Logo   Banner   TopRight
TopUnder
Transparent
Digital Storm's Core i5 System Reviewed
Transparent
Date: Dec 17, 2009
Section:Systems
Author: Joel Hruska
Transparent
Introduction and Specifications


When Intel introduced its LGA1166 Lynnfield processors earlier this fall, one of its goals was to introduce Nehalem's advanced technology at lower price points. To that end, Digital Storm has designed a high-end system around the P55 chipset and Core i5 processor. At a price of ~$1900, the customized system isn't what you'd call cheap, but it's still less expensive than some of the Core i7 PCs we've seen recently, several of which landed between $3100-$4000 dollars. At first glance, the 2.67GHz Core i5 might not seem to be capable competition for Core i7 processors running at 3.2GHz and above, but Digital Storm has a trick up its sleeve. The relatively pedestrian quad-core has, in this case, been punched up to 3.8GHz from the factory. DS claims that the processors it ships in this configuration have been certified to operate "100% Stability and Reliability." We'll test that claim later in the review. It's worth noting that the company doesn't actually guarantee any single clockspeed if you opt to have the processor factory overclocked. For $45, the company will overclock the CPU to 3.3-3.9GHz, while $199 buys you entry in the 4GHz+ club.



 
Digital Storm Gaming System
System Specifications

Processor
Intel Core i5-750 2.67GHz (base, 3.8GHz Factory OC)

Motherboard

EVGA P55 FTW Edition (SLI Supported)

Operating System
Windows Vista Home Premium 64-bit Edition SP1
Windows 7 Home Premium Upgrade

Memory
4GB Mushkin Enhanced
DDR3-1600MHz 8-8-8-24

Graphics Cards

2x EVGA GeForce GTX 275 

CPU Cooling

Asetek Liquid CPU Cooler (120mm Radiator)

Audio

Onboard

Hard Drive
Western Digital Caviar Black (1TB, 7200RPM, 32MB Cache)

Optical Drive
DVD-R 20x / CD-RW 48x

Accessories
None


Case
Cooler Master HAF 932
Width: 9.0" (229mm)
Depth: 22.7" (577mm)

Height: 21.5" (546mm)
Weight: 29lbs (Empty) (13.22kg)


Power Supply
1000W Chieftec PSU

Available Expansion Slots
1x PCIe x16

1x PCIe x1

2x PCI


Front Panel I/O Ports

4 USB 2.0
1 FireWire 1394A

1x eSATA

1 Headset
1 Mic


Rear Panel I/O Ports
1 PS/2 keyboard

1 CMOS Reset

6 USB 2.0
2 RJ-45 Gigabit Ethernet
1 IEEE 1394A
2 eSATA
10-channel audio I/O
1 SPDIF Digital/Optical Output

Warranty and Support

3 year limited parts + labor

(4-5 year options available)


Price:
$1,996.00 USD (as tested)


Our test configuration is built around the Cooler Master HAF932,a fairly common high-end case these days. The system includes two GeForce GTX 275 cards (896MB of RAM), 4GB of DDR3-1600 (2x2GB), a 1TB Western Digital Caviar Black hard drive, and a 1kW Chieftech power supply. Digital Storm offers a three year warranty on parts and labor, and will cover the cost of shipping the system there and back—provided it's within the first 30 days of ownership. The company ships an actual copy of Windows rather than relying on a recovery disc alone and includes lifetime telephone technical support with any system purchased.
 

The Warranty:
Digital Storm's baseline warranty is three years parts+labor on the system only. The company's website states that it "will cover all shipping costs (back and forth) as long as it's within the first 30 days of receiving your machine." Four and five-year coverage packages are also available for an additional fee.
Transparent
The Case: Cooler Master's HAF 932



Digital Storm uses the Cooler Master HAF (High Air Flow) 932 in several default configurations. The case is well-built, roomy, and provides a number of options for routing cables behind the motherboard as opposed to in front of it. DS exploits this to maximize airflow throughout the case's interior. The case shots here are courtesy of Cooler Master.



The HAF 932 may be designed to maximize airflow, but CM clearly had an eye on the tower's acoustic profile when they designed it. The front, side, and side fans are both 230mm (110CFM, 19dBA). Digital Storm has replaced the single 120mm rear fan with a pair of 120mm's (one on each side of the AseTek radiator), and the front fan (originally a third 230mm) has been swapped for a 120mm as well. It's technically possible to remove the top 230mm fan and install a pair of 120mm fans instead, but there's no practical reason to do so.




Here's a shot of the top of the HAF 932. The large 230mm fan is visible here. There's a fill port underneath the rubber mat, for those with water-cooling rigs.


 
Here's CM's stock photo of the FP connectors and fill port. Front-panel goodies include four USB ports, FireWire 400 (1394a), eSATA, and microphone/headphone jacks.

The HAF932 definitely isn't a case that everyone is going to like. Cooler Master isn't kidding when they call it "High Air Flow." Not only are the left side and top punched full of ventilation holes, the entire front of the case is an open grille. Both the side and top fans are exhaust fans. Add a 1kW PSU, two GTX 275s, and a pair of 120mm rear fans, and you've got a tremendous amount of air being pushed out of the case—and since nature abhors a vacuum, that means you've got quite a bit of air being pulled into the case too, mostly at the front. Even with the front fan disconnected, there's a steady stream of air flowing through the system.

The unavoidable side effect of pushing or pulling a ton of air across all those little holes is the creation of fan noise; at full load, the HAF932 sounds as though it's attempting to become a wind tunnel. Its physical characteristics and total lack of any sort of ventilation filters also makes the HAF932 an absolute dust magnet. If you've got shag carpet, six kids, and assorted dogs + cats, this might not be the case for you.
Transparent
Build Quality


Our review system, in the, err, flesh. The assembly team at Digital Storm appears to study origami in its spare time; every system cable is routed behind the motherboard or secured in such a way as to create minimal airflow disruption. The rest of the power supply's modular connectors are included in separate packaging, as are several molex-to-SATA and molex-to-PCIe power converters.



Here's a closeup on the CPU socket and Asetek liquid cooler. The unit is entirely self-contained, but the fans are Digital Storm's choice—they don't appear to ship standard with the cooling unit+radiator. That distinction is relevant to the system's noise level, which we'll discuss at the end of the review.



Next up, the rear of the case. Both GTX 275's are visible, as are the intake/outflow water pipe ports. The EVGA P55 FTW motherboard used in the rig packs a full range of external I/O capabilities; further internal headers are available if you  find yourself needing to hook up, say, a few more external hard drives. Dual eSATA ports are also a nice touch. The little red switch one expansion bay above the power supply is actually much more interesting than it looks. Ostensibly, it's a simple on/off switch for times when you might want to shut off the internal cold cathodes and dampen the flood of cerulean light that pours from the front of Digital Storm's creation. Believe it or not, however, Digital Storm's decision to put the switch where it is knocked up to 40 percent off our review system's 3D performance (we'll explain why on the next page). The switch can be moved to a different bay, but only after removing several cable zip ties and rerouting the wires—as delivered, it only fits in the port shown.

This is something of a problem.
Transparent
A Less-Than-Perfect-Storm


Our experience with the test system was marred somewhat by two unrelated video card issues.

Unstable Video Overclocking.
Initially, our two GTX 275 cards were each clocked at 700MHz, a speed the company claimed it had verified 100 percent. In real world 3D tests, however, that speed proved unsustainable; the system crashed almost immediately in every test. Oddly, each card individually was capable of running at 700MHz in a non-SLI configuration, but enabling the feature immediately destabilized the system until the GPU clockspeed was lowered. We contacted the company and backed the cards down to a core clock of 675MHz, at which point we were able to test without further incident.

When we inquired about the company's method of verifying video cards were stable at overclocked speeds, we were told that Digital Storm uses both Crysis and
Furmark as stability tests, with Furmark considered the higher bar. After downloading and running Furmark (the program bills itself as an overclocking stability test and OpenGL benchmark) on the two GTX 275's in SLI at 700MHz, I can confirm that the cards are stable at that speed in the program's stress test mode. Try to run the actual benchmark, or any benchmark, and the cards promptly crash.

Digital Storm shipped its test suite (and several screenshots) installed to our review system, which allowed us to do a bit of cross comparison. Based on the Furmark test results reflected in one of the screen captures, it appears the stress test was run on just one card (or with SLI disabled).

We don't know if Digital Storm tested each card individually at 700MHz and then assumed they'd run correctly at that speed in SLI or if another factor confounded our results, but we've recommended the company use a more robust set of benchmarks for determining overclock stability. Furmark does an excellent job at heating up a GPU and draws a prodigious amount of power, but this test should be augmented by at least one of the mainstream 3DMark tests. It's not always possible to predict which programs or games are going to destabilize an overclocked component. The 3DMark programs are pretty good GPU stability tests—all of them since '03 have been sensitive to GPU instabilities—but even a 3DMark run does not, in and of itself, prove stability. The only way to do that is one program at a time; it's for this reason that any company (including OEMs) shipping video cards in an overclocked configuration should have a suite of tests.

Incorrect GPU Configuration:
The second problem is a bit more complex. When the Digital Storm i750-based system arrived, it was configured exactly as shown in our photos, with a GTX 275 in each of the first two PCIe physical x16 slots. At first, we didn't realize this was incorrect until after putting the system through a complete series of 3D benchmarks. It wasn't until we were actually writing this review that we stumbled across the following on page 22 of the FTW's
manual. (PDF)  "For SLI use, it is very important to use only PCI Express Slots 1 and 3 (primary and third) for optimal performance."

They aren't kidding. Here's why it matters:



The P55 FTW is built on Intel's P55 chipset, as shown above. 16 of the chipset's 24 PCIe lanes are Gen 2 and attach directly to the processor. The remaining eight are first-generation lanes and hang off the southbridge. The Gen 2 lanes directly off the CPU can be allocated to provide either a single video card with an electrical x16 connection or split between two cards at x8/x8. Since PCIe Gen 2 offers twice the bandwidth of PCIe Gen 1, two x8 connections offer each card 3.2GB/s of unidirectional bandwidth when in SLI mode (6.4GB/s bidirectional). In the case of EVGA's P55 FTW, it's the first and third slots that run at x8/x8. Digital Storm actually put its own review system at a severe disadvantage when it shipped the video cards in a Slot 1/Slot 2 configuration. Not only is the second slot limited to a fraction of the first's bandwidth (800MB/s, 1.6GB/s bidirectional), it's also attached through the southbridge rather than directly to the CPU. This adds additional latency to an already-lopsided linkage and hamstrings the configuration further.

EVGA didn't just put this bit of critical information in the manual—they actually silkscreened it on the board.


Pop out the first two video cards and there it is, plain as day.

It's hard to know what to make of this. Obviously it's not the sort of thing any company does on purpose, but it's not the sort of mistake that an enthusiast-oriented, custom-designed OEM wants to make either. If EVGA had simply noted the issue in a manual footnote that might be one thing, but the requisite information is plainly written. On other EVGA motherboards, the primary PCIe slots are a separate color from the secondary southbridge connection (which the company occasionally refers to as a PhysX slot.). If the difference between a Slot 1/Slot 2 configuration and a Slot 1 / Slot 3 configuration was minimal that would be one thing, but as you'll see, the performance gap is significant.

This is also where the ill-placed red-button from Page 3 causes trouble. As shipped, the button blocks the use of the third PCIe x16 slot. Normally, moving the backplate would be the work of seconds, but Digital Storm's tight cabling work prevents this. Moving the button to a different slots required rewiring and partial removal of the power supply, as well as adjustment at multiple points—the wiring to that button runs behind the entire case before connecting to one of the cold cathodes and the front fan.
Transparent
Test Systems


We tested the Digital Storm as configured below. Relevant 3D performance data was gathered with the video cards in the shipped configuration (x16/x4) and in the correct, optimal configuration (x8, x8). Specific configuration details are noted in each test, should you care to duplicate them. One thing to note as you check the benchmarks is that at $1896, Digital Storm is fighting well above its weight class when it takes on the iBuyPower and Alienware systems. The CyberPower system is a more direct competitor.

HotHardware's Test Systems
Performance Comparisons


System 1:

Digital Storm i750

Intel Core i5 750 @ 3.8GHz
EVGA P55 FTW ATX
4GB Mushkin DDR3-1600
NVIDIA GeForce GTX 275 SLI
1TB WD Caviar Black
Vista Home Premium x64


Price: $ 1896.00 USD


System 2:

Alienware Aurora ALX

Intel Core i7 975 3.33GHz
Alienware X58 MicroATX
6GB Corsair DDR3-1600
ATI Radeon 5870 CrossFire
1TB Seagate HDD RAID 0
Vista Home Premium x64


Price: $ 4,074.00 USD

System 3:

CyberPower Gamer Extreme 3000

Intel Core i7 860 2.8GHz
Asus P7P55D Delux P55
4GB Kingston DDR3-1600
EVGA NVIDIA GTX 295
1.5TB Seagate HDD
Vista Home Premium x64

Price: $ 1,599.00 USD

System 4:
iBuyPower Gamer Paladin F970

Intel Core i7 965 3.2GHz
Asus P6T Deluxe V2 X58
12GB Corsair DDR3-1333
NVIDIA GeForce GTX 295
128GB SSD / 1TB HDD
Vista Ultimate 64-bit

Price: $ 3,111.00 USD

Multi-core Performance and Hyper-Threading:

One performance comparison our Digital Storm testbed unintentionally raises is the impact of Hyper-Threading. The other systems tested here are all quad-cores with Hyper-Threading enabled, which gives them the ability to execute up to eight simultaneous threads across four virtual cores and four real ones. The i750, while limited to four "real" cores and just two DDR3-1600 memory channels, is clocked 15 percent higher than its closest competitor in that regard, the Aurora ALX. What makes this comparison interesting is the fact that Hyper-Threading's ability to deliver additional performance is limited by the degree of parallelism that the CPU's scheduling unit can extract from a given application. Windows "sees" eight cores and reports the existence of such, but there's a distinct difference between a quad-core with HT enabled (4L+4V) and an actual octal-core processor.

Based on the characteristics of typical consumer-level applications, we expect the quad-core Lynnfield to match or even surpass the performance of the slower-but-wider Core i7s in some situations. This might not be the case if we focused on professional software that's been designed and optimized for parallel execution; apps that can significantly utilize >4 cores will likely buck the trend.

Transparent
Windows Performance Optimizations: Fact or Fiction?

One of the unique features Digital Storm claims it provides is the option to optimize Windows Vista (or Windows 7) performance. According to the company, this involves fine-tuning the services and processes running in the background of the operating system. In theory, this improves system speed and response time without compromising stability or functionality in any way. To be honest, it wasn't a claim we paid much attention to one way or the other initially, until we saw the PCMark Vantage scores below. In the graph, the "OEM" label refers to the performance of the Core i5-750 system when the OS has been reinstalled using the provided company restore discs. "Retail" refers to our performance results when the system was loaded with a retail version of Vista from a standard pressed disc. Both systems used identical drivers, hardware, and were fully patched.


See the 9,338? That's over ten percent higher than the Alienware Aurora, which, while slightly slower, has two hard drives configured in a RAID 0 array. PCMark Vantage's results can be tricky, but this gap is larger—and more consistent—than can be explained within the margin of error. The Core i5 trounces all competitors here, beating even the Aurora ALX by 10.5 percent. If that seems strange to you, it did to us as well—the Aurora may be slower, but it's got more RAM, a (theoretical) higher IPC rate, an additional memory channel, and it's configured in a RAID 0 array. It's not the presence or absence of AHCI, because Digital Storm ships the system with the HDD configured in legacy IDE mode. (This means no NCQ and no hot-swapping hard drives—we actually intended to test the performance difference between AHCI mode and IDE mode when we installed the retail copy of Vista in the first place.)

Despite being identically patched and using identical drivers, our standard Vista install fell well short of the system's first score with an 8450. Multiple Windows reinstalls and OEM image installations confirmed that the 10.5 percent difference in PCMark Vantage was consistent. As for where the differences occur, the distribution is not random. PCMark Vantage's final score is calculated based on the system's performance in seven distinct categories. These are:
  • Memories
  • TV & Movies
  • Gaming
  • Music
  • Communications
  • Productivity
  • HDD
All of the performance differences we mapped occurred in the Music, Communications, Productivity, and HDD suites. When we broke down the individual tests within those seven categories, we noted a distinct trend—all of the gains, when they occurred, occurred in benchmarks that involved searching, reading, or writing to the hard drive. Actual computation tests that were processor, GPU, or memory-bound did not show the same variations.

Are The Optimizations Real?
In order to check against the possibility of some benchmark shenanigans, we installed and ran PC World's Worldbench 6, Beta 2. Although the finalized version of the benchmark shipped about a month ago, Beta 2 was the only release available for over two years. As such, it's been thoroughly vetted against several years of hardware and proven itself consistent and comparable over time. Unlike PCMark Vantage, which runs all of its tests within a customized shell, Worldbench relies entirely on standard desktop applications. This doesn't guarantee against some form of trickery, but it at least makes it less likely. Worldbench 6 returns an aggregate score measured against PC World's original baseline testbed, but also returns the individual times, in seconds, that it took a system to complete each application test.



Again, we see a performance gain of approximately 16 percent. If we break out the results of each individual application, the two OS installations track each other closely in virtually all cases; the second-largest gap is 6.8 percent in Roxio Videowave 1.5 percent. The OEM image only smashes our baseline install's performance in Worldbench's Nero 7 Ultra test.


Worldbench's Nero 7 Ultra test is ostensibly a benchmark of disk controller performance, and again, this is where we see the jump.

There are three basic variables in play here. The performance differences we've seen could be the result of changes to the hard drive, customized driver optimizations, or changes to Windows and how it performs certain operations. We can throw the first out immediately—we have not changed hard drives, updated firmware, or changed default block sizes. The second remains a possibility, the third is the company's explanation. To be frank, we'd be happier if we'd been able to duplicate (or come within a reasonable margin of duplicating) the company's results based on the changes we've made to the OS environment.

So are the performance differences "real?" As far as we can tell thus far, yes. The degree of performance acceleration is consistent, we see it when particular types of operations are performed, and we've yet to uncover evidence that the company is somehow warping test results. Furthermore, these changes do not appear to have a downside. Over the course of testing this system we've copied, moved, pushed, pulled, and teleported hundreds of gigabytes of data across SATA, USB2, DVD, CAT6, and llama. The only troublesome part was the llama, which refused to haul data at 4.8GB/s. At no time did we encounter a problem or stability issue in performing these operations, and the actions in question never appeared to be occurring at abnormally slow rates.

At this point, we'd state—with considerable caution—that the results appear valid, which is to say that these numbers are reflected in reality, not merely within benchmark results. Even if Digital Storm had concocted a unique secret sauce to boosting performance (and the company makes no such claim), the results could still be accurate. If we find evidence to suggest anything to the contrary, you'll definitely hear about it.
Transparent
SiSoft Sandra



Preliminary Testing with SiSoft SANDRA 2009
Synthetic Benchmarks


We began testing with SiSoftware's SANDRA, which stands for System ANalyzer, Diagnostic and Reporting Assistant. Since our 2.67GHz Lynnfield proved rock-solid at the 3.8GHz overclocked speed Digital Storm shipped us, we kept the chip running at that frequency for these tests and for all the tests forthcoming. 




Sandra's CPU test shows the Core i5's integer performance winning past the slower-clocked i7's, despite the latter's use of HyperThreading. The tables turn in FPU code, where the assembled Core i7 processors are able to make better use of their virtual cores. Here, the Core i5 lags even the CyberPower system by 20 percent, despite its 35 percent higher clockspeed.


If you're wondering about the difference between the arithmetic and multimedia tests, it's laid out in Sandra's help file. Both benchmarks measure CPU performance, but the Dhrystone/Whetstone benchmarks measure a processor's raw capability to execute integer or floating point instructions. Sandra's multimedia suite is designed to test the various SIMD capabilities of a processor. In the company's own words: The test involves the generation of Mandelbrot Set fractals that are used to realistically describe and generate natural objects such as mountains or clouds. By using various multi-media extensions better performance is achieved.

HyperThreading makes a difference here, but you can also see where the 3.8GHz core cock on the i5-750 keeps that chip near the head of the pack.
 


The Core i5 is limited to "just" two memory channels, making its bottom-end showing in this particular benchmark a foregone conclusion. Note, however, that Lynnfield isn't missing that extra channel all that much, even in a theoretical test. The CyberPower and iBuyPower only lead the Digital Storm rig by about 20 percent—a significant gap, but not as large as one might expect based on raw bandwidth numbers. Don't expect the bandwidth difference to have much of an impact on our actual tests—consumer applications and games are typically more latency-sensitive, than bandwidth-sensitive.



Transparent
PCMark Vantage


 

Next, we ran the test systems through Futuremark’s latest system performance metric built especially for Windows Vista, PCMark Vantage. This benchmark suite runs through a host of different usage scenarios to simulate different types of workloads including High Definition TV and movie playback and manipulation, gaming, image editing and manipulation, music compression, communications, and productivity. We like the fact that most of the tests are multi-threaded as well, in order to exploit the additional resources offered by quad-core processors. We used the 64-bit version of the benchmark and defragmented the hard drive immediately prior to running it. The test was looped 3x. One thing to keep in mind when comparing PCMark Vantage results is that the benchmark's margin of error is fairly wide—we'd estimate 5-7 percent. Relevant factors include  whether or not the hard drive was defragmented immediately prior to the run and whether Vantage was run immediately following OS+driver installation, or only after a full suite of tests and other benchmarks had been run.


In order to generate the cleanest results possible, we tested Vantage immediately after OS installation and defragmented the hard drive immediately prior to benchmarking.
 

Futuremark PCMark Vantage
Simulated Application Performance



Vantage might be multi-core friendly, but it's a big fan of HDD performance and clockspeed / IPC as well. The i5-750's high performance and the meaning behind the (OEM) and (Retail) qualifiers is explained in more detail here.

Transparent
Cinebench R10

 
Cinebench R10
3D Rendering

Cinebench R10 is an OpenGL 3D rendering performance test based on Cinema 4D from Maxon. Cinema 4D is a 3D rendering and animation tool suite used by 3D animation houses and producers like Sony Animation and many others. It's very demanding of system processor resources and is an excellent gauge of pure computational throughput. Cinebench is a multi-threaded, multi-processor aware benchmark that renders a single 3D scene and tracks the length of the entire process. The rate at which each test system could render the entire scene is represented in the graph below.


Cinebench is a perfect example of how HyperThreading can boost CPU performance over and above a frequency increase. Granted, the Core i5-750's four cores at 3.8GHz are almost as fast as the HyperThreaded Core i7's, but the Core i7 may well be the more power-efficient of the two. We do, ironically, see a limitation of HT technology—if we're talking about single-threaded code, there's no substitute for clock. Here, the Core i5 leads its closest competitor by 7.2 percent.

Transparent
3DMark 06, 3DMark Vantage

Futuremark 3DMark06
Synthetic DirectX Gaming


3DMark06

3DMark06 is an aging benchmark now, but it still serves as a useful benchmark of GPU capability, particularly when combined with its bigger brother, 3DMark Vantage. We ran the test at default settings and looped it three times to obtain a more reliable result. Texture and AA settings were left in the default position (optimal and off respectively).


Our first 3D benchmark, and the first concrete example of just how much an x8/x8 configuration boosts performance over an x16/x4. Configured as shipped, the i5-750 is 6 percent behind the Aurora ALX. Configured properly, it's three percent faster. That's a difference of 10.1 percent in all. Let's see how that shifts when we swap to 3DMark Vantage, which is significantly newer and includes an updated set of tests.


Futuremark 3DMark Vantage
Synthetic DirectX Gaming


3DMark Vantage

The latest version of Futuremark's synthetic 3D gaming benchmark, 3DMark Vantage, is specifically bound to Windows Vista-based systems because it uses some advanced visual technologies that are only available with DirectX 10, which isn't available on previous versions of Windows.  3DMark Vantage isn't simply a port of 3DMark06 to DirectX 10 though.  With this latest version of the benchmark, Futuremark has incorporated two new graphics tests, two new CPU tests, several new feature tests, in addition to support for the latest PC hardware.  We tested the graphics cards here with 3DMark Vantage's Performance preset option, which uses a resolution of 1280x1024. As always, tests were looped 3x.

In 3DMark's newer, more stressful benchmark, the difference is a tad bigger at 12.1 percent. In the old configuration, the DS 750 is neck-and-neck with the Aurora in this test; in the new configuration it's clearly faster. 
Transparent
Left 4 Dead, Batman: Arkham Asylum

Left 4 Dead
DirectX Gaming Performance


Left 4 Dead

Left 4 Dead is a co-operative, survival horror, first-person shooter that was developed by Turtle Rock Studios, which was purchased by Valve part-way into development. Like Half Life 2, the game uses the Source engine, however, the visual in L4D are far superior to anything seen in the Half Life universe to date. The game pits four Survivors of an apocalyptic pandemic against hordes of aggressive zombies. We tested the third mission (Blood Harvest) with a complete run through "The Bridge" level,  8X anti-aliasing and 16x anisotropic filtering were both enabled and all in game graphical options were set to their maximum values.



The Aurora's framerates look a bit more pedestrian here as even the "wrong" GTX 275 configuration outpaces Alienware's beast. Again we see a 10 percent gap between x16/x4 and x8/x8. What's that gap look like when we move to Arkham Asylum—definitely one of the hot new games of the last few months?

Batman - Arkham Asylum
DirectX Gaming Performance


Batmat:Arkham Asylum

Batman: Arkham Asylum is a third-person action game developed by Rocksteady and is arguably the best comic book videogame of all time. Cast as the Dark Knight, you track down the Joker and bring him back to Arkham Asylum, Gotham's home for the criminally insane. However, the Joker escapes and takes over the asylum filled with hundreds of villains, which you must battle in this dark and creepy world. Gameplay consists of fist fighting, attacking from the shadows, and exploration. We tested the game using the built in benchmark at 1900x1200 with no anti-aliasing, PhysX off, and every graphical options set to their maximum values.



The Aurora pulls ahead comfortably. Again, we see a 10 percent difference between x8/x8 and x16/x4 configurations.

Transparent
Far Cry 2, Crysis

FarCry 2
DirectX Gaming Performance


FarCry 2

Like the original, FarCry 2 is one of the more visually impressive games to be released on the PC to date.  Courtesy of the Dunia game engine developed by Ubisoft, FarCry 2's game-play is enhanced by advanced environment physics, destructible terrain, high resolution textures, complex shaders, realistic dynamic lighting, and motion-captured animations.  We benchmarked the graphics cards in this article with a fully patched version of FarCry 2, using one of the built-in demo runs recorded in the "Ranch" map. Of the three available, we used the Ranch (Short) option.  The test results shown here were run at a resolution of 1920x1200 with 8X AA and Ultra High Quality presets enabled concurrently.


There are two performances of note in the above graph. First—Holy Hannah—the Alienware Aurora just frackin' tears the doors off the benchmark. The Dunia engine and the new ATI HD 5800 series are, apparently, the best of friends. The second item of interest is the performance gap between the x16/x4 Digital Storm configuration and the x8/x8. Up until now, all the deltas we've seen have been in the 10 percent range. Not this one. Shift a video card, boost your frame rate 40 percent—and lead the pack, save for the Aurora.

Crysis v1.21
DirectX 10 Gaming Performance


Crysis

Crytek's game engine visuals in Crysis are some of the most impressive real-time 3D renderings we've seen to date on a computer screen. The engine employs some of the latest techniques in 3D rendering like Parallax Occlusion Mapping, Subsurface Scattering, Motion Blur, and Depth-of-Field effects, as well as an impressive use of Shader technology. The single player, FPS Crysis is a smash-hit, and rightfully so. We patched the game to v1.2 with all of the game's visual settings to 'High' at 1920x1200 resolution to put a significant load on the systems' graphics engines being tested.


It would've been entirely fair to think our Far Cry 2 results were just an oddity—but they weren't. In x16/x4, the i5-750 hits a respectable (if unremarkable) 51.9 FPS.  Put the darn thing together correctly, and FPS goes up 44 percent. RTM—sometimes, kids, it really does matter. 

Transparent
Decibel Level, Thermal Testing


Sounding Off:
When the Digital Storm system arrived, we initially installed it in the test lab, within about 10 feet of my own desk. By the third day of testing it, I'd moved the system into a separate room altogether. Not only was the DS i5-750 system intolerably loud, it was loud at a particular range of frequencies that, while not immediately nerve-wracking, turned out to be the aural equivalent of Chinese water torture. The first two suspects were the power supply and case. The former was seemingly confirmed as the cause when we spoke to Digital Storm themselves—they "highly recommend" a much quieter and more expensive PSU from Corsair. When we went scrounging for information on the Cheiftech's decibel rating, what little data that was available pointed towards a dBA level between 42-47.

Initially, we took Digital Storm's word that the noise levels were caused by the power supply, but this turned out to be somewhat inaccurate. Near the end of my performance tests, we swapped out the Chaintech for a 750W PC Power & Cooling Silencer. This did help quiet the system—but only a little. Our second suspect—the HAF932—wasn't the direct culprit either. There's a short explanation and a long explanation for the system's dB output, you can choose which you prefer.

The Short Explanation:
The i5-750's 66dB boot volume and 54dB idle volume is not caused by any single component, but by several. The power supply, dual EVGA GTX 275 cards, and the company's choice of case and radiator fans all directly and materially contribute to the problem. The design of the HAF932 exacerbates the situation. Because the case is basically a giant sieve, pushing air through it creates more wind noise than is typical. Place your hand on the front of the case (which is nothing more than a giant grille, from top to bottom), and you can feel a distinct stream of air being pulled inside.

As constructed and shipped, the system is at least acoustically disagreeable; system noise can be substantially reduced with virtually no impact on the system's operating temperature. Given this, we strongly suggest Digital Storm modify the cooling options available from its website. Despite the fact that DS identifies our system as having shipped with "stock chassis fans," none of the fans in question are stock models from Cooler Master or Asetek—offering the option to install different, quieter fans should be no problem.

The Long Explanation:
We'll start with the Digital Storm i5-750's dB ratings as measured at a distance of one foot at the front, side, and back. Unlike a standard benchmark or temperature reading, decibel ratings are logarithmic. 63dB is twice as loud as 60dB, 66dB is four times as loud as 60dB, etc. A 40dB sound is 100x louder than a 20dB sound. Remember, a system is only as quiet as its loudest measured dB level. If the back of the case is kicking out 66dB, it doesn't really matter if the front is somehow just 20dB. The dB meter we used can only detect sounds that are 50dB and upwards.

The system's high boot volume is driven exclusively from the back of the case, which is much louder than the side. If you think about it, this makes sense. The side of the case has a single, 240mm low-speed exhaust fan, while the rear of the case is the exhaust point for the CPU radiator fans, two GTX 275s, and the power supply. Once the system boots and the GPU fans drop back to 40 percent, the system's sound is more evenly distributed.

The rear of the case was the obvious place to focus—and since we didn't want to change the review system's hardware configuration, that left the CPU radiator fans and front case fan. That, folks, is where things get really interesting.

Mi Casa, Akasa:
As we noted earlier in the review, the DS i5-750 uses two fans, one on either side of the radiator. This seemed a bit odd, given that one purpose of watercooling is to use less fans, but we assumed the company had outfitted the radiator with two low-speed, ultra-quiet fans, and was using two of them rather than a single, higher-speed unit. This turned out to be decidedly untrue. The two fans attached to the radiator are both manufactured by Akasa, model number: DFB122512L. There's a third, identical fan mounted in the front of the case.


It's not entirely clear what the specs on these fans actually are. It's not listed as a product on the manufacturer's website. The image is of a visually identical blue LED model, but the specs don't appear to match our baby (which could also be one of these beasts). The DFB122512L is listed by part-number over here, where it's specced as pushing 35.12 CFM and running at 25.24dB. For lack of a better reference, we're assuming the CPC-given specs are accurate.

Watch what happens to the i5-750's dB level when we disconnect the entirely spurious front fan and swap out the radiator fans for a
single Thermaltake TT-1225. (For the record, this one also glows blue, and matches the Digital Storm color theme nicely.) In the configuration measured below, the TT-1225 is mounted closer to the CPU and is pushing air over the radiator and out the back, rather than pulling it.

A measurement of "<50" means the decibel level was below the detectable threshold of the meter. Replacing the two radiator fans with a single, different fan cut the dB level at boot significantly. Once in idle mode, we couldn't pick up a reading at one foot from any of our previous test distances. Even if we assume the system's dB level is hovering right below the limit of the meter, at 49dB, we've still cut the system's sound by nearly four times. One thing to note is that the dB drop at the front of the system was caused by the disconnection of the front case fan, rather than by the radiator configuration changes. The reverse is also applicable—the front fan's status didn't impact our dB measurements at the back of the case.

The logical question at this point is to what degree these changes impact the temperature of the processor, as discussed on the next page.
Transparent
Decibel Level, Thermal Testing (Cont)


Thermal Analysis, Continued:
Instead of just relying on a CPU-centric program to task the processor, we opted to pound both video cards and the CPU simultaneously. For processor testing, we used DIEP. DIEP is a chess simulator developed by Vincent Diepeveen. The program scales excellently across multiple cores, and chess simulation itself is an inherently good stress test. The CPU is tasked with calculating every single possible move in a chess game—first in a single round, then two, then three, etc. This process hammers the L1/L2 cache, tests the efficacy of the processor's branch prediction unit, and indirectly measures ALU performance.

In order to simultaneously stress the video cards, we (ironically) fired up Furmark's stress test. It may not be a good predictor of GPU overclocking headroom, as we previously discussed, but it's fabulous at sucking down power and raising GPU temperatures. Under Furmark, the temperature of both GTX 275's rose steadily, peaking between 91-93C. We used CoreTemp 0.99.5 to record CPU temperatures; the numbers given here represent the average temperature across all four cores.

One side note: Simultaneously running DIEP and Furmark puts a substantial load on the PSU, often drawing 685-700W. The Cheiftech unit performed flawlessly in our tests, even under a sustained load of 12-16 hours.

Here's the configurations we tested:
  • 1 Akasa DBL122512L (CPU-side), pushing air over the radiator.
  • 1 Akasa DBL122512L (case-side), pulling air over the radiator.
  • 2 Akasa DBL122512L (stock configuration) simultaneously pushing and pulling air over the radiator.
  • 1 Akasa DBL122512L + 1 Thermaltake 1225, Thermaltake fan pushing.
  • 1 Thermaltake 1225 pushing air over the radiator.


Allow for some variance in ambient air temperature and for the fact that these numbers are averaged, and the already-small gap between these values shrinks even more. That's bad news for the stock configuration, acoustically speaking. The variance between the single Thermaltake fan and dual Akasa fans indicates that the user "pays" for a 3.5 degree temperature reduction with a system that produces 4x the noise.

The Storm and the Fury:
As shipped, Digital Storm's review system has a significant acoustic flaw in our opinion. It's not just loud—it's pointlessly loud. The additional volume delivers virtually no thermal benefit and contributes nothing to system stability. Even if the company wanted to include three fans, there are much quieter options on the market that are demonstrably capable of getting the job done; the per-fan premium would not add much to the total system cost either.

Transparent
Conclusion


Performance Summary: Once configured properly, the Digital Storm i5-750 was a great performer. The one-two punch of its high CPU overclock and dual GTX 275 cards performed well in light of the competition. The Digital Storm rig is an indirect advertisement for NVIDIA as well—dual GTX 275s competed relatively well against the Radeon HD 5870s in Crossfire of the Alienware system, and unlike the Radeon HD 5870, the GeForces actually available for sale right now.The system's high overclock, meanwhile, suggests that the only people who'd need a Core i7 are those with unlimited budgets, professionals using applications with a proven love of HyperThreading, or those who know, for certain, that the primary applications they run are actually limited by memory bandwidth. Even when the Core i5 can't match the i7, it's never far off, and the dual-channel DDR3-1600 delivers more than enough bandwidth for the vast majority of situations. While we had trouble with the overclocked GPUs, Digital Storm's 3.8GHz processor was absolutely solid, never crashing once in weeks of testing.

 

When the Digital Storm i5-750 is great, it's great. The company has clearly invested time and energy in certain places; the cable routing was done as neatly as we've ever seen and the system components struck a near-perfect price / performance balance. Customer service appears to be good, the OS image discs are intuitive, and the default OS installation isn't stuffed with lousy, performance-sucking bloatware. Evaluated on such merits, the DS i5-750 deserves nothing but praise.

It is, however, impossible to ignore the fact that the system only performed as well as it did after significant intervention. Evaluated individually, any one of the three problems we had—overly-agressive GPU overclocking, misconfigured SLI, and noise—are no big deal. Mistakes happen. Taken in aggregate, however, the situation changes. Based on our experience, we'd recommend Digital storm expand its repertoire of tests before it certifies video cards as "overclocked as much as possible with complete stability," as the Digital Storm customization documentation states.

Second, the company needs to ensure that the enthusiast, boutique-level systems it's selling are actually configured properly. The P55 FTW was a new board when this system shipped, but EVGA spelled out the proper configuration in both the manual and on the motherboard. This is the type of mistake that one wouldn't expect in a $399 white-box, and it has a substantial impact on 3D performance. That's another way of saying it has a substantial impact on the buyer's satisfaction level, should he or she go looking for comparative performance results. The entire rationale behind buying a custom-built system is that there's a certain degree of high-performance tuning baked in. If the end-user has to come along post-purchase and retune / rebuild the system properly, it rather kills the point of buying it in the first place.

At present, the only noise-related option Digital Storm offers on any of its systems is a $99 "Noise Supression Package," in which they install custom-tuned fans at specific speeds and add insulation material. For comparison, the various custom overclocking options start at $35. Fan upgrades are available on the fully customizeable systems, but these are identified as "Zalman Performance Fans," not quiet fans. Given the results of our decibel and thermal tests, there definitely needs to be more customization options here, and possibly a configurator notification if a customer selects a certain set of hardware that's been identified as cumulatively loud. A simple "the configuration you've selected may exceed XX decibels when gaming, we recommend x, y, or z" would do the trick.

Would we recommend it? That's tricky, as laid out in the pros and cons above. If the system Digital Storm delivered was identical to the one we received, it's tough to overlook the missteps. If, on the other hand, the system that dropped on your doorstep was identical to the review system as it is today, we would, in a heartbeat. If Digital Storm closes the gap between the two—and none of the issues are hard to fix—then we'd recommend it wholeheartedly.

Update - December 17, 2009: Digital Storm has provided some information regarding the issues we experienced during our review and the QA improvements the company has made to address them.  You can read the company's response in its entirety right here.

 

  • Top-tier performance in all categories
  • Neat, clean interior
  • Stable CPU overclocking

 

 

  • Sub-optimal initial configuration
  • Noisy

 


 



Content Property of HotHardware.com