Maximum PC in the lab RSS Feed

Everything You Need to Know About GDDR Memory

Michael Brown — Tue, 16 Sep 2008 01:11:36 -0500

We invariably refer to the video memory in modern videocards as GDDR, differentiating it only by version (GDDR2, GDDR3, GDDR4, and now GDDR5), but the technology’s full acronym is actually GDDR SDRAM, which stands for Graphics Double Data Rate Synchronous Dynamic Random Access Memory.

“Double data rate” describes the memory’s capacity for double-pumping data: Transfers occur on both the rising and falling edges of the clock signal. This endows memory clocked at 800MHz with an effective data-transfer rate of 1.6GHz. “Synchronous” refers to the memory’s ability to operate in time with the computer’s system bus. This allows the memory to accept a new instruction without having to wait for a previous instruction to be processed, a practice known as instruction pipelining.

GDDR2 memory was never a very popular solution among GPU manufacturers: The technology required 2.5 volts to power its input buffers and core logic (i.e., VDD voltage), which is the same as GDDR. GDDR2 operated at much higher clock speeds than its predecessor, however, which produced a tremendous amount of heat. The fact that GDDR2’s VDDQ voltage requirement (the electricity needed to power the memory’s output buffers) was only 1.8 volts didn’t compensate for this problem.

Survival of the Fittest

GDDR3—an open standard developed by ATI in conjunction with the standards organization JEDEC Solid State Technology Association—is the most widely used graphics memory technology in use today. Ironically, Nvidia introduced the first graphics processors designed to use GDDR3: The GeForce FX 5700 Ultra, followed by the GeForce 6800 Ultra. ATI didn’t deploy a GDDR3 solution until it shipped the Radeon X800.

GDDR3 improved on previous GDDR designs by supporting higher clock speeds while requiring less power. These chips consume less electricity, so they produce less heat and can rely on simpler cooling hardware (GDDR3’s VDD and VDDQ voltage requirements are both 1.8 volts). GDDR3 also has separate read and write data strobes, which contributes to a much faster read-to-write ratio (meaning the turnaround from a read operation to a write operation occurs much more quickly) than GDDR2 supported. GDDR3 chips have a hardware reset feature that can wipe their memory clean to start receiving new data should such an operation be necessary.

ATI and Nvidia (in conjunction with JDEC) both had a hand in establishing the specification for the next generation of graphics memory, GDDR4, but Nvidia has so far decided not to use the new technology in any of its reference designs. ATI, meanwhile, incorporated the new memory first in its Radeon X1950 XTX cards and subsequently in several models of its Radeon HD 2000, 3000, and 4000 series.

Evolutionary Dead End

GDDR4’s improvements over GDDR3 were mostly incremental. It seemed to offer a power advantage in that it could operate with just 1.5 volts, compared to GDDR3’s 1.8 volts. Board designers, however, quickly discovered that they needed 1.8 volts anyway to ensure stability at higher clock rates.

Two other GDDR4 enhancements are more significant in that they increase the memory’s overall performance: The new memory doubled the size of GDDR3’s prefetch scheme from 4 bits to 8 bits, and its burst length was locked at 8 bits (GDDR3 supports either 4- or 8-bit burst lengths). Prefetch enables the memory chip to anticipate the need for data and grab it before the GPU asks for it, reducing the time the processor has to wait. Burst length defines the amount of data sent in burst mode, a process in which data is transmitted without waiting for input from another device, such as the GPU.

GDDR4’s 8-bit burst length might be one reason Nvidia ultimately passed on this type of memory: Nvidia’s processors support only 4-bit burst lengths. With ATI (now AMD) being the only major customer for GDDR4, just two manufacturers—Samsung and Hynix—decided to manufacture it. This circumstance has kept the price of the memory relatively high.

Successful Mutation?

GDDR5 is the next major development in graphics, and as with GDDR4, AMD’s ATI division has already paired it with its higher-end GPU: the Radeon HD 4870. Nvidia continues to hang back, professing satisfaction with the performance of GDDR3.

GDDR5 requires just 1.5 volts of electrical power, which should make the memory run cooler—a feature that could aid in overclocking, reduce manufacturing costs, and extend battery life if used in a notebook PC. The new memory’s prefetch and burst length remain the same as that of GDDR4: 8 bits on both counts.

GDDR5 technology supports densities ranging from 512Mb to 2Gb, so it would require just four 2Gb modules to create a 1GB frame buffer (here again, however, real-world parts are currently limited to 512Mb and 1Gb). Boasting a raw theoretical data rate ranging from 3.6Gb/s to 6Gb/s (although we won’t see that upper limit for several years), GDDR5 promises to deliver twice the memory bandwidth of GDDR3 running at the same clock frequency.

More practically, that high data rate also enables a GPU manufacturer to achieve nearly the same memory bandwidth with an economical 256-bit interface as it would by building a much more expensive 512-bit bus into its GPU.

Nvidia’s professed ambivalence toward GDDR5 hasn’t stopped a third major memory manufacturer—Qimonda—from joining Hynix and Samsung in the market for GDDR5 memory. Hmm, is anyone taking bets that Nvidia’s next-generation GPU will tap GDDR5?

In the Lab: Will Smith Gets Burned by the Apple TV

Will Smith — Tue, 29 Apr 2008 15:38:06 -0500

As part of my testing for this month’s cover feature, I spent a few quality days watching movies from the iTunes Store on my PC and in my living room. By necessity, I had to integrate a newly updated Apple TV into my entertainment center, which is a fairly common closed cabinet with a few air vents in the back. All of my other electronics devices—my receiver, my TiVo, my Xbox 360—live happily in this environment (although I do open the cabinet door when I fire up the 360).

After leaving the Apple TV connected overnight, I detected the faint smell of burning plastic in my living room. After investigating, I found that the Apple TV was actually too hot to touch—around 140 F. I don’t know about you, but none of the other consumer electronics devices I own (save a towel-wrapped Xbox 360) gets this hot.

After digging into the documentation and Apple’s support site, I was assured that the passively cooled Apple TV “may feel warm when touched.” But that it was “designed to operate quietly in an environment warmer than room temperature, including within media cabinets.” Eventually, I discovered a standby mode, which leaves the Apple TV running but disables the graphics hardware to minimize heating. You can access standby by holding down the Play/Pause button for a few seconds. That’s great, but if I forget to manually put the Apple TV into standby, it won’t automatically do it for me. That’s just irresponsible and dumb. There’s absolutely no reason for a modern CE device that gets this hot not to automatically shut down when it’s not being used.

In the Lab: David Murphy Doesn't Like Microsoft Mucking with Standards

David Murphy — Fri, 04 Apr 2008 16:33:27 -0500

What do you do with a media-streaming device that doesn’t connect to anything? That’s the question posed by the HP Media Vault NAS, which I reviewed this month (page 79). I found that I was unable to stream a movie on the device’s hard drive to any connected platform in my house, including Windows Media Player 11 and the Xbox 360.

But determining the culprit in this media mess wasn’t as cut-and-dried as I expected. I was ready to hoist the sails and speed away from the sinking HP device, but a little bit of research brought a new target to bear: Microsoft. The company has allegedly modified the open UPnP protocol the Xbox uses to connect to devices. Company reps won’t admit to any outright fudging, but something seems suspicious in their response: “The Xbox 360 console is optimized to communicate with Microsoft UPnP streaming devices… to provide the best possible consumer experience with the Xbox 360 dashboard.”

The blog Mediaholic interviewed the CTO of TwonkyVision at the 2006 Consumer Electronics Show—TwonkyVision being the creator of TwonkyMedia, the streaming service the QNAP TS-109 Pro NAS box uses to connect to the Xbox. This works, according to the CTO, because Twonky reverse engineered the protocol to end-run Microsoft’s Windows Media Connect-only restriction.

If that’s correct—and I suspect it is—that’s why I’m hacked off. An open standard is dubbed “open” for an explicit reason: A product that follows the standard should be able to interoperate with other standard-friendly devices. You’re not following the standard if you’re tweaking it to work with only your branded platforms.

HP might share some blame for not figuring out a way to make streaming work, but why should a company have to reinvent a wheel that Microsoft’s broken?

In the Lab: Gordon Mah Ung Re-Examines RealTek Audo

Gordon Mah Ung — Fri, 04 Apr 2008 16:32:24 -0500

In December 2006, I took RealTek to the woodshed for its cheating drivers, which made many EAX-enabled games sound simply awful.

For months, we’ve been rattling motherboard makers, including Nvidia (which uses RealTek parts on many of its reference motherboards), about the issue, but RealTek seemed not to get the message, as nothing changed—until recently, that is. Indeed, I was pretty surprised when testing XFX’s nForce 780i-based motherboard to find that RealTek had finally fixed the problem with EAX support.

The problem was that RealTek HD Audio drivers would tell games that EAX support was enabled, which would cause games to pass off the audio mixing to the soundcard drivers. In RealTek’s case, that often (though not always) produced terrible sound. A tank that was supposed to sound as though it was 200 meters away sounded like it was 10 meters away. And instead of a grenade sounding muffled by the presence of a wall, it sounded like it was next to you.

With older RealTek drivers, a simple position test would show that EAX filtering wasn’t working correctly.

Why would RealTek do this? Falsely claiming to have EAX support is akin to a videocard driver claiming to run16x AA filtering when, in reality, no filtering is occurring.

But RealTek’s R 1.86 drivers dated February 1, 2008 seem to finally enable EAX. Using RightMark’s 3D Sound positioning utility, I found that the occlusion effect finally worked as it does on all other EAX-enabled cards. The obstruction effect almost works but is a bit heavy-handed, blocking all sound when you should still hear some audio.

In real-world gaming tests, however, EAX support seemed sporadic. Battlefield 2 audio was as poor as it was previously when EAX was enabled and the game let the RealTek part do the mixing. But when we fired up the EAX-enabled BioShock and ran the new drivers through their paces, with EAX and reverb enabled, the experience wasn’t bad. Compared to an X-Fi XtremeGamer soundcard, the onboard RealTek component’s audio was still clearly inferior but at least not intolerable, as it was with Battlefield 2.

I’ll continue to follow this issue, but at least it appears that RealTek is finally addressing the problem.

In the Lab: Gordon Mah Ung Wants to Kill ATX

Gordon Mah Ung — Mon, 03 Mar 2008 14:44:01 -0600

Would you use a ball mouse? A VL-Bus graphics card? A Socket 7 board? Then why the hell are enthusiasts still embracing the 13-year-old ATX formfactor? It’s time we started thinking about moving beyond ATX.

Today, we’re running quad-core boxes with two or more GPUs in a formfactor created when people used serial and parallel ports and the Pentium was the hot chip in town. In the near future, USB 3 will appear on motherboards in the south-bridge chips. To route the ports, motherboard vendors must run traces all the way from the south bridge to the rear I/O shield. You might be able to do this with USB 3 data rates on a four-layer board, but can it be done with USB 5? If it requires more layers, it’ll add to the cost of the board.

The failed BTX formfactor included many forward-thinking features.

Of course, Intel tried to fix these problems with its BTX formfactor, which cratered because of resistance from case-makers, a new emphasis on cooler CPUs, and complete resistance from AMD. But if I were hardware dictator for a day, I’d propose a new formfactor called GTX (Gordon TX) that mandates:

A minimum motherboard stand-off height, so wires can be routed safely and easily under the motherboard
RAM and expansion slots that are parallel to air flow in the case
A larger board area and I/O section to accommodate the dizzy- ing array of connectors a modern power user needs
Standardized front-panel connectors for reset, power, and LEDs
Two more expansion slots. The seven in ATX aren't enough with the multi-GPU machines we’re building
Less distance between the south-bridge and north-bridge chips
A CPU cooling scheme that accounts for liquid cooling or venting from an area other than the front of the machine. While we’re at it, let’s build in more cooling for the GPU.

This probably sounds crazy because the push is for smaller, rather than larger, PCs, but I say it’s time. Average people are moving toward smaller machines or notebook PCs. Full-tower ATX boxes are increasingly focused on the workstation market; we really shouldn’t be handcuffed by formfactors designed to appeal to the masses.

In the Lab: Gordon Mah Ung Thinks ESA Is Long Overdue

Gordon Mah Ung — Thu, 07 Feb 2008 14:42:36 -0600

PC geeks like dials and gauges and greater hands-on involvement, so Nvidia’s Enthusiast System Architecture (ESA) should come as a welcome gift to us all. In a nutshell, ESA is an open standard for adding communication capabilities to normally “dumb” components. Using USB as the basic protocol, ESA adds intelligence to devices such as water coolers, power supplies, and cases, so you’ll have more insight into how your PC is running and be able to effect changes.

I finally kicked the tires on ESA’s basic functionality with this month’s Hypersonic Sonic Boom OCX (reviewed here). The PC came with an ESA-enabled PC Power and Cooling 1,200 watt PSU and CoolIT Systems Freezone Elite. At this stage, the software, drivers, and hardware are pretty rudimentary, but I like what I saw.

Of course, your own appreciation of ESA depends on how geeky you are, but I like knowing that my PSU is eating about 34 amps on the 12-volt rail at idle. And just how hot is it inside my power supply? Well, it’s 31 C, and the fan is spinning at 100 percent. Likewise, just what is the temperature of the coolant? You’ll know if you have an ESA-enabled water-cooling system. Sure, some water-cooling rigs already give you basic readouts and manual control over flow, but ESA will eventually let you control functionality from the OS or monitor a device remotely across the Internet.

ESA will finally let you know the actual load on your PSU!

Ultimately, if ESA takes hold, we could use it to troubleshoot problems—we’d know if the power supply is overloaded or if the coolant is low. Cases with ESA will let you toggle lights and control fans from within the OS, functionality which has been limited to large OEMs until now, and I’m sure some crafty geek will figure out a way to let you remotely control a USB device via ESA, so your Peltier cup-cooler has your beer nice and cold by the time you get home from the cubicle.

ESA actually stands a better chance of widespread adoption than Nvidia’s EPP profiles for RAM. While EPP couldn’t make it past the August JEDEC memory council, ESA will be submitted to the USB-IF—the folks who approve USB standards. Nvidia is even willing to forgo branding the spec with its name to ensure its competitors feel more comfortable adopting it.

Overall, ESA is cool and will give power users yet another reason to upgrade.

In the Lab: Nathan Edwards Flies the Friendly Skies

Nathan Edwards — Thu, 07 Feb 2008 14:42:17 -0600

In my review of the Hypersonic Sonic Boom OCX on page 68, I focused on the PC’s performance and stability—crucial qualities in any gaming rig. But I’d be doing this machine a disservice if I didn’t talk a bit about its flight-simulator setup.

Hypersonic shipped the Sonic Boom with three 19-inch monitors controlled by a Matrox TripleHead2Go Digital Edition—enabling a resolution of 3840x1024 across a single desktop—as well as Saitek’s Pro Flight yoke and rudder pedals. The good news: Flight Simulator X looks fantastic with this panoramic view. But is it worth it? Are three relatively small monitors better than one huge one? For comparison, I also tested the Hypersonic with one 30-inch Gateway XHD3000 at 2560x1600.

The verdict? Although I and others here thought we’d prefer a large single-panel display, when it came to the flight sim, I actually preferred the smaller panels’ wraparound effect, which felt more lifelike. Once I returned to the desktop for real work, however, the happy feelings vanished. The triple monitors are especially irksome when you have to go into the BIOS. Instead of being confined to one panel, the BIOS screen is stretched across all three, which makes changing settings rather difficult. It was hard to tell if I was adjusting the CPU core voltage or RAM voltage with the TripleHead2Go enabled.

Given the choice, I’d go for the triple-panel display for racing, flying, or any other immersive sim—maybe even an MMO. But for most other purposes, I’ll take a single large panel any day.

In the Lab: David Murphy Explains Case Reviews

David Murphy — Tue, 05 Feb 2008 17:10:03 -0600

The art of testing cases at Maximum PC is a lot like the sword ceremony scene in Kill Bill. There’s a lot of razzmatazz and showmanship at first, but in the end, a worthy case is treated with honor and delicacy as it’s gently placed back in the Lab; wretched cases are also moved to the Lab… to be used as pedestals upon which we rest the worthy enclosures.

I’m being only a bit facetious when I say that because reviewing a case really does require delicacy. I start by giving a case’s exterior a full inspection. Aesthetics play the smallest role in our overall verdict, but as our September 2007 review of Dynapower USA’s Hachiman case illustrates, I do point out the look of a case when a vendor tries something new—or when an exterior is worthy of contempt (although I do recognize that one person’s pile of garbage is another person’s Pieta).

The real fun begins once I’ve popped off the side panel—and if that’s not a pleasant experience, the chassis earns its first ding. Inside, I’m looking for elements such as screwless mounts that are easy to use but secure, convenient and accessible drive bays, and cooling potential. Mounting a motherboard in the case typically exposes any flaws in the overall design: We experienced this with an early version of Antec’s Nine Hundred case, which didn’t allow certain cables to be connected. (The problem was fixed prelaunch.)

I like to come at case reviews as if I’m a basic user, and in doing so, I ask myself a series of questions during the review process: Is the case difficult to work in? Is this case easy to manipulate and fill with components? Is there a better way to do what I’m doing? Do I need extra parts, tools, or products to complete my rig? And once the rig is built, are the provided cooling solutions too loud? Does the case adequately muffle my noisy components? Is its cooling sufficient? Does the case lack anything that would be necessary for me to build the perfect rig?

There are two primary case design styles: classic and over the top and gamer themed. Depending on your needs, even a 10 Kick Ass case might not be exactly what you’re looking for.

A case can have a lot of features and still have a horrible design. Conversely, a case can be a little skimpy with its add-ons—holes for water-cooling tubes and included tri-speed fans and LCD display panels—but still deliver an amazing experience for rig builders. Reviewing cases is part exact science, part surprise and delight.