Maximum PC proprietary RSS Feed

Murphy's Law: Apple Opens Up to Closed Standards

David Murphy — Thu, 28 Jan 2010 10:00:00 -0600

I'm not sure which of these is a more compelling criticism of the Apple iPad: "They named it what?" or "Where's the Flash?"

It's no secret that Apple harbors no love for Adobe's Flash architecture. John Gruber over at Daring Fireball recently wrote up a wonderful treatise as to why this is the case. If you have a spare hour or so, I recommend giving it a look-see. I'll spoil the ending for the sake of continuing on with this column: Flash is a proprietary architecture that Apple has no control over. Thus, when Flash-based elements wreak havoc on the stability of Apple platforms, Apple can't do much to fix the issue--nor can the company convert the 32-bit Flash binary over to Apple's goal of a system-wide, 64-bit experience.

Naturally, Apple's only real choice has been to dump support for Flash in certain use situations--namely, the company's iPhone (and surely the iPad as well, given the supposed similarities in their underlying operating systems). In order to incorporate a similar level of interactivity and video processing as delivered by Flash, Apple's turned to a combined replacement of HTML5 and JavaScript. But the media codec that Apple's thrown its support behind for HTML-based video rendering is H.264. Ain't nothing open about that.

The enemy of Apple's proprietary enemy might be the company's friend, but it's no friend to the Internet.

Unlike the video compression Ogg Theora--the competing solution for HTML5 video rendering that's been championed and incorporated by Mozilla--H.264 is patented. You can't redistribute H.264 codecs per the license, nor will you likely be able to stream your own H.264 content without paying the piper at some point in the future. To Mozilla, this is just a complete slap in the face to the company's belief in a royalty-free Internet. Without the innate ability for anyone to freely code some HTML, churn CSS, or hack JavaScript, the Web would not be in the state it is today. Licensing core technologies stunts innovation.

Oh, and H.264 also violates the GPL for those seeking to spin off of Mozilla's initiatives--the free distribution rights guaranteed by the GPL doesn't mesh with code that's patent-encumbered. Or, as said by Mozilla's vice president of engineering, Mike Shaver:

"We want to make sure that the Web experience is good for all users, present and future. I want to make sure that when a child in India or Brazil or Kenya discovers the internet, there isn’t a big piece of it (video) that they can’t afford to participate in. I want to make sure that there are no toll-booth barriers to entry for someone building a whole new browser, or bringing a browser to a whole new device or OS, or making and using tools for creating standard web content."

Thus, we are left with fragmentation between those that stand behind H.264 and those that, for whatever reasons, would prefer to support the competing open-source codec Ogg Theora. Apple doesn't believe that Ogg Theora has enough hardware support to warrant the switch, not to mention the confusing patent situations the codec could find itself in were it to attempt to improve its quality to H.264-levels. Google, who commands considerable power in the discussion thanks to YouTube, has almost single-handedly decided matters by opting for H.264-based encoding (even though the Chrome browser supports both codecs). Mozilla, as mentioned, refuses to incorporate the H.264 codec on principle alone.

But can you really blame Apple? I suppose the company is caught up between a digital rock and hard place--with the Flash architecture closed off to Apple's tweaking, the company has no choice but to opt for the combination of HTML5 and H.264. Yet this, in itself, presents a situation that won't necessarily close Apple off to delivering rich media for its devices, but one that impacts the general openness of the Web as a whole. It's going to take a bigger iPad to wipe up all that bad Internet karma.

David Murphy (@ Acererak) is a technology journalist and former Maximum PC editor. He writes weekly columns about the wide world of open-source as well as weekly roundups of awesome, freebie software. Befriend him on Twitter, especially if you have an awesome app or game you're dying to recommend!

Release Notes: Everything Just Needs to Be More PC

Will Smith — Tue, 08 Dec 2009 12:15:00 -0600

There are dozens of different computing devices in my home, ranging from the common—TVs, PCs, smartphones, and digital picture frames—to the unusual. Some of the more eclectic gizmos, like smart alarm clocks and various types of music streamers, deliver kick-ass functionality on their own, but there just isn’t much communication between these devices. There are dozens of different protocols and software interfaces designed to foster communication betwixt electronics kit, but none of the manufacturers use them. Seems like all the cutting-edge hardware we buy these days uses proprietary cables, software, and communications protocols.

Sometimes propriety is the price of progress: A product includes some new functionality that requires more than existing technology allows. Sometimes a vendor chooses one standard over a different competing standard. And sometimes it’s just sheer bloody-mindedness on the part of the manufacturer. But regardless of the reason, it’s unacceptable.

Apple does a great job of integrating its gear with other Apple products, but is notoriously bad about integrating with third parties. For example, I still can’t pull photos from my Flickr account into my iPhone without using a third-party app. Likewise, there’s no way to stream the music collection stored on my Windows Home Server to an AppleTV, unless I use Apple’s proprietary iTunes software.

Microsoft is just as bad—some MS products from the same product lines can’t even talk to each other. While my Windows Home Server will stream video and music to my Xbox, the server doesn’t recognize many common file types, including ones the Xbox supports! And while we’re talking about it, why doesn’t my Home Server speak Homegroup, the new scheme Windows 7 uses to share files and printers?

Of course, the humble PC stands in diametric opposition to all these closed, proprietary systems. There are multiple, complementary interfaces to connect to the PC—fast and slow, internal and external, wired and wireless, hardware and software. And, because these specs are open and (in many cases) royalty-free, anyone can connect to, and extend, the PC. This open platform, combined with the speed gains and cost savings delivered by Moore’s Law, powers the technological revolution we’re living in right now. Were the PC not an open platform upon which everyone in the industry innovated, the PC wouldn’t be as fast and cheap as it is today.

So, what’s my point? If you’re a vendor, stop worrying about protecting your market share with proprietary cables and interfaces, and instead focus on differentiating your products the old-fashioned way: Make them better than the other guys’. If you’re just a person who buys this stuff, only spend your money on products that play well with others. If your gear doesn’t work as you’d expect, call the support line and demand to know why. Your gear should never be stymied by conflicting “standards,” whether that aligns with the vendor’s corporate strategy or not.

Novell: It's Tough to Make Money with Open Source, "Keep it Proprietary"

Paul Lilly — Fri, 20 Nov 2009 07:24:20 -0600

Despite the growing popularity of open source software, there's still the issue of how to make money with it. No easy task, warns Miguel de Icaza, Vice President of Novell, who also heads up the firm's open source Mono project.

"If your livelihood depends on the product that you're selling, until you can figure how you're going to make money on that thing, I say, keep it proprietary," de Icaza said.

The VP went on to say that it's "incredibly difficult" making an open source business. His remarks were in response to an audience member at the Microsoft PDC (Professional Developers Conference), who raised the question of making money via open source. The issue of making money by selling support also came up.

"You need to take those steps carefully in my opinion," de Icaza said. "And support, by the way, is a horrible business. I want to be writing code, and I want to be paid to write code."

The VP did note that if you're a young developer without a lot of obligations, like a family and tuition, then it's far easier to consider doing open source projects.

Image Credit: daveyost.com

Murphy's Law: What is Skype Up To?

David Murphy — Wed, 04 Nov 2009 12:00:00 -0600

The recent announcement of Skype turning quote-unquote open source has me twirling a finger with delicious glee. It's not that I dislike Skype. And it's not that I'm about to get into one of my 1,500-word debates on the differences between the definition of "free" and "open-source," I promise. This is nevertheless an important premise of Skype's entire move, as some Internet commenters are crying foul that Skype is only half-opening its popular application to the crowd. The GUI code will be yours to play with as you please. The underlying Skype protocol... nope!

To them I say: Duh.

I don't want to put words where they don't exist, but I'm willing to bet that Skype's sudden shift toward open-source waters has more to do with applying a giant, universal band-aid to staggered Linux development. It's not quite an altruistic gift to the community so much as it is a package and a bow with the phrase, "you fix it" written on the label. And that's fine. Let the community create the functional GUIs for Skype. It would be suicide for the company to release its heavily encrypted voice protocols for common use.

So what, then, is Skype up to?

Stronger Integration

Linux isn't just for geeks in the basement anymore. The open-source operating system powers an innumerable amount of devices, including super-small netbook PCs, equally small tablet PCs, mobile phones, et cetera. While opening up the Linux GUI still requires the underlying architecture of the Skype client in order to have any semblance of functionality, this is at least a first step toward smoothing the integration of Skype into Linux-based devices of all shapes and sizes. Developers could freely embed the Skype client and skin it to match the look and feel of a particular device--surely a boon versus, say, waiting for Skype to come into the app center of-sorts for each and every permutation of product.

Third-Party Takeover

And we now venture into the realm of the hypothetical. Imagine, if you will, that Skype's opening of the Linux GUI is but the first step in a gradual elimination of the Skype Client as a whole. Sure, this downloadable application would still exist for those that care to use it, but what if Skype was looking to offload the client's functionality into any software platform that wants to take it? Skype would still control the underlying protocol (depending on how the legal battle sorts out), yet this "naked GUI" approach, as it's been dubbed, would allow one to perform the same call functions using an application like Pidgin, for example.

For consumers, this approach would be a marvelous way to integrate Skype into existing "favorite applications." But there's also a fairly large flaw surrounding this plan. Skype would need to come up with some ingenious way to keep its services relevant, because you can bet that I'd want to add more VoIP services than just Skype to my communications app of choice. I'd love to be able to receive calls on Skype and call other people for free on Skype... but if there was a way to use a cheaper provider for outbound calls, I'd cast off the Skype experience like a boot on the end of a fishing line. Integration with third-party apps is truly a double-edged sword.

Skype as a Service?

Suppose Skype opts for option two and extends its service into third-party applications. What if this is not a measure of altruism, but rather, an investment? With countless developers around the world working on their own programming variants that somehow tie back into the underlying Skype protocol, what's to stop this mass intelligence from coming up with The Next Best Thing for Skype integration?

Both TechCrunch and I envision a future where even the desktop client has been rendered useless by the Skype service. No more installations; no more new applications to load up. Skype would be as permanent and unobtrusive a fixture in your operating system as Windows Contacts. Or, better still, you'd be able to dial up Skype directly through a Web application. As soon as you were to click on a phone number, the call would launch right in your browser of choice.

These are all far-off ideas. Some could even be ideas that only exist in a dreamy haze. Skype is promising a "larger offering" to come centered on its open-source Linux GUI announcement. When that big thing happens, do give me a ring.

AMD: PhysX and Other Proprietary Standards Will Die Off

Paul Lilly — Fri, 12 Dec 2008 11:21:05 -0600

The promise of in-game physics has yet to be fully realized, but the technology doesn't appear to be going anywhere. Leading the charge is Nvidia, who has a vested interest with its acquired PhysX technology. But in an interview with Bit-Tech, Godfrey Cheng, Director of Technical Marketing in AMD's Graphics Product Group, downplayed the proprietary standard.

"There is no plan for closed and proprietary standards like PhysX," said Cheng. "As we have emphasized with our support for OpenCL and DX11, closed and proprietary standards will die."

The comment came in response to questions about EA's and 2K's decision to license Nvidia's PhysX technology across all of their worldwide studios. And while Cheng said he can't comment on competitor's business models, he did say that AMD views "Havok technologies and products to be the leaders in physics simulation," pointing out that game developers share that same view. If true, it would be reasonable to assume EA and 2K have gone against their development studios' wishes by adopting PhysX.

"People need to scrutinize various announcements on what is beling 'licensed,'" Cheng pointed out. "Is it to replace the whole physics simulation / tool stack within a game or within the whole studio? Is it for a specific physics simulation product or just a couple of titles? Remember PhysX also has game physics libraries in addition to its new GPU based products."

Cheng went on to say that Havok physics on Radeon videocards is still forthcoming, possibly by early 2009, but noted that this is just the beginning of in-game physics.

Can Open-Source Software Help a Recession?

David Murphy — Fri, 14 Nov 2008 12:00:00 -0600

If you haven't noticed the general collapse of the financial system around you, coupled with the massive switch to corporate cost-savings mechanisms (including a healthy dose of "rightsizing" by every company under the sun), then you need to stop playing Wrath of the Lich King and flip on the news. Money is important, but perhaps never as important to the general corporate well-being as right now.

It's no surprise then that good ol' open source hardware and software platforms are being thrown into the mix now more than ever. Semantic arguments aside, the open source movement is generally consider a cheaper, if not free alternative to proprietary, commercial software in the enterprise market. But that doesn't mean that open-source software comes without a cost, nor are these companies necessarily immune to the financial movements of the technological industry. So where, then, does open-source development rest in the spectrum? Can these solutions do enough to save the bottom lines of big business? Or are open-source companies just as doomed by a market slowdown as the software vendors on the other side of the fence?

The Argument Against

Here's the deal: Open Source isn't freeware. A business might be able to get its hands on a program's source code (or a non-licensed variant thereof), but that doesn't mean that it will be able to have a brand new enterprise server up-and-running at the touch of a button. Open Source software developers make their money in a bundle of ways: selling the reference manuals for the software, selling a custom installer for the open-source application, selling open-source and proprietary versions of the software. And those aren't even the big hits. Numerous companies have found success in releasing the software for free, yet charging for services like technical support, books, speaking engagements, consulting, et cetera. In short, the business is built around the software.

With all of the big tech companies taking a hit from the ongoing recession, this means that open-source developers have also been forced to scale back on costs. Sack the workforce, and there go the consultants and support. Sack the developers, and there goes your ability to hit the market with more versions, updates, and add-ons to your product. Sack the managers, and you've just lost the drum-banger for the ship, not to mention important bits of business like marketing programs, financial accountability, and other important MBA functions. Open source companies are still technology companies, and seeing as they're not taking swims in the ol' coin vault out back like, say, Microsoft, economic downturns have the capacity to hurt them as well as any other company. Nobody is immune from the dangers of this recession.

The Argument For

This example spells it out perfectly: if your average Fortune ### company can save $68,000 by switching from in-house computer systems and Microsoft software to a Google data center setup that uses Google Apps, that is a compelling cost-savings. Factor in the economy of scale and a business could make the cost-cutting measures it needs--while retaining similar functionality as its previous hardware and software--by switching to open-source solutions.

The switch is already occurring. Red Hat has seen a 29 percent increase in revenues this past quarter compared to the same time period last year. And while the growth might not be as astonishing for every open-source company, there's still an element of hand-over-fist growth. Unlike their dot-com counterparts, open-source companies seem more prepared to weather the financial storm as well. There's less over-hiring, overfunding, and over-exuberance-- perhaps indicated by the less-than-crazy returns we previously noted. It's analogous to spreading one's stock picks over a wide spectrum of companies versus dumping your retirement fund into, say, Google: the returns might not be huge, but the losses won't crush you either.

There will always be a "service" component to open source. But the one-time learning curve require to switch hardware or software services feels like a drop in the bucket compared to, say, investing in expensive new servers versus turning to a cloud-computing solution. But we caution that these savings can be a panacea in their own right. Switching a business to a new open-source platform isn't going to be the new captain that rights the ship. At the end of the day, building out (or maintaining) a strong customer base is going to keep a company afloat during these unsteady times. But the cheaper investment opportunities allotted by open source might very well go a long way: to keeping business afloat in the recession, and keeping open source developers rich with new business.

The Final Word

Expect to see slow, steady gains in the open source market. But the recession will do its part in taking out the stragglers and those on the cusp of pushing their business into the black. Consider it a tightening of the open-source belt. The general swath of companies will survive with less damage than their proprietary counterparts in the technology marketplace, but they will not be immune to the harsh financial times. And if things get too bad, expect the more prominent failures of big business and startup companies to take a big chunk out of the open-source developers' client lists--weeding the field thinner yet.

We Can Rebuild It

Gordon Mah Ung — Tue, 27 May 2008 12:24:59 -0500

So your store-bought PC is getting a little long in the tooth and its performance is showing signs of age. You might be tempted to just toss the machine and replace it with a newer model, especially when you see that more powerful OEM PCs can be had for as little as $500. But before you get on the horn to Dell, ask yourself this: What would a power user do (short of having built their own rig to begin with, naturally)? The answer: upgrade.

In many instances, you can achieve even greater performance gains with a $500 upgrade than you can from a new $500 machine. Even Dell’s proprietary builds can be retrofitted into better shape.

The skill is in knowing when it’s right to upgrade and how much upgrading is warranted. Too often, what begins as a simple upgrade can end up as a box full of regrets. Instead of achieving blistering performance, people often find that they’ve just thrown good money after bad hardware.

This brings us to the first rule of upgrading: Know your needs. Are you after higher frame rates and the ability to game at higher resolutions? You’ll need a new graphics card. Are you tired of waiting hours for your video editing to finish or the eternity it takes to edit your photos? A new CPU is in order.

After you’ve determined your goals, set a budget. Can you spend $200 or $2,000? Finally, the hardest question will be whether it makes sense to even perform an upgrade. This is the part that usually trips us all up. Folks are often compelled to upgrade old machines out of loyalty, as though that box of silicon, tin, and plastic was the starship Enterprise. The truth is that it’s just a bunch of commodity parts that you probably can’t sell on eBay for a quarter of what you originally paid.

So ask yourself, does it make sense to spend $250 on a 3.4GHz Socket 478 Pentium 4 for that old 2.6GHz P4 box? Do you really want to buy a $200 AGP card for your Athlon XP system? We’re not so sure.

Even worse, like plumbing and car repairs, oftentimes upgrades can cause you to replace more parts than you originally intended. Say you buy a hot, new AGP card for your Athlon XP 3000+ box. You’ll soon discover that the PSU in your vintage machine is underpowered. New PSU. Cha-ching. But now that swanky PSU doesn’t fit in your case. New case. Cha-ching. In the end, you just spent $500 for minimal performance gains.

Over the following pages, we’ll walk you through three real-world upgrade scenarios using three very different old off-the-shelf computers, so you can see first-hand our approach to bringing these rigs up to modern standards. Your upgrade needs won’t necessarily be the same, but by understanding how we made our decisions concerning what to upgrade, you’ll be more prepared to tackle the task yourself.

Next: Our first case study - the Dell P4

Case Study 1: Dell P4

This is the very bottom of what we'd consider upgrading

When it was released almost three years ago, Dell’s Precision 370 (a business-class workstation version of the Dimension) was state of the art with its PCI Express and DDR2 RAM. Today, the box is an ancient hunk of junk that wouldn’t fetch $150 on Craigslist.

The Precision 370 features a 3.8GHz Pentium 4 proc, 1GB of DDR2 ECC, a FireGL V3100 graphics card, an 80GB SATA hard drive, and a combo DVD burner. Our initial plan was to replace the single-core Pentium 4 with a Core 2 or even a Pentium D, but the board uses an Intel 925X chipset, which doesn’t support any dual-core procs. There was one processor upgrade option left, however: The 3.8GHz Pentium 4 in the Precision is a 1MB L2 Prescott. Intel later released a Pentium 4 670, which was pretty much a 3.8GHz P4 but with 2MB of L2 cache.

Another seemingly easy upgrade is also complicated by the machine’s workstation roots. The Precision 370 uses ECC DDR2 DIMMs. That RAM is more difficult to find and typically carries a slight price premium.

Given all of this motherboard’s negatives, we thought about replacing it outright, but this is a Dell, and like many large OEM rigs, this machine sports a proprietary design. There’s no way in hell you’re going to drop a standard ATX motherboard into this case.

The weakest link in the Precision is the FireGL V3100 graphics card. This 128MB frame-buffer card with a four-pixel pipeline is just about unusable. The card’s best attribute is that it’s PCI Express. Overall, this machine is borderline recycle-bin material, but the support for PCI Express and DDR2 convinced us to give it the old upgrade try.

In its current configuration, it is absolutely useless for gaming and is a mediocre machine for video and photo editing. A new $500 box (see sidebar below) beats it up and down. But let’s see what kind of performance improvements our upgrades can muster.

Sidebar: To Upgrade or Start Fresh?

Before we started, we established a minimum basis for comparison. After all, if you spend $500 on an upgrade, will you really be getting better performance than what a new budget PC with a warranty offers? To establish our low bar, we looked at the $500 configs of several large OEMs and built a comparable baseline machine to see how well our upgrade boxes would perform against a brand-new, inexpensive PC. Turns out, you can get a surprisingly good machine for five Bennies—we’re talking a dual-core 1.8GHz Pentium E2160, with 2GB of DDR2, a 250GB hard drive, an Intel Q33-chipset motherboard, and even an 8500 GT–class card. That’s good eats for the price and a good platform for future upgrades. Later on, you can go quad core with either a 65nm or 45nm Intel CPU and drop in a faster GPU.

This baseline config should serve as a guide for deciding whether an upgrade is worthwhile, and if so, how much is reasonable to spend. If you don’t think your upgraded machine can outperform this configuration, it’s time to start fresh.

Next: Our Dell Upgrades

Our Dell Upgrades

Total Upgrade Cost: $369

We turned this turkey into a usable machine that would give us moderate gaming performance and handle most medium-intensity tasks. We concentrated our upgrades on graphics, RAM, and a drop-in hard drive for additional storage capabilities.

A $160 GeForce 9600 GT is several times faster than the ancient ATI FireGL V3100 that came with the Dell.

GPU

Since the 3.8GHz P4 CPU is still serviceable (albeit barely), our first upgrade target was the substandard FireGL V3100 graphics card. We looked at two contenders: Nvidia’s GeForce 9600 GT and the GeForce 8800 GT. Both feature 512MB frame buffers, but the 9600 GT sports the newer G94 core. The 8800 GT is faster at higher resolutions, but given the potential of a single-core Pentium 4, we felt it was more prudent to save our ducats for now. Still, putting in a $160 card takes us from a machine that’s worthless at gaming to one that can play most of today’s popular titles at normal resolutions. Even better, GPUs are in flux now. AMD and Nvidia are in a GPU price war, so expect the 9600 GT to be even cheaper by the time you read this.
Before the GPU upgrade, the Precision was about 75 percent slower than the $500 box in our Unreal Tournament 3 benchmark. After the upgrade, the Precision was 158 percent faster.

Instead of hunting down ECC RAM, we replaced the rig’s pair of 512MB DIMMS with a pair of 1GB DIMMs. Don’t mix ’n’ match ECC and non-ECC though.

RAM

With RAM as cheap as it is these days, it’s an obvious upgrade choice. It’s so cheap, in fact, we decided to ditch the two 512MB ECC DIMMs in the Precision and replace them with a pair of standard 1GB Corsair DDR2/667 modules for $40. Some people believe that ECC RAM slightly hinders performance, but we were more motivated by the performance benefits we’d gain by moving the machine from its 1GB of RAM to 2GB—the optimal amount for a 32-bit OS.

CPU

A $169 Seagate 750GB drive makes good upgrade sense, as it can be easily transplanted into a new machine.

As we mentioned earlier, the system’s 925X chipset prevented us from using dual-core processors, to say nothing of quads. The one option available is the 2MB L2 version of the Pentium 4 Prescott that Intel released near the end of that CPU’s run. The P4 670 can still be had if you search the nooks and crannies of the Internet, but once you find one, be prepared to shell out some bucks for it. The two sites we found that stocked the processor were charging $199 for it. Even the secondary market—that’s fancy speak for “used”—wanted $100 for a 3.8 P4. That just ain’t worth it. Doubling the L2 cache gets you, what, maybe a few percentage points of improvement in a few applications? It’s not worth the money or the hassle. The highest clocked single-core Pentium 4 is still a dog, but not dead meat on a stick, so we’re sticking with it. Our ProShow Producer test pegs the P4 at about 30 percent slower than the $500 rig’s 1.8GHz dual-core Pentium, while PCMark05 puts the P4 slightly ahead.

Hard Drive

You know things are bad when your iPod has more storage capacity than your computer. The 80GB drive in the Dell isn’t even big enough to be puny. To supplement our storage, we dropped in a $169 Seagate 750GB Barracuda drive for secondary storage. Hard drives (and to some extent optical drives) usually break the upgrading rules on spending because they’re easily transportable. Even if we paid $500 for two 1TB drives, we could easily move those to the next machine that we build.

BENCHMARKS
	$500 PC	Pre-Upgrade	Post-Upgrade	% Change
ProShow (sec)	2,528	3,606	3,606	0%
PCMark05 Overall	4,785	3,460	4,029	16%
PCMark05 CPU	4,635	4,805	4,806	0%
PCMark05 RAM	3,966	4,400	4,425	1%
PCMark05 GPU	3,750	1,653	10,390	529%
PCMark05 HDD	5,877	4,497	4,576	2%
U3 Omicron_Bot (fps)	18	5	46	838%

Next: CyberPower Core 2 Duo

Case Study 2: CyberPower Core 2 Duo

Can a PC have a midlife crisis? This one did. Watch it go from pudgy and out of shape to pimpin’!

Our CyberPower box is easily the most capable machine in this upgrade story. A circa 2006 rig, the CyberPower’s configuration is still quite powerful. At the heart of it is an Intel Core 2 Duo E6700. That’s a 2.66GHz dual core on a 1,066MHz front-side bus. The chip is riding in an Intel D975XBX Bad Axe board with 2GB of DDR2 and a 250GB hard drive for storage. In the GPU department, an ATI Radeon X1950 XTX runs the show.

But while these components are still quite serviceable, they’re far from what most power machines are running today. In fact, we’d say everything in the CyberPower would qualify as mediocre by present-day standards. The Intel 975X board, though, somewhat limits our upgrade options. Built for the Pentium D processor, the 975X— at least the revision here—works with most 65nm Core 2 CPUs but not the newer 45nm CPUs. The board also lacks official support for 1,333MHz front-side bus CPUs, but it will unofficially recognize and work with the Core 2 Extreme QX6850 CPU and other 1,333FSB 65nm parts.

Clearly, the weakest component here is the machine’s hard drive, which is sorely lacking in capacity. After that, it’s really a toss-up as to whether the Core 2 Duo E6700 or the Radeon X1950 XTX is more in need of replacement. In gaming, the X1950 XTX is long in the tooth but still quite capable of playing most modern games at 1280x1024 resolution. Push the resolution to fill a 24-inch or 30-inch flat panel, however, and the card runs out of steam with more punishing titles such as Crysis. The Core 2 Duo E6700 is one of the original Conroe chips launched almost two years ago. It has since been eclipsed by quad-core Conroes (and, of course, Penryn-based chips). Nonetheless, upgrading either of these components will give the machine a reasonable peppy boost in performance.

Next: Our CyberPower Upgrades

Our CyberPower Upgrades

Total Upgrade Cost: $920

As we’ve said, the upgrades here are borderline optional, as this machine is still very useful. But presented with an unexpected windfall of cash or permission from the family CFO, here’s how we’d bring this already capable machine up to modern specifications.

GPU

At $350, the EVGA GeForce 9800 GTX is a hell of a deal and outperforms the original X1950 XTX card.

The Radeon X1950 XTX was a competitive card in its day, with full HDCP support (albeit single-link only). Unfortunately, GPUs age far faster than most components in your PC. So our two-year-old GPU is akin to a three-year-old CPU or a four-year-old optical drive.

We thought about adding a second Radeon X1950 XTX to the machine but got cold sweats at the thought of trying to use the original permutation of CrossFire. Instead of the internal bridge that’s used in SLI and modern CrossFire setups, the old system uses a clumsy, unreliable dongle cable. Worse yet, we couldn’t just grab any two X1950 cards, one of the cards had to be an X1950 XTX “master card.” Well good luck, buddy. We looked around and couldn’t find any reputable establishments selling master cards. In the used market, people wanted $250 for them, so we opted to just remove the X1950 XTX and replace it with today’s top of the line.

Well, almost top of the line. For $350—or $100 more than what you’d have paid for a used X1950 XTX master card—we were able to get Nvidia’s new GeForce 9800 GTX card. The card supports dual-link HDCP, uses the new G92 core, and, fortunately, does not require the newer 8-pin/6-pin power connectors. That means we can continue to use the 600-watt Thermaltake in the machine. However, we encountered another unexpected twist. The longer 9800 GTX board just barely fit into the midtower CyberPower case. If it had been even a few millimeters tighter, we would have had to buy a whole new case too. So before you upgrade your GPU, make sure your case can accommodate it.

We achieved performance gains of 92 percent in PCMark GPU and 72 percent in UT3, our two GPU-centric benchmarks. Not too shabby. And that’s at normal resolutions. At 1920x1200 or 2560x1600, the 9800 GTX eats the X1950 XTX’s lunch, dinner, and midnight snack.

CPU

Call us fools, but we still like good audio in games, which you get in spades from Creative Labs’s X-Fi XtremeGamer.

The D975XBX motherboard isn’t actually certified for quad-core support, but it runs fine with a Conroe quad. We reached for Intel’s 2.4GHz Core 2 Quad Q6600 as a fairly inexpensive way to gain more performance. While tempting, making the jump to a $1,000 quad Core 2 Extreme seems, well, extreme for this box. Selling for $240 today and possibly even less by the time you read this, the Q6600 gives you the performance benefit of four cores. And in apps that scale with cores, such as most video encoders, a quad pays off. With something as simple as slide-show creation, we were able to cut a third of the time off the project’s encode. If a job typically takes three hours, what would you pay to have an extra hour for playing games instead of twiddling your thumbs in front of your dual-core box?

Hard Drive and Soundcard

Don’t believe you can use 1TB of hard drive space? Someone once said the same about 5MB of storage, buddy.

A 250GB drive doesn’t cut the mustard anymore, so we dropped in the biggest, fastest drive around: Samsung’s F1 terabyte drive. It doesn’t hurt that the drive sells for as little as $250. Our second drop-in was a Creative Labs X-Fi XtremeGamer card. An $80 upgrade, this greatly improves the gaming audio experience on our machine.

BENCHMARKS
	$500 P	Pre-Upgrade	Post-Upgrade	% Change
ProShow (sec)	2,528	1,594	1,176	36%
PCMark05 Overall	4,785	7,045	8,110	15%
PCMark05 CPU	4,635	6,819	7,734	13%
PCMark05 RAM	3,966	5,477	5,313	-3%
PCMark05 GPU	3,750	7,668	14,691	92%
PCMark05 HDD	5,877	5,280	5,228	-1%
U3 Omicron_Bot (fps)	18	61	105	72%

Upgrade Scenarios that Lack Clear Answers

Lawyers like to say that hard cases make bad laws. The same logic applies to system upgrading—borderline configurations really challenge you to take a long, hard look at what to do with your machine. In two scenarios we’re really torn on whether or not to upgrade.

AGP: As a general rule, we feel that AGP isn’t just dead, it’s dead, buried, and has turned into fossil fuel. However, there are situations in which adding a $150 card will give a Pentium 4 or Athlon 64 AGP machine a short reprieve. Nvidia has pretty much called it quits on AGP, but AMD is planning on launching a new Radeon HD3850 with the old-school interface. If it’s priced right, the card should buy AGP Pentium 4 users and Athlon 64 machines another year of life. Still, you’ll have to really ponder whether this is a wise investment, as even a bottom-end Pentium dual-core system will spank either CPU.

SOCKET 478: The same upgrading trepidation applies to Socket 478. Beyond dropping in a little more RAM, you’re left with the prospect of chasing down an elusive Gallatin-core P4 Extreme Edition and then paying $400 for it, which seems unwise to us.

Next: The Alienware Athlon 64 3000+

Case Study 3: Alienware Athlon 64 3000+

Our upgrades put the spookiness back into this old Area 51 machine

We’re pretty sure that three years ago somebody uncrated this Alienware Area 51 PC and was amazed by the rig’s unearthly technology and out-of-this-world gaming performance. Today, it’s about as special as a bucket of mud.

That’s a sad comment on how far the Athlon 64 3000+ inside the Area 51 has fallen. Today, this Venice single-core Athlon 64 is probably best suited for web browsing and email. Don’t believe us? In our ProShow Producer benchmark, the lowly Pentium 4 created our photo slide show in about an hour. It took the A64 3000+ almost two hours!

The rig’s GPU, a GeForce 7900 GTX, is not the original videocard that came with the system, as the machine’s other parts are vintage 2005 and the 7900 GTX came out the following year.

The Area 51 is an interesting upgrade challenge. It’s the opposite of the Dell, which had a fairly serviceable CPU but a complete dog of a GPU. The Alienware has an almost useful GPU but a total dog of a CPU. The benchmark scores back this up. Against the new $500 rig, the poor Athlon 64 doesn’t stand a chance. The Alienware PC is able to redeem itself only in the gaming tests, in which even the ancient 7900 GTX outstrips the budget 8500 GT.

So here’s where it gets tricky. Our first instinct was to drop in a Socket 939 dual-core processor and a second GeForce 7900 GTX card. That gives you dual cores and SLI, man! But is that the right upgrade? Read on.

Next: Our Alienware Upgrades

Our Alienware Upgrades

Total Upgrade Cost: $1,197

Remember our third rule of upgrading: Does it make sense? With the Alienware Area 51, it does and it doesn’t.

CPU and Motherboard

It didn’t make fiscal sense to keep the Asus Socket 939 board, so we swapped it for an MSI nForce 750i SLI board.

Upgrading the CPU was job one on the Alienware box. At first we figured, heck, let’s just drop in a Socket 939 Athlon 64 X2 4800+ and call it a day. Then we looked at prices. On Pricewatch.com, the cheapest new 4800+ was $439. Even on eBay, 4800+ procs were moving for about $100. That’s for a used processor. Our second option was that little darling the Opteron 185. A 2.6GHz dual-core 939 CPU, this chip works perfectly fine as a stand-in Socket 939 Athlon 64 dual core. Prices for this chip, however, are closer to $280. If we got a paper route and saved our money, we’d be better off buying a $500 bargain PC instead of buying this proc. Even with the Opteron 185 or 4800+, the box would still be a chump next to the Pentium E2160 in the $500 machine.

That tipped the scales for us. The situation clearly called for a more thorough overhaul. Thanks to Alienware’s use of industry-standard ATX parts, an extreme makeover was not only possible but quite easy. We replaced the Athlon 64 3000+/Asus A8N32-SLI combination for a Core 2

Corsair Dominator modules are pricey but give us low latency and full EPP support for our nForce board.

Quad Q9300 and MSI P7N-SLI Platinum. Intel’s $300 Q9300 is based on the new Penryn core and is the class leader of cheap CPUs. The $175 MSI P7N-SLI Platinum mobo uses nVidia’s nForce 750i chipset and gives us the SLI option, as well. In a nutshell, we’re talking budget SLI with full quad-core Penryn support. We, of course, recycled the 2GB of DDR in the box, but it’s not like DDR2 is expensive today. In fact, it’s even cheaper than DDR.

Thanks to our CPU/mobo upgrade, we went from a slide-show encode that took almost two hours to one that took 21 minutes. That’s a performance increase of more than 400 percent.

GPU

Nvidia’s new GeForce 9800 GTX is the top single-core card in town today.

This one’s easy, right? Just drop the original GeForce 7900 GTX into the upgraded Alienware, buy a second 7900 GTX, and SLI the two, right? Wrong. First, almost no one sells the GeForce 7900 GTX anymore and those who have them want beaucoup bucks. Newegg, for example, had an open-box MSI GeForce 7900 GTX card available—for $300. On eBay, prices for the cards ranged from $300 to $350. Holy GPU, Batman! Computer components are supposed to get cheaper as time goes by, aren’t they? Not when you’re talking about high-end parts, apparently. Still, isn’t it better to pay for a second card rather than buy a single newer part? Not necessarily. Performance might be close in some situations, but generally, a single top-end modern card will outperform cards of an earlier generation, even if they run in SLI mode. What’s more, SLI 7900 GTX still doesn’t give you DX10. And from what we can tell, the drivers aren’t a priority either.

Thus, for this upgrade, we decided to remove the 7900 GTX card and replace it with a brand-spanking-new GeForce 9800 GTX. While the 9800 GTX costs $350, we could probably sell our used 7900 GTX for $250 and end up paying just $100 for the GPU upgrade.

The switch certainly pays off. We went from 38 frames per second in Unreal Tournament 3 to 101fps, or the equivalent of a 166 percent increase in frame rates. That’s at a standard 1280x1024 resolution, too. If we cranked the res up to 1920x1200, the gap between the 7900 GTX and 9800 GTX would be far wider, as the 7900 GTX would undoubtedly run out of steam while the 9800 GTX would keep sailing along.

Hard Drive and Soundcard

As with the CyberPower, we performed a couple upgrades on this rig mostly for livability. We added an F1 1TB Samsung hard drive to the machine as well as a Sound Blaster X-Fi XtremeGamer soundcard and called it a day.

BENCHMARKS
	$500 PC	Pre-Upgrade	Post-Upgrade	% Change
ProShow (sec)	2,528	6,745	1,297	420%
PCMark05 Overall	4,785	4,235	8,786	107%
PCMark05 CPU	4,635	2,643	8,060	205%
PCMark05 RAM	3,966	3,591	5,684	58%
PCMark05 GPU	3,750	7,731	14,830	92%
PCMark05 HDD	5,877	8,244	8,250	0%
U3 Omicron_Bot (fps)	18	38	101	166%

Some Upgrades are Just Plain Stupid

Tempting as it might be to upgrade, it’s not always prudent.

Pentium III: It’s dead, Jim. Really. It’s really, really, really dead. You could shoot a P3 onto the Genesis planet and the only possible result would be that you wasted a photon-torpedo casing. Just give up.

Athlon XP: Instead of throwing good money into an Athlon XP box, just send us the cash. At least then somebody would be happy when all is said and done.

Rambus: There’s a surprising number of Direct RDRAM boxes in service. If you have an RDRAM machine, don’t even bother cracking it open. Unless you can afford to spend a million bucks on memory, you’ll never see more than 1GB of RDRAM. RDRAM also limits you to a 533MHz FSB CPU, so there ain’t no way you’ll ever drop in that 3.4GHz Pentium 4 Extreme Edition.

Dell Drop Kicks Proprietary Parts

Gordon Mah Ung — Tue, 26 Feb 2008 13:25:52 -0600

Dell, long dinged for using proprietary hardware in its gaming PCs, has seen the light. The company said such annoying traits such as proprietary motherboards and power supplies is now a thing of the past.

The first XPS to shed the proprietary hardware will be the new budget XPS630 gaming machine. Based on the nForce 650 ichipset, Dell claims you can swap the board, PSU out for any other ATX-spec hardware.

The change is a long overdue. In the past, Dell has used designs that looked ATX-like but were actually not. If you tried to swap the power supply in your Pentium III Dimension XPS B 733R years ago, you would have been greeted by charred motherboard as the company actually wired its PSU’s differently than the industry but did not key them differently. For years, PC Power and Cooling has made small side business selling Dell upgrade PSUs. More recently, the company has been called out over BTX support and even using a proprietary power plug in its more recent XPS gaming rigs.

Why the use of proprietary designs? Cynical observers have said the company was just trying to lock customers into buying parts exclusively from Dell. The company has long defended the practice by saying that the variations from spec were because its engineers found the specs to be lacking. But the heat from critics and machines such as Hewlett-Packard’s Blackbird 002 going all ATX apparently have forced Dell to see the light. Company officials said the mantra in Dell engineering is that varying from the spec’s must be avoided at all costs.

The change from proprietary parts won’t be the only new trick for the XPS630 though. Dell has taken a page from Hewlett-Packard and claims the XPS 630 will support either Nvidia’s dual-card SLI or AMD’s dual-card CrossFire cards. How can Dell do this? Maximum PC spoke to AMD graphics officials who said the capability is being offered only to select OEMs who take the responsibility for making sure the drivers fully work with the BIOSes on the motherboards.

So why not just release such a driver to the public to let any nForce-user run CrossFire? The company said it is worried that a certain company could affect the performance of its cards when CrossFire is run on an SLI board so public support just isn’t going to happen. In the case of Dell and HP, AMD feels both have enough influence to keep performance problems from cropping up on their systems. For now, the CrossFire support is only through drivers obtained directly from Dell.
Dell says the XPS 630 will also be the first tier one OEM system to support Nvidia’s Enthusiast System Architecture. In the XPS 630’s case, ESA will let the user control the lights in the system. ESA support for the PSU or other components will not be initially supported.

The $1,300 version of the XPS 630 will ship with a 2.4GHz Core 2 Quad Q6600, GeForce 8800GT and a 750 watt power supply. Dell said the BIOS on the XPS 630 will support overclocking and is upgradeable to both dual and quad-core Intel Penryn CPUs.