Page 1
    Page 2
   

 
 
By William Van Winkle
  
 
If you want to become an artist with a Dremel drill, we can’t help you. If you want to dabble in liquid nitrogen and enough cold cathode tubing to light up a city block, you’re reading the wrong article. Sure, that stuff is fun. No honest geek could claim otherwise. But unless you want to make hobby-type extreme computing your core business, fighting for sales here is going to be an uphill battle—you versus a fleet of specialty etailers. Most likely, your main business is in mainstream systems. We think there’s a niche for enthusiast-class systems within the mainstream market, systems that deliver nearly all the performance benefits of the “L337" (that’s “Leet,” meaning “elite,” for the uninitiated) crowd but do so using the components and vendors with which you’re already familiar. Once you have the knowledge to identify and maximize these deceptively common pieces, you’re ready to reclaim your share of enthusiast sales.

  
 
There are two schools of thought in consumer computing: get stuff done and get stuff done faster. Gamers are acutely aware of this. Most people just want to play, but some people thrill in the competition, and consistent winning in many games requires not just skill but raw hardware performance. Extreme gamers will drop thousands of dollars on a competition-class system; the rest of the gaming world dreams of having such luxury but makes do with less, trying to get the best bang possible from the bucks they have. This is no small group. According to a 2007 study sponsored by AMD, 61% of new PC buyers in North America plan to use their machines for gaming. Among the 263 million PC gamers worldwide, only 13.1 million are “enthusiasts.” The rest just like gaming.

Beyond the fragging field, some consumers need workstation-type machines for everything from video editing to CAD design. Some might be day traders driving four applications across six screens in real-time. Some may be running home businesses and habitually leave a dozen or more applications open concurrently.

Beyond the Scope
Zalman Reserator towers are fun and sexy, but water cooling is often the domain of high-end enthusiasts. Only a handful of mainstream enthusiasts are likely to bear the extra expense and footprint.

Performance matters. At the same time, most consumers watch their wallets, and if more performance can be had for less money—or at least less money than you’d expect to find for a tricked-out, turbo-charged enthusiast tower—than they’re happy as clams. We call this group “mainstream enthusiasts,” people who have mainstream mindsets and budgets but performance needs more common to the enthusiast sect. There are far more mainstream enthusiasts out there than true enthusiasts, just as there are more people who enjoy playing basketball than there are pro basketball players. There are a few opportunities for system builders in the high-end enthusiast market, but this is a tough space to target. OEMs like Alienware, Voodoo PC, and Falcon Northwest have a near lock on this market, and the consumers not buying OEM boxes tend to be the DIYers buying components through etail.

Roughly speaking, the high-end enthusiast space, on a good day, might constitute 2% of the market. We’d rather see you go after the 20% share constituted of mainstream enthusiasts. You’ll make close to the same ASPs and margins as with the elite systems but be aiming at a much bigger bull’s eye. How does one do this? It’s not about flashy looks, nor about picking the most expensive item on the component menu. Instead, your challenge is to find the parts with the highest performance value and then know how to deploy them for maximum benefit. This is what will set you apart from the mass merchants, etailers, and OEMs and establish you as an authority in the enthusiast mainstream space.

A Question of Caveats:

Performance discussions inevitably turn to overclocking, so let’s jump right to it. Overclocking is your No. 1 tool for getting more speed from regular PC components, but it’s not a practice that has sat comfortably with system builders over the years. Overclocking at its simplest means increasing the clock speed, or frequency, of a component beyond the factory-recommended level. Vendors like Intel and AMD have spent many years cultivating a love-hate (mostly the latter) relationship with overclocking. In the ‘90s, before many of today’s protective measures were built into processors, overclocking was much more likely to burn out a chip, which in turn left the vendors in a nasty RMA situation with enthusiasts. Besides this, the ability to quickly make a few BIOS changes and knock a CPU up two or three speed bins, effectively turning a $300 chip into a $500 one, had obviously negative implications for vendor bottom lines.

The love side of the overclocking relationship pops up occasionally when the stars of technology, public opinion, and competitive expedience align. For instance, in May of 2006 before the launch of the Core 2 microarchitecture, Intel was at the height of a price and market share war with AMD in the consumer processor space. The 2.66 GHz Pentium D 805 launched as a low-end ($130) SKU on a 533 MHz front side bus. This was one of the lowest speed bins on that chip stepping, meaning that this particular stepping group had a lot of frequency overhead beyond that 2.66 GHz level. In fact, it didn’t take long for tweakers to discover that the Pentium D 805 could be overclocked to above 4.0 GHz, outperforming even the top-tier Extreme Edition CPUs of the day. Did system builders take advantage of this opportunity to deliver a lot more performance to buyers at a significantly lower price? Smart ones did. (Note that CPU price deltas don’t simply boil down to frequencies. Lower-end CPUs are likely to feature less L2 cache, lower bus speeds, and other things that may not show up in benchmarks but could impact real world performance.)

Now the bad news: Just because a chip can be run at higher-than-spec frequencies doesn’t mean it should be. Moderate overclocking for short periods is a far safer route than an aggressive overclock left running 24x7. Imagine microwaving a raw egg for 10 seconds versus 10 minutes. A little heat is OK, but too much is going to leave you with a nasty mess. Similarly, that raw egg may keep for weeks in your refrigerator, but leave it on the counter at a 35-degree-higher ambient temperature and you know that the time to failure (rot) is going to be significantly accelerated.

ENTIRELY TOO COOL
If your customer has to resort to liquid nitrogen for CPU cooling, you probably have an RMA in the works. Still, enthusiasts at Intel’s 2007 Taiwan Overclocking Live Test Tournament pushed the envelope.

Chip manufacturers use the mathematical relationship between heat and longevity to arrive at their MTBF numbers. One example at Active Hardware (shows how a chip rated to run for 30,000 hours (eight years and three months) at 25 degrees Celsius drops to about 13,500 hours (three years and nine months) when run at 50 degrees Celsius. Manufacturers combine these variable equations with heat chamber environments to backtrack their way into an average use MTBF spec. If the manufacturer determines that its product should run for 10 years under normal, recommended conditions and six or seven years under lightly overclocked conditions, that’s still ample room to feel safe in offering, say, a three-year warranty. But remember that MTBF is about averages. The more you scoot that MTBF bell curve to the left along the time axis, the more failures you’re going to see in only one or two years—before the regular replacement cycle of that PC. And manufacturers state that they can clearly tell when failures have been caused by non-spec temperature and/or voltage conditions, which are a warranty violation.

“Heat can cause electromigration problems,” says John Bruno, senior manager of the chipset ASIC design division at AMD. “It can cause on-board components to unsolder themselves. You can get shorts. And you can run into actual transistor-level wires that burn. We have intricate routes on these chips that are very, very fine. You start increasing voltage and you can very easily burn things. That might scare away everybody, but the truth is that there are ranges that are fairly safe. I’m not going to say just how much I think is safe because it varies product by product—this or that process, this or that foundry, on this board with this heatsink. There’s so much variance that it’s difficult to give an exact number. But there are ranges where you don’t significantly affect the mean time between failure. If you keep your temperatures under control and don’t go crazy with the voltage, you’ll still enjoy a full life out of the PC.”

Another point to consider is the legal ramification of overclocking. If a user goes nuts with overclocking, the chip ignites the motherboard, and the user’s house burns down, that poses a potential liability nightmare for both the chip manufacturer and the company that resold it. This is why you see Intel and AMD continue to make disclaimers and admonitions against overclocking even while producing ever more overclocking-friendly parts and tools. It’s also why you need to proceed very cautiously and prudently with overclocking. Uninformed buyers who frequent enthusiast Web sites can find themselves in the same boat of preconceptions as people who think that women should all look like the models in Cosmopolitan and Vogue. Just because there are a lot of pictures of exceptional examples doesn’t make those exceptions the norm, and you’d be a fool to think they were. The extreme overclockers posting into those forums, whatever their backgrounds, often have money to burn (literally) on overclocking hardware to death in months or even weeks. Most of us, particularly in business settings, need our systems to perform well and dependably for several years.

Overclocking can be a tremendous asset for your mainstream enthusiast sales efforts, but, like most things in life, too much of a good overclocking thing can come back to bite you.

CPUs

When it comes to system performance, there’s no getting around the CPU being the system’s star player. To add value in a mainstream enthusiast system, you’re almost bound to start here. Of course, there are only two camps to discuss, Intel and AMD, and we’re not here to play nor pick favorites. Each side has its own benefits. True, if you take a spin through the major Web hardware reviewers, you’ll quickly see that today’s 45 nm-based Intel Core 2 processors hold the top spot in the lion’s share of benchmark tests, and a mainstream enthusiast is likely to pay close attention to such reviews. That said, we also know that mainstream enthusiasts are going to be varied in their budget restrictions, and AMD does have an intriguing play here with its triple-core designs

You’ll Never Go Back
AMD Phenom Black Edition processors feature surprisingly low pricing and an unlocked multiplier, allowing system builders and users far more flexibility in fine-tuning and optimizing performance.

The AMD Phenom X3 group currently has three SKUs—the 8450 (2100 MHz), 8650 (2300 MHz), and 8750 (2400 MHz)—all of which have a power consumption of 95 watts. The xx50 model naming notes a part with B3 stepping, the TLB bug-free successor to the prior B2 stepping. Online pricing for the 8450 starts at $145; the 8750 starts at $195. To take a reference point, Newegg’s least expensive Intel quad-core chip is the 2.4 GHz Core 2 Quad for $200. Picking a $160 price point, we find the Core 2 Duo E6550 at 2.33 GHz. So for about the same price and frequency, AMD throws in an extra processing core. Buyers a bit tight on cash but with apps or usage models that favor multi-threading will find this very persuasive.

Here’s one more AMD play for mainstream enthusiasts—Black Edition chips. As with Intel’s Extreme Edition processors, Black Edition CPUs are just like their non-Black counterparts, only the multiplier has been unlocked. For those new to processor tweaking, the two ways to alter a CPU’s speed is through its frequency (clock speed) and the multiplier value applied to that frequency. The Phenom X4 9850 Black Edition, for instance, contains a base frequency of 200 MHz. Because this is a 2.5 GHz part, you know that 2,500 GHz divided by 200 MHz yields a multiplier of 12.5. With a standard 9850, you can go into the motherboard BIOS and increase the frequency but not the multiplier. With the Black Edition, you can change both. A 15 multiplier on a 9850 running at 200 MHz gives you a 3.0 GHz final speed. A 240 MHz clock with the default multiplier of 12.5 gets you to the same place. So which is better to change—the frequency, multiplier, or both? Different system configurations will respond differently to changes in the multiplier and frequency, and the only way to know for sure is to try out the various possibilities.

“Sometimes, to get the really high frequencies, you actually need to decrease the multiplier and increase the front side bus frequency,” elaborates Intel product marketing engineer Erik Cubbage. “With a default speed of 400 and a multiplier of 8 on a 3.2 GHz quad-core processor, say I take that up to 500. That might give me better performance than simply changing the multiplier. I still get 4.0 GHz, but, depending on a number of factors, I could end up with a more stable platform.”

Hold On Tight
Intel’s latest 45 nm-based quad-core processors are today’s performance champ. With models spanning from $200 to over $1,000, mainstreamers can own as much speed as their budgets allow.

With AMD’s Black Edition CPUs, you get the added flexibility to experiment with different speed configurations and arrive at the best blend of performance and stability. The added bonus with Black Edition chips is that AMD has priced these unlocked parts far below Intel’s unlocked equivalents. For example, if your buyer only needs a dual-core processor, the 2.6 GHz Athlon 64 X2 5000+ Black Edition sells for under $80. Presently, there isn't an Extreme Edition Intel chip on Newegg for under $1,000. Obviously, these are apples and oranges, and Intel does have older dual-core Extreme chips you might find through secondary channels, but that carries its own set of complications, not the least of which may be warranty concerns. For those who want to overclock on a budget, Black Edition is the way to fly.

On Intel’s side, there’s a broad spectrum of choices—with excellent options spanning from under $200 up to well over $1,000. While AMD is working to carve out a niche for its triple-core models, Intel is doing everything in its power to make quad-core ubiquitous throughout desktop computing. In particular, we would recommend a 45 nm-based model for mainstream enthusiasts. The Penryn processors are more power-efficient and often feature higher overhead capabilities than their 65 nm counterparts.

On this basis, perhaps the choice starter SKU for our audience might be the Core 2 Quad Q9300. With 6MB of L2 cache and a 1333 MHz front side bus, this 2.5 GHz chip is one of the lowest speed bins in its group. We haven’t tried overclocking the chip in-house, but the odds are decent that this model has good potential. To reinforce an earlier point, though, note that the 2.83 GHz Core 2 Quad Q9550 sells for about $560. Could the Q9300 achieve a 13.2% overclock to 2.83 GHz without posing any serious stability or heat risk? Very probably. But the Q9550 features 12MB of L2 cache, so if your buyer has apps that benefit from more cache, then that price doubling may well be worth it. There are times when cache is king, and if this is one of those times, be aware that the 2.66 GHz Q9450 sells for over $200 less than the Q9550, making it our favorite price/performance compromise SKU for the mainstream enthusiast sector.

There’s no question that Intel’s quad-core Extreme Edition CPUs are the fastest chips on the market. AMD can’t touch them today, and enthusiasts who demand the utmost in speed will gladly pay four figures. Extreme chips will offer more overclocking overhead and dispense with the multiplier lock. We mention the Extreme chips here because they are the best, but their applicability to a predominantly mainstream audience will be limited. The Q9550 is representative of the kind of part you should be considering to nail that mainstream enthusiast sweet spot.

Before we leave the subject of CPUs, we should touch on cooling. Enthusiasts often embrace liquid cooling systems, but we’re going to shy away from that product group here. Again, we’re looking for component pieces that will be familiar to you, parts from large, reliable vendors that require almost no learning curve. Liquid cooling can be simple when you buy a self-contained kit, but it veers away from our focus and pertains more to true enthusiasts than mainstream enthusiasts.

COOL BUT QUIET
Stock cooling tends to be noisy, especially on overclocked systems. By paying a bit more for a premium cooler like Thermaltake’s MaxOrb EX, buyers get fewer decibels and greater headroom.

Instead, we prefer active air cooling products from the likes of Thermaltake and Zalman. Thermaltake’s $70 MaxOrb EX is an excellent example of a premium cooler for mainstream enthusiasts. A 120 mm fan sits within a ring of all-copper radiator fins shot through with copper heatpiping. It’s a big honker (143 x 144 x 95 mm), and you’ll need to make sure it fits in your case without disturbing other components or chassis walls. But with a noise output of 16 to 24 dBA, this kind of chiller can shave many degrees from a heavily loaded CPU in near-silence. And if you have to make a nod to some cosmetic bling, this is a good place to do it. To offer most of the cooling benefits for almost half the price, consider an aluminum rather than copper option, such as Thermaltake’s GunMet Orb.

Keep in mind that these premium coolers will probably offer a more incremental than radical improvement versus stock heatsinks. Depending on the model, you may only see a five degree Celsius variance. For customers only interested in default speeds, this may be unnecessary. For overclockers, a premium cooler is a cheap and sexy insurance policy. Interestingly, the need for such coolers may actually diminish over time.

“The more we get into the future,” says AMD’s Bruno, “the more we run into something called temperature inversion. You actually see less margin at lower temperatures. We’ve reached the point where this is a real concern. Forever, we’ve gone under the assumption that if you keep a piece of silicon colder, it’ll run faster. That may not be true these days. We can find scenarios using these very fine geometries—65 nm, 45 nm, and beyond—where reducing the die temperature actually gives you less margin. It’s a pretty complex thing that happens at a transistor level, and there’s no easy way to explain it, but that’s something we’re dealing with now at these geometries. So cooling benefits used to be a 100% statement. Now, it may only be a 90% statement, and it’ll likely continue from there.”

MOTHERBOARDS

 CPUs may get most of the headlines when it comes to high performance, but a great CPU can still be crippled by an insufficient motherboard and/or chipset. As the heart of the motherboard, the chipset details are like life and death to a successful mainstream enthusiast system. If your buyer goes buck wild and buys a new Intel Extreme Edition processor based on a 1600 MHz front side bus connection but plants this on a motherboard based on last year’s P35 chipset, the chip will work, but the processor will be limited by the chipset’s 1333 MHz bus. Moreover, not only are different chipsets built to handle varying amounts of overclocking, but motherboard manufacturers are free to dictate what BIOS options are available in each board model for manipulating the system components.

BUILT TO BE BETTER
Intel’s DX48BT2 (“BoneTrail 2") board competes with other enthusiast platforms, but its superior engineering lets users dispense with a lot of the extraneous and potentially risky copper add-ons.

The end-user confusion compounds when you figure that the motherboard reference designs so often reviewed by the press don’t actually make it onto store shelves. (This is less of an issue for Intel, which more or less turns its reference designs into finished SKUs because it controls the entire process from chipset design to motherboard production. Obviously, third-party manufacturers are free to make whatever changes they please, but we can’t help but feel that this end-to-end control is largely responsible for Intel’s reputation for stability.) A various number of motherboard manufacturers create their own versions of the reference designs put out by Intel and AMD. Some of these are skewed toward low-cost markets, forcing compromises in order to meet a certain price target. This is why you can’t judge a board by its manufacturer. Most make excellent products, and most of the same names make stuff that many would call junk. You have to judge the specific model.

Even an experienced tech can’t tell a board’s quality through visual examination. Sure, you can make educated guesses based on the board thickness, how well the manufacturer centered components within the PCB’s white placement lines, and the type of components used. You know that a P45 chipset is probably going to deliver a more robust performance experience than a G33. But the real differences are in the things you can’t see, such as the quality of the voltage regulator module (VRM). A 100W CPU turned up to 120W probably won’t roast before your eyes; Intel and AMD implement enough thermal management tools to prevent such catastrophic failures. However, those extra 20W cycled through a cheap VRM might toast one of the board’s metal-oxide semiconductor field-effect transistors (MOSFETs). This isn’t a “catastrophic failure” per se, but it’ll kill the motherboard just the same. These low-end boards are built to run at spec and nothing more. You should never sell one to a mainstream enthusiast just to close a deal; both of you will regret it soon enough.

“We’ve seen tier-ones build boards for, say, Far East markets that are stripped way, way down,” says AMD’s Bruno. “It’s common to see them have a problem if you start overclocking. They’re built and validated just enough so that at stock speeds everything is OK. These same people build high-end boards that you can put through liquid nitrogen baths and crazy voltages and they will not burn. It’s not about this or that vendor. It’s about what kind of board you have. I wouldn’t expect an IGP microATX budget board to survive as well as a platinum-class, quad graphics, mega-heatpiped, six-layer board. Same vendor, same basic chipset, totally different overclocking results. We control our chips, and we design them with excessive margin knowing that people might do this. But in the end, it’s more than the chip that will determine the results.”

This touches on the issue of voltage, which is an integral piece of the overclocking puzzle controlled through the motherboard’s BIOS. Just as you’re going to burn more gas when you accelerate a car, increasing voltage to a component is often required if you’re going to overclock it. Higher frequencies demand more electricity, and that means higher voltage, whether we’re talking about the CPU, chipset, GPU, or system memory. You can increase the voltage to all of these components, but you’ll be pumping more electricity into a part that wasn’t necessarily designed for it. This is when you need to start watching temperatures more closely. If a customer wants to engage in voltage overclocking for any extended period of time, we strongly recommend liquid cooling.

Easier Total Control
Forget monkeying with BIOS settings. Intel Desktop Control Center and AMD Overdrive give you precise, GUI-based control over practically every speed feature in the desktop, including handy presets

“I can’t speak to the likes of ASUS, Gigabyte, or MSI,” says Intel’s Cubbage. “We don’t do those kinds of competitive tests. But on the BoneTrail 2 [model DX48BT2], we actually space the voltage regulators out farther than most other board manufacturers do. We space out the individual phases farther and also space them farther from the CPU. That helps eliminate a lot of electrical cross-talk. And the pins that come off the voltage regulators, we extend those out to the bottom of the board, make them a little bit thicker, and those serve as a back-side heatsink. You can’t see it. Even if you look at the board, it would be tough to tell the difference. But on the back of the board, we actually use the copper traces to dissipate heat. If you were to rip off the heatsinks that we put on our voltage regulators on the BoneTrail 2, you can run an Extreme Edition CPU at 150W with no airflow and no heatsinks. Now, a lot of boards depend on the airflow generated by the CPU fan to blow across the voltage regulators and other components, including the memory, to cool it down and run within spec. But a lot of enthusiasts use liquid cooling, and with that you don’t get the airflow on those other components. Like I said, with no airflow and no heatsinks, you can take the BoneTrail 2 up to 150W. With airflow from a lateral or top-mounted case fan—what a typical airflow would be in a decent chassis with a couple of fans—our voltage regulators can handle over 200W.”

We should point out that quality of motherboard construction isn’t a black or white affair. For instance, an Intel P45 platform will cost about $100 less than an X48 platform. Never mind the difference in FSB speeds (1333 MHz versus 1600 MHz, respectively). The X48 features a slightly higher quality VRM and more board layers for better electrical communication at the trace level. This translates into the extra performance edge demanded by high-end enthusiasts. Does this mean that mainstream enthusiasts have to buy an X48 platform? Not at all. Most buyers will be pleased as punch with the P45, which should exceed most mainstream enthusiast expectations. Should their passion for the technology continue to increase, an upgrade is always possible.

These sorts of engineering-level details are what you don’t get from a spec sheet. What you do see, in many cases, are ever-growing webs of board-mounted copper heatsinks and heatpipes. We love the eye candy as much as anyone else, but the reseller’s challenge is to assess how much of this added bling is conveying true value. Cubbage’s point about liquid cooling is an excellent one. How much heat dissipation will those on-board heatsinks provide when there’s no central CPU cooler driving airflow? You need to watch for this and increase case ventilation accordingly. Also consider that some motherboards feature hard-mounted chipset heatsinks, which may or may not be up to the task of handling the extra heat from running under prolonged overvoltage. A good mainstream enthusiast board will let you remove every heatsink element and upgrade it if necessary.

“I guess the reason we don’t use all those things is because we don’t need them,” says Intel’s Cubbage. “If you design the board well to begin with, with an overclocking user in mind, we can get more from our little heatsinks than competitors do with their fancy copper heatpiping. And as a side benefit of that, it significantly reduces the complexity on the board. Yeah, some people want to pimp their rig, but a lot of users just want something simple that’s going to work really well. More heatsink and board-mounted fan complexity injects more failure points. If you ship the system, you now have a shock and vibe issue working on those extra parts. We have slow-motioned videos of boards in shock and vibe tests bending a quarter of an inch or more. Consider the potential impact of that bending on extra board-mounted components.”

We’ve done a fair amount of motherboard overclocking work, and it’s one of our least favorite aspects of working with enthusiast products. As with fast cars, overclocking can be a passion, a science—even an art. Monkeying with frequencies, multipliers, and voltage has historically meant a seemingly endless loop of small increases to BIOS settings, a reboot, and running through a battery of benchmark tests. If everything passes and yields a higher set of scores, you reboot into the BIOS and repeat the process, over and over, for hours until there are signs of instability or outright failure. This could take the form of a system lock-up so hard that you have to unplug the power cable, remove the CMOS battery, and reset the CMOS switch. (This is why enthusiast boards now feature power and CMOS clear buttons, because they’re more convenient than jumpers.) Worse than this is when an overclock corrupts the operating system, and you don’t have to endure this too many times before figuring out what a great idea it is to clone the original test image before you start overclocking.

Some people enjoy this; we don’t. Perhaps it was for more mainstream-minded enthusiasts like us that Intel and AMD both created Windows-based overclocking tools that come bundled with several motherboards based on their platforms. Intel bundles its Desktop Control Center utility with Extreme Edition and some higher-end mainstream boards, including today’s DX38BT and DX48BT2 models, and works with Passmark’s BurnInTest and PerformanceTest to establish a performance baseline and then evaluate subsequent increases. Users can monitor and fine-tune major component frequencies, memory timings, chipset parameters, fan speeds, and thermal set points. There are also preset modes, such as “quiet” and “gaming,” that can make modifications even more convenient. With this, there’s no more need for the endless loop of reboots. Users can make changes and test the results in real-time. If nothing else, it’s a handy tool for resellers to use for preconfiguring mainstream enthusiast systems with extra performance before they go out the door. One of our favorite uses is figuring out how far frequencies can be pushed before additional voltage is needed and then we pull back a couple percentage points and lock down the settings. The Desktop Control Center lets us do this in minutes whereas, previously, the process would have taken hours of BIOS twiddling.

AMD’s Overdrive utility functions in a manner similar to Intel’s Desktop Control Center, but the interface is more detailed. There is an easy novice mode that allows for changing clock speeds but not more advanced aspects such as voltages and multipliers. We also like Overdrive’s auto-overclock tool, which will crank up the system until it fails and then establish an optimal configuration. It’s possible that an expert overclocker could arrive at a higher performance configuration, but the difference will be in the single-digit percentages. For the time saved, Desktop Control Center and Overdrive are close enough to perfect for practically any mainstream enthusiast.

We also recommend pairing these performance tools with a simple device like a Watts up? meter, which measures voltage and the power consumption of a system from the wall. Because we’re talking about mainstream buyers, power consumption may be more of a consideration than with pure enthusiasts. Customers should understand the full implications of overclocking, and while they may not want to pony up $100 for a base-level power monitor, your being able to show in real-time the effects of elevating mainstream enthusiast PC’s performance is another value-add that other providers are unlikely to offer. ...more
  
         
    Back to top
Page   1 2
   
   
Copyright © 2008 RAM Magazine. All rights reserved.
Do not duplicate or redistribute in any form.