0.0, 6.61GB MXF Project to H.264 to H.264 Blu-ray, Output 1920x1080, Maximum Quality |
| Adobe Acrobat X Pro | Version 11.0.0.379: Print PDF from 115 Page PowerPoint, 128-bit RC4 Encyption |
| Productivity | |
| ABBYY FineReader | Version 10.0.102.95: Read PDF save to Doc, Source: Political Economy (J. Broadhurst 1842) 111 Pages |
| Autodesk 3ds Max 2013 | Version 15.0 x64: Space Flyby Mentalray, 248 Frames, 1440x1080 |
| Blender | Version 2.67b, Cycles Engine, Syntax blender -b thg.blend -f 1, 1920x1080, 8x Anti-Aliasing, Render THG.blend frame 1 |
| Visual Studio | Version 10.0, Compile Google Chrome, Scripted |
| Compression | |
| 7-Zip | Version 9.28, LZMA2, Syntax "a -t7z -r -m0=LZMA2 -mx=5" Benchmark: THG-Workload-2012 (1.3 GB) |
| WinRAR | Version 4.2, RAR, Syntax "winrar a -r -m3" Benchmark: THG-Workload-2012 (1.3 GB) |
| WinZip | Version 17.0 Pro, Syntax "-a -ez -p -r" Benchmark: THG-Workload-2012 (1.3 GB) |
| Synthetic Benchmarks and Settings | |
| 3DMark 11 | Version: 1.0.3, Performance Suite |
| 3DMark Professional | Version: 1.2.250.0, Fire Strike Standard and Extreme |
| SiSoftware Sandra 2013 | Version: 2013.01.19.50, Processor Arithmetic, Cryptography, Memory Bandwidth Benchmarks |
While we remain confident that Intel's Haswell-based Core i3 processor delivers a satisfying gaming experience, we also have to accept that a Hyper-Threaded dual-core CPU fixed at 3.4 GHz is a clear step down from the quad-core Core i5 in parallelized workloads. At the same time, the Core i3 struggles to keep up with less expensive processors from AMD. So, you’ll have pardon the redundancy, but I'm doubling up on a number of benchmarks, testing with the latest versions wherever possible, but also including data in common with our six-month-old $650 effort. I thought the extra time would be worthwhile, since the FX-6300 and GeForce GTX 760 both continue to earn recommendations in their respective price brackets.
3DMark

By overclocking this quarter’s GeForce GTX 770, I was almost able to match the $800 build’s graphics scores, since the potent factory-overclocked Radeon R9 280X had virtually no headroom left to improve its GPU frequency.

On the flip side, today's build suffers from lower Physics and Combined scores, suggesting that, in threaded workloads, it's going to struggle with the $650 Vishera-based configuration.

SiSoftware Sandra

Averaging ALU and iSSE3 Arithmetic measurements in SiSoftware Sandra 2013 puts Intel's Core i3 just below AMD’s FX-6300.

But both Intel Core-based machines benefit from AES-NI support and higher memory bandwidth, yielding greater throughput in the Cryptography module.


Although budgetary limitations commonly hamper my ability to procure tweaking-friendly hardware, I still managed to squeeze an extra 400 to 600 MHz from last quarter’s Core i5-3470, depending on how many cores were active. And earlier, the modest stock cooler and AMD 970-based motherboard were enough to push an FX processor to 4.0 GHz, yielding a meaningful speed-up in threaded applications. Unfortunately, the current machine is at a major disadvantage, since its Core i3 processor can't overclock at all.

In our single-threaded iTunes and LAME MP3 workloads, the Haswell architecture running at 3.4 GHz is fast enough to match Ivy Bridge at 3.6 GHz.


The Hyper-Threaded Core i3 is no match for four physical cores in our more demanding transcoding benchmarks, though.


Intel’s Core i3 offers respectable per-clock performance, which really helps is excel in lightly-threaded tests, primarily because quad-core processors go under-utilized. But the dual-core chip in this quarter's setup is unable to compete with the older $650 machine's three Piledriver modules in our more taxing Premiere Pro and Photoshop benchmarks.
Really though, the star here is the $800 system's Core i5, which, in comparison, appears to do all things well thanks to a balance between four cores and Turbo Boost technology.





The oldest build's FX-6300 CPU totally outclasses Intel’s dual-core offering in threaded productivity-oriented applications, though similar performance within Visual Studio helps tighten up the averages a bit. Thankfully, as a purpose-built gaming box, this quarter's $750 PC was never intended to be a workhorse. I just have to make sure it can keep up so I'm not blown out of Thomas' Day 4 value comparison.





My current build gets hammered by the Core i5- and FX-based builds in 7-Zip, which is well known to effectively utilize available processing resources. Otherwise, the Core i3 holds its own in WinRAR, which isn't very well-optimized, and WinZip. The issue with WinZip is that this is an older version of the software; newer builds are better able to exploit multi-core CPUs.


This quarter's machine even scores a victory, as its GeForce GTX 770 best handles the OpenCL-accelerated WinZip sub-test.


Battlefield 4
Last quarter, we cycled in the newest edition of EA’s Battlefield franchise. But of course, we're lacking the data for my $650 machine from two quarters ago. As a result, I'm going to call upon Battlefield 3 one more time for the sake of drawing a three-way comparison.

Signs of the Core i5-based machine's superiority disappear as I crank graphics quality up to the game's Ultra preset, where both rigs pack enough graphics muscle for playable frame rates at 1920x1080. Neither survives 4800x900 without reducing settings, however.

Battlefield 3
At the high resolutions we're most interested in, frame rates in our Battlefield 3 single-player test sequence are typically limited by graphics performance, and not by modern CPUs. To better reflect the graphics loads you'll experience in-game, we shoot for an average of about 45 frames per second as our target.

At our lowest settings, my current build's overclocked GeForce GTX 770 appears bottlenecked by the multiplier-locked Core i3 processor, whereas the $800 PC simply bumps up against the game’s maximum cap of 200 frames per second.

Even in stock form, my newest effort remains viable through Battlefield 3’s Ultra quality preset, including 4800x900. In comparison, the $650 machine’s GeForce GTX 760 required hefty overclocking to eke out playable frame rates in Surround.

F1 2012
All three of these machines are mainly processor-bound in F1 2012, demonstrated by the Ivy Bridge-based Core i5 that distances itself from the more affordable CPUs, in turn bolstering the $800 PC’s overall frame rate totals.
We also know this game to be heavily bandwidth-dependent, and thanks to Sandra's Memory Bandwidth module, we have evidence to show that same Core i5-based system capable of pushing more than 21 GB/s.

Of course, our final evaluation will emphasize the settings gamers want to use most, which is good news since even the least-expensive rig is able to deliver at least 60 FPS under the Ultra preset at 4800x900. I'd hardly count that as a loss.

Grid 2
Although more GPU-bound at its Ultra detail preset than F1 2012, we still find that only the most taxing settings load our GeForce GTX 770 enough for GPU Boost to kick in and drive the highest available GPU frequencies.

Most important, both rigs remain playable in Grid 2 through all tested settings and resolutions.

Arma 3
Packing a Core i5 processor helps the $800 PC pull away at lower settings and resolutions.

This quarter's rig takes a lead at my targeted settings of 1920x1080 and the Ultra preset. Although minimum frame rates are 5 FPS higher on my newest box, Arma 3 is plenty smooth on both machines.

In the end, both PCs run into the same ceiling. At 4800x900, lower quality settings are required to sustain more than 30 frames per second.

The Elder Scrolls V: Skyrim
When we pack in high-end graphics muscle, Skyrim is mainly CPU-bound. The performance drop at 4800x900 is a result of the wider aspect ratio, and not just the demands of pushing more pixels.

The AMD-based PC trails, but still fares well enough. Best of all, even at stock clocks, neither machine with an Intel CPU drops below 60 FPS all the way through 4800x900.

Far Cry 3
Far Cry 3 is the most graphically-demanding game in our System Builder Marathon suite, and one of those rare titles able to justify high-end graphics for medium-resolution play. It also happens to be the only one of our former titles remaining in the line-up moving forward.

At high-quality settings without anti aliasing, we see evidence through 1920x1080 that both overclocked PCs with GeForce cards in them are held back by their respective CPUs. In my opinion, though, even the stock $650 PC remained playable though 4800x900. This title doesn’t necessitate high frame rates for a smooth experience, and minimums never dropped below 34 FPS.
If you insist on higher sustained performance, you might prefer the 47 to 52 FPS floors established by the two costlier configurations.

At Ultra quality with 4x MSAA, a GeForce GTX 760 is as low as I'd go to play at 1920x1080. The $650 PC averaged less than 37 FPS, but never dropped below 32 in our test sequence (which is about as demanding as this game gets). The overclocked GeForce GTX 760 and factory-overclocked Radeon R9 280X are capable of staying above 40 FPS. But the newest machine wins, sporting the highest average and minimum frame rates, stock and overclocked.
None of my budget-oriented PCs survive through 4800x900 at these cranked-up settings. However, spending time in-game determined that simply disabling MSAA made the $800 machine playable. And today's configuration fares even better. In fact, after putting in hours testing for stability, I was able to turn on 2x MSAA and keep things smooth.

Power-saving features are typically left on for each round of benchmarking. Only the overclocked $650 system didn't benefit from them. When it came to tuning the FX-6300's clock rates, I jumped into the firmware and disabled any setting that might negatively affect stability or performance consistency.

Sporting an efficient Haswell-based Core i3 processor and powered by an 80 PLUS Gold-rated power supply, my most recent effort sips power at idle and under host processing load. There's an overclocked, voltage-bumped GeForce GTX 770 in there, but still, peak draw from the wall remains under 300 W. You most certainly don't need the 600 W power supply recommended by Zotac. Our 450 W PSU has oodles of output in reserve.

Low power means that this quarter's PC is also the coolest-running in today's competition. Notice that CPU and GPU temperatures drop after applying an overclock. Simply, the CPU is left alone, and airflow increases by speeding up the cooling fans through Asus' firmware (from the Standard setting to Turbo). I also disabled control over the rear case fan by simply running it at 100% duty cycle. Finally, Zotac's thermal solution was quiet, so I created a custom fan profile to blow more air over the card and its components.
Of course, faster fans generate more noise. While this is by no means a quiet PC, it wouldn’t take much to pull its output down to a whisper. Surprisingly, overclocking didn't really add much noise since the 120 mm intake fan up front is the loudest. It doesn't vibrate, and is by no means obnoxious. But there is an unquestionable sound of rushing air. Needless to say, my fondness of the Line-M enclosure certainly diminishes when its bundled fans are running all-out.
That was a ton of data spanning six months worth of benchmarking. So, let's finish up by comparing three gaming-oriented builds to each other.
Given the ongoing relevancy of the $650 machine, I'll use that as our baseline and factor in only the four titles tested throughout 2013's System Builder Marathons.

Tallying average frame rates using every resolution puts a big emphasis on CPU performance. The Q1 2014 PC drops its processor budget by almost one-third, which would seem scary. And yet it still serves up at least 90% of the Core i5-equipped rig’s frame rates in three titles. It only trails further behind in F1 2012.
The more affordable $650 PC takes a hit in CPU- and GPU-limited scenarios, but its Achilles heel is The Elders Scrolls V: Skyrim. You just have to keep in mind that it was still going strong using the Ultra preset, never dropping below 50 FPS at 1920x1080.
But nobody builds a machine this strong to game at 1280x720 with low-quality details and jagged edges. So why even run those benchmarks, which mess with the averages? The data points do serve a purpose; they let us identify bottlenecks and gauge the potential of other subsystems. We can factor them out of our calculations, though, and draw conclusions about the resolutions and settings that matter most to gamers.

Cranking up the eye candy at 1920x1080 obviously shifts demands over to graphics hardware, and once we overclock the GeForce GTX 770, this quarter's PC trades blows with the more expensive rig. If we dropped F1 2012, which never dropped below 72 FPS anyway, today's effort actually matches the $800 build’s overall frame rate measurements in stock and tuned form. Performance across three panels is pretty darned strong, too.
Remember, even the $650 box survived 1920x1080 in all four games at Ultra details. The respectable overclocking headroom of its GeForce GTX 760 gave us higher GPU Boost clock rates when we needed them. But when it came to 4800x900, quality had to suffer in the name of smooth game play.

Sandwiched between the two previous machines in cost, my Q1 2014 setup delivers great frame rates in most modern titles. Of course, that was the goal all along; my hardware choices reflect the tough decisions needed to push 1920x1080 and taxing image quality options.
But the System Builder Marathon tends to illustrate overall system performance with a ton of other tests, too.

In threaded workloads, AMD's six-core FX-6300 is more capable. But the Haswell architecture's ability to get more work done per clock cycle, along with the GeForce card's solid OpenCL compute performance, means today's platform is roughly on par...at least until overclocking is taken into consideration, too.

These are gaming machines, first and foremost. So I weigh their overall performance accordingly, rather than using the same formulas Thomas will use tomorrow. On Day 4, my PC is undoubtedly going to struggle. Its gaming aptitude will only account for 20% of its overall rating. And although it doesn't quite hang with the $800 machine in today's evaluation, it slides in ahead of the cheaper system.

The two Intel-based PCs are between 56% and 71% more efficient. Overclocking both GeForce cards reduces overall efficiency compared to the stock configurations, too. While our current PC is the overall winner, we can't forget that it's driven by a pricier 80 PLUS Gold-certified power supply.
Value: Is There A Winner?
I only chart value when it’s difficult to pick a bang-for-the-buck winner through an easy calculation. This time, I have to admit that true value gets more subjective, depending on your specific needs. Even picking the right price to use in a comparison is tough, since the numbers are changing on a daily basis. We paid $300 for a Radeon R9 280X last quarter, which then shot up to $420 by the time my story went live. When it came time to order the parts for today's PC, 280Xes were still going for $420 and up. Now, the cheapest 280X in stock is $340. And the Zotac GeForce GTX 770 I used is bouncing up and down between $330 and $350.
My three PCs originally cost us $650, $799, and $773. Now, with some minor substitutions accounting for in-stock alternatively, they're closer to $685, $857, and $779. But I can almost guarantee they'll change again depending on the day you read this. Any attempt to lock in a price is futile.
I'll gauge value on two factors, then. First, the original purchase price of the components, which was set in stone, and then based on an adjusted platform price. All three builds were outfitted with 8 GB of affordable DDR3-1600 and the same 1 TB hard drive. Every time, I grabbed the most affordable DVD burner I could find and relied on bundled CPU heat sinks. All three machines could get by with the same case and power supply, which typically sold for about $45 each. Those parts are a matter of personal preference, so a fixed $235 covers the basic memory, power, storage, and case. From there, I add the cost of what really separates the platforms: processor, motherboard, and graphics. The adjusted prices come in at $675, $842, and $745.

Using only the 2013 test suite, I originally got the most value from the $800 PC in stock and overclocked form. The $650 build even slightly edges out today's effort. Of course, we can't forget that the current machine is penalized by $55 in mark-ups from the RAM, storage, optical drive, and chassis that I picked.
When we use current pricing and level out the playing field with complementary components, the current Core i3 and GeForce GTX 770 combination secures a narrow value victory right out of the box. However, all three builds are within 2% of each other once I overclock them. I don't think we can pin down a definitive winner; this is as close as you can get.

This quarter, I set out to build a more affordable gaming equivalent to my previous $800 PC, which, in three months, rose in cost by an astounding $137. Unfortunately for me in this piece, but good for gamers as a community, Radeon R9 280X prices aren't as bad as they were a month ago. It'd be great to see them back down at $300, though the cheapest models as of this writing sell for $350 on Newegg.

We recently introduced some new tests to the System Builder Marathon benchmark suite, and those were the current build's target apps. In them, my current effort serves up 90% of the $800 machine’s average frame rates. That number jumps to 94% once its GeForce GTX 770 is overclocked. Interestingly, I get the same exact totals for stock and overclocked performance in our previously-tested games. Of the titles we used to run, only Far Cry 3 carries over.

But my comparison doesn't end there. In fact, I need a few more charts to keep telling this story. The games we just added are unquestionably less processor-bound. As a result, sporting similarly-quick graphics hardware, the new $750 system competes readily at the most interesting resolutions against my old build, going so far as to beat it once I overclock them both.

The gains seen at high resolutions help this quarter's PC pick up a few percentage points of overall performance, too.

Spending more on the case, power supply, memory, and graphics card prevents the Q1 2014 build from matching the as-purchased value of my $800 machine. However, when I adjust both systems to reflect today's pricing and isolate the performance-oriented parts, my $750 box scores a victory in its stock form. Here's the thing, though: once I overclock the Core i5 in last quarter's machine, value evens out again.

If all you care about is gaming at 1920x1080 or higher, today's cheaper PC is just as capable as the last one I built for more money. Your only concern should be running out of graphics memory at Surround resolutions, which might not be playable anyway, depending on the settings you're using.
Of course, the averages don't give us the whole story, so we turn to the individual games. My $750 machine leads in Far Cry 3, Battlefield 3, and Arma 3, while the $800 PC scores big in Grid 2, F1 2012, and The Elder Scrolls V: Skyrim. Both rigs trade blows in Battlefield 4, which either favors the Core i5 processor or Radeon R9 280X more than its predecessor. In a blind test, I'm willing to bet that you couldn't tell the difference, though.
The good news for me is that, despite lofty prices on some components, this most recent machine is a success. The competition is a fundamental tie, and I'd need to test more games to try pinpointing a winner. Intel’s Core i3-4130 again proves itself to be a capable value-oriented gaming processor able to keep up with Nvidia's GeForce GTX 770 in the seven titles we tested. At the same time, I don't have anything bad to say about the previous gaming builds either. As prices drop, AMD’s Radeon R9 280X re-joins the GeForce GTX 760, Core i5-3470, and FX-6300 as worthy options for your next gaming PC.