Maximum PC amd RSS Feed

Stoked about AMD's HD 5970? Good Luck Finding One in Stock

Paul Lilly — Fri, 20 Nov 2009 11:15:49 -0600

AMD's newest HD 5970 is the fastest videocard on the planet, and as it turns out, it's also the hardest to find. And not just in the U.S. either, the card has been equally elusive in the EU.

According to news and rumor site Fudzilla, several AMD launch partners confirmed there would be limited availability of the card, and that's been the case so far. Newegg showed some stock in the wee morning hours on November 18th, but by the time noon (PST) rolled around, stock was gone. The $625 price tag doesn't appear to be slowing demand.

Fudzilla says it's been told to expect to see stock of the XFX Radeon HD 5970 and 5970 Black Edition sometime today, but it's unclear if that will include the U.S. market, or be limited to Europe.

Seen the HD 5970 in stock somewhere? Post a link in the comments section below.

First Programmable Quantum Computer Created

Ryan Whitwam — Wed, 18 Nov 2009 19:27:22 -0600

Moore’s Law states that approximately every two years, the number of transistors that can be placed on an integrated circuit doubles. This has held true for the last 50 years. But there will come a point one day when physics puts a stop to that. Eventually the boundaries of atomic scale will limit transistor density. However, a new breakthrough in the field of quantum computing may provide hope for future advances. Until now, a quantum computing device had to be designed for one, and only one, operation. But scientists from the National Institute of Standards and Technology (NIST) have constructed the first programmable quantum processor.

Quantum processing units are fundamentally different in a number of ways. First, where a regular bit can be only 1 or 0, a quantum bit (or qubit) only assumes a value of 1 or 0 when it is observed. Additionally, Quantum computers aren’t bound by Boolean operators like ‘and, ‘or’ and ‘not’. Finally, two qubits can be “entangled”, meaning they will always have the same value when observed, even if separated.

The NIST computer consists of two quantum gates, one single qubit gate and an entangled two qubit gate. The gates utilized two beryllium ions stimulated with UV lasers to represent operations. The test programs run came back with 79% correct results. Certainly not perfect, but a huge step forward. You won’t be dropping one of these into a socket on your motherboard anytime soon, but maybe someday.

ATI Catalyst 9.11 Now Available for Download

Paul Lilly — Wed, 18 Nov 2009 11:30:49 -0600

ATI on Tuesday released its Catalyst software suite, version 9.11, for Windows XP, Vista, and Windows 7. The latest release appears to put a heavy focus on squashing bugs as opposed to injecting performance enhancements to specific gaming titles.

New features include GPU acceleration of H.264 video content using Adobe Flash Player 10.1 beta, and high quality downscaling for video transcoding MSE. Everything else in the release notes is geared towards resolving issues for various OSes. Some of these include:

Users can now enable and disable CrossFire when three displays are configured in extended mode (all OSes)
Catalyst Control Center no longer stops responding when setting Eyefinity SLS mode for extended HDMI display (all OSes)
Resolved an issue where high bit rate audio from Blu-ray discs might not output when using PowerDVD 10 (Windows 7)
Wolfenstein no longer stops responding when running a 2560x1600 resolution (Windows 7)
No more green lines at the bottom of the screen when playing some interlaced content (Vista)
PowerDVD no longer intermittently terminates when playing HD MPEG2 content in extended/clone mode (Vista)
Transcoding of AC3 files no longer shows corruption in transcoded files (XP)

There are plenty of other bug fixes, as well as some lingering known issues, all of which you can view in the release notes here (PDF).

Catalyst Download

Image Credit: AMD

ATI Radeon HD 5970: The Undisputed Performance Champ

Loyd Case & Gordon Mah Ung — Wed, 18 Nov 2009 11:00:00 -0600

AMD’s Radeon HD 5970 takes the performance crown, and now offers DirectX 11 GPUs top to bottom.

You can forgive AMD for stealing a line from Nvidia’s playbook. From the name and marketing materials, it’s not obvious that this card is a dual GPU card. One AMD chart even refers to the card as the “ATI Radeon HD 5970 GPU,” much like Nvidia’s 295 GTX is a dual GPU card that’s sold as if it were a normal graphics card.

Let’s take a quick look at the speeds and feeds of the new card, and then discuss additional features. We’ll compare them to the Radeon HD 5870 single GPU card; there are differences in core and memory clock speeds.

ATI Radeon GPUs Compared

Model	HD 4870	HD 5970
GPU	Single HD 5870 GPU	Dual HD 5870 GPUs
Stream Processors	1600	3200
Core Clock	850MHz	725MHz
Memory	1GB 1200MHz GDDR5	2GB 1000MHz GDDR5
Texture Units	80	160
ROPs	32	64
Memory Data Rate	4.8Gbps	4.0Gbps
Compute Performance	2.72TFLOPs	4.64TFLOPs
Idle Power	27W	42W
Max Power	188W	294W

Most features simply double, since there are two identical GPUs with the same memory per GPU. However, note the memory bandwidth is lower, since the memory clock is lower. Also, overall compute performance isn’t quite double, since the engine clock for each GPU defaults to 725MHz, rather than the single GPU 5870’s 850MHz.

In addition to the paired GPUs, the card maximizes performance with a new second generation PLX PCI Express 2.1 bridge chip. This speeds up communications between the two GPUs.

The older Radeon HD 4870 X2, AMD’s last dual GPU card, often suffered from serious overheating problems, so AIT has taken steps to improve overall thermal efficiency. The first step is to used specially screened GPU chips with the minimum amount of leakage current. The thermal system consists of a vapor chamber, with heat dissipation via a fully vented exhaust system. Finally, a multipoint, programmable PWM fan controller ensures fan speeds consistent with overall thermal output. The whole affair is rated to dissipate 400W, or over 100W of headroom.

The use of low leakage GPUs suggests that the core clock speeds could be pushed higher than the default 725MHz. Given that the card can dissipate 400W of overall thermal output, ATI has decided to leave the card unlocked, letting users overclock it to their heart’s content. The company is even making available an overvoltage tool, and suggests that the card has enough headroom to push the engine clock to above the 850MHz of the single GPU HD 5870 and the memory speeds up to the 1200MHz of the 5870. In The HD 5970 also sports sophisticated digital voltage regulators, real time power monitoring and Japanese pure ceramic capacitors, which facilitate higher clock speeds. (The reason for the more conservative default clock speeds is that the company needs to make allowances for the thermal environment inside PC cases, which may be less optimal.)

What about power efficiency? The HD 5970 offers the same clock gating capability as the single GPU variant. In addition, the card offers an ultra low power state that puts one GPU to sleep, which cuts the overall power draw of that GPU by half the normal idle performance. The card idles at 42W, only 15W higher than the 27W of the HD 5870. Of course, the card sucks 294W at full throttle – and that’s at the default clock speeds. So you’ll need a power supply with the later PCI Express 8-pin power connector – the HD 5970 uses one eight pin and one six pin connector from the PSU.

The onboard connectors are slightly different from the original HD 5870: two DVI-I connectors and one mini-DisplayPort attachment, first popularized with Apple’s MacBook Pro laptops and iMacs. All three can be attached simultaneously, and drive three displays at the same time. Note that most DisplayPort enabled monitors which aren’t Apple displays ship with standard sized DisplayPort cables, so you’ll need to obtain a mini-DisplayPort to standard DisplayPort cable if you want to take advantage of that connection option.

Performance

We tested three graphics cards: the Radeon HD 5970, the Radeon HD 4870 X2 and the GeForce 295 GTX. Note that the 295 GTX only ships with 1792MB of GDDR3 memory (896MB per GPU). The Nvidia GPUs on the 295 GTX have a full complement of 240 shaders, like the 285GTX, but only a 448 bit wide memory bus, similar to the GeForce GTX 260 line of graphics chips.

All games were run at two settings: 1920x1200, 4xAA and 2560x1600 4xAA. All detail settings on all games were maxed out. We also report the results from 3DMark Vantage’sdefault, performance setting and the extreme setting. In addition, we’re tossing in the results from the eVGA GeForce 285 GTX SSC and XFX Radeon HD 5870 at the 1920x1200 resolutions, so you can get an idea as to how much you gain from the dual GPU cards.

We used our standard graphics test system for benchmarking graphics cards:

Test System

Component	Model
CPU	Intel Core i7 975 @ 3.33GHz
Motherboard	Asus Rampage II ExtremeX58
Memory	6GB Corsair Dominator DDR3-1600 @ 1333MHz
Hard Drive	Seagate 7200.12 1TB
PSU	Corsair 850W
Operating System	Windows 7

The HD 5890 seems much better behaved than its predecessor, the HD 4870 X2. Although the HD 5970 fan noise became quite audible at high loads, it seemed much quieter than the 4870 X2. More telling was our experience removing the cards immediately after completed benchmarking runs. The 4870 X2 was extremely hot to the touch – we had to handle it carefully to avoid burnt fingers – while the 5970 was cooler to handle (though still very warm.) Much of this was no doubt due to the vapor chamber covering the back side of the card.

As we expected from our previous experience with the Radeon HD 5870, the dual GPU HD 5970 pretty much smokes the competition. At its targeted $599 price point, the performance needs to be stellar – and it is. Let’s check out the performance charts.

1920X1200 4XAA

Model	eVGA 285 GTX SSC	Radeon HD 4870 x2	GeForce GTX 295	XFX 5870	Radeon HD 5970
3DMark Vantage (Perf)	13941	14458	19342	17089	21089
3DMark Vantage (Extreme)	6276	6574	9241	8312	12084
HAWX	62	78	93	68	104
Far Cry 2 (Action)	47	67	62	62	75
Far Cry 2 (Ranch Long)	56	77	73	74	116
BattleForge (DX10)	46	36	33	47	61
Crysis (DX10)	22	33	29	32	44
Resident Evil 5	87	126	115	100	135
X3: Terran Conflict	93	101	100	101	107
STALKER: Clear Skies	27	38	40	36	53

2560x1600 4XAA

Model	Radeon HD 4870 X2	GeForce GTX 295	Radeon HD 5970
HAWX	53	54	76
Far Cry 2 (Action)	18	59	65
Far Cry 2 (Ranch Long)	50	55	82
BattleForge (DX10)	23	20	39
Crysis (DX10)	19	18	24
Resident Evil 5	88	85	102
X3: Terran Conflict	90	59	101
STALKER: Clear Skies	22	20	31

There’s no question the Radeon HD 5970 is the fastest single desktop graphics card you can buy today. The real question is if it’s worth $200 more than a single Radeon HD 5870. Certainly if you’re running a 1920x1200 display, we’d recommend the single GPU card. But if you’re running triple displays, or a 30-inch, 2560x1600 monitor, and you want to push polygons at full resolution, you should consider the Radeon HD 5970. But make sure you’re willing to pay the cost.

Another note: this card is enormous. It’s an actual foot-long graphics card; make sure you’ve got a case deep enough to handle it.

Then there’s the whole issue of CrossFire. The HD 5970 is essentially a pair of downclocked HD 5870s in CrossFire mode. When the game can take advantage of CrossFire – as it does in our benchmarks – you can see huge performance gains. If the game can’t make use of CrossFire, then you own a pricey single GPU card that doesn’t run as fast as the HD 5870.

But what about running TWO of these cards in CrossFire mode? Find out on the next page.

Dual-5970 CrossFire Performance

We were fortunate enough to get not one, but two 5970 cards to test in our lab. The two cards came in Falcon Northwest's brand new Talon PC, which is the first 5970-equipped system we've reviewed.

You can read our full review of the kick-ass Falcon Northwest Talon here, but the system specs are below for your reference. Note that the Talon is a Lynnfield-based Core-i7 system, overclocked to 3.93GHz, as opposed to the 3.33GHz Core i7-975 Bloomfield proc we used in our testbed to benchmark the lone 5970 card.

Specifications

Processor	Intel 2.93GHz Core i7-870 @3.83GHz
MOBO	MSI P55-GD65
RAM	8GB Crucial DDR3/1600
Videocard	Two MSI Radeon HD 5970 in CrossFire mode
Soundcard	Onboard
Storage	Two Intel X25-M 80GB in RAID 0; 1TB Samsung Spinpoint 7,20rpm hard drive
Optical	Lite-On 22x DVD burner
Case/PSU	Silverstone case with Exotix paint job and 1,000 Silverstone PSU

In addition to a 48-hour plus burn-in period, we wanted to push Falcon Northwest’s dual-Radeon HD 5970 cards on something a bit harder so we connected the FNW Talon to a 30-inch 2560x1600 panel and cranked several benchmarks to maximum. The result? Impressive. Although, the drivers we used were still fairly early, the performance of the card is unquestionable. We had some concerns of running the card in a dual x8 mode (a limitation of the P55 chipset and Lynnfield platform) instead of full x16’s that an X58 platform would give us but the results don’t seem to show too much to be concerned about. The cards seem to offer better than expected scaling, even with early drivers.

Dual-5970s on Falcon Northwest's Talon PC

	HD 5970 Crossfire On	HD 5970 Crossfire Off
3DMark Extreme Overall (1900x1200)	18954	11686
3DMark Extreme GPU (1900x1200)	18763	11384
3DMark Extreme CPU (1900x1200)	23498	23561
3DMark Extreme GPU (2560x1600)	13494	9076
3DMark Extreme CPU (2560x1600)	23387	23399
Crysis, Very High, (2560x1600)	34.4 FPS	27.1 FPS
Far Cry 2 Ultra High (2560x1600)	185.2 FPS	107.4 FPS

Crysis, 1900x1200 (NO AA)	66.3 FPS	(did not run)
Crysis, 1900x1200 (16X AA)	64.3 FPS	(did not run)

Note that in Crysis, enabling 16X Anti-Aliasing doesn't seem to affect the framerate with CrossFire enabled.

Finally, a note on overclocking. The AMD overvolting tool has two settings for core and memory. You can overvolt the GPUs, overvolt the memory or both. Using the tool, we pumped up both voltage settings, then found the card to be stable at 870MHz core and 1250MHz memory (17 and 25% respectively.) Curiously, though, we saw little actual performance gain. 3DMark Vantage (extreme setting) only saw a 1% increase, from 12084 to 12182. We got two additional FPS in Crysis and either no increase or slight decreases in frame rate for Far Cry 2 and STALKER: Clear Skies. We probably need to do a better job of balancing memory and core clocks, but our initial experiences suggest that overclocking may not be worth the effort.

Maingear Stuffs Two HD 5970 Videocards into Shift PC

Paul Lilly — Wed, 18 Nov 2009 08:30:13 -0600

A little prerequisite reading is in order. If you haven't done so already, be sure to check out Lloyd Case and Gordon Mah Ung's article covering ATI's Radeon HD 5970, a videocard we've dubbed "The Undisputed Performance Champ." Once you read through that feature, you'll be in a better position to appreciate the awesome power that comes from cramming two HD 5970 videocards into a CrossFire X configuration, and that's exactly what Maingear has done.

The boutique system vendor today announced the immediately availability of said cards into its Shift PC. By doing so, Maingear says gamers are afforded 10 teraflops of pixel pushing power when enhanced with the company's Redline technology.

"The Maingear Shift was designed specifically to harness the incredible amount of GPU power that is on the horizon," said Wallace Santos, CEO and founder of Maingear. "The ATI Radeon HD 5970 truly ushers in a new era of everyday supercomputing for the consumer. Besides having a total of 3200 stream processors each for an incredible gaming experience, with Windows 7 and DriectCompute, these cards can be used to accelerate everything from video editing and transcoding to advanced photo manipulation."

As we previously reported, the Shift can be configured with either an Intel P55 or X58 foundation, and you can outfit either one of them with a pair of HD 5970 videocards. All that power doesn't come cheap, however, and you'll be looking at about an additional $1,500 over the base configuration, which includes an Nvidia GeForce GTS 250.

Image Credit: Maingear

Hard Case: Signs of Life at AMD?

Loyd Case — Tue, 17 Nov 2009 13:47:35 -0600

In the past year, AMD seems to have been taking a sort of “strategy du jour” approach. We ship low-cost processors! We do low-power CPUs! Our parts are great for overclockers! We love home-theater PCs!

Those messages weren’t really different from anything Intel, the 900 pound velociraptor in the CPU business, would offer up, but there was always a tinge of desperation. This became more noticeable as Intel slowly and methodically stripped away whatever technology edge AMD had. Intel’s Nehalem was really the last straw: AMD couldn’t even claim “true quad core” any longer.

The exception to this has been the company’s graphics division. For several years now, analysts have suggested that AMD’s acquisition of Toronto-based ATI was a distraction, and likely to bring down both companies. In the past twelve months, though, the graphics division has been surprisingly resilient. This, despite all the soap opera shenanigans that have plagued AMD--the latest being Hector Ruiz’s resignation as chairman of Global Foundries (the former AMD fab business), due to the Galleon Group insider trading scandal.

As the CPU division was forced to ship underperforming products that often sucked too much power, the graphics guys were kicking serious ass. The Radeon HD 4000 series GPUs weren’t the fastest you could buy, but they did have the best price/performance and forced Nvidia into an unanticipated price war. ATI then delivered the Radeon HD 5800 series, which proved to be the fastest GPUs available on the market. AMD has been selling all the 5800s they can manufacture, and would likely sell even more, if it weren’t for yield issues with TSMC’s 40nm manufacturing process.

Even with the successes on the graphics side, AMD still seemed like a bifurcated company: CPUs on one side, graphics on the other, with some platform (chipset) technologies the only link. Now, however, AMD is positioning itself as the only company with a unified approach to GPUs and CPUs. Nvidia, AMD notes, is only a GPU company while Intel is still only strong on the CPU side.

We’ve heard inklings of this, with the earlier announcements of its Fusion CPU, a melding of graphics functionality onto the CPU die. Now, however, AMD is taking its approach to CPUs–the melding of graphics and central processing–and applying that idea to its overall corporate strategy. AMD is trying to position itself not as a GPU or CPU company–instead, it’s a single company with feet planted firmly in both sides of the computational equation. Only AMD, with its expertise in building CPUs and GPUs, can really take advantage of the coming age of visual computing.

It’s about time.

To be sure, it’s a risky approach. Trying to excel in both arenas may simply result in creating mediocre products. On the other hand, the strategy plays to AMD's last real strength. Intel is struggling to get Larrabee out the door. Its current integrated graphics products are an also-ran compared to AMD and Nvidia’s chipset-level graphics. Nvidia, meanwhile, can’t get its Fermi architecture out the door, and continues its ongoing verbal and legal tussle with Intel.

So AMD has an opening it can exploit, however small and tenuous. Whether or not AMD will be successful depends entirely on execution at this point. On one side, AMD has a world class GPU design team which has been firing on all cylinders. On the other side, the CPU designs have been… less than competitive. AMD has had a poor track record in picking a strategy, then executing a product plan based on that strategy. It’s attempting something new, melding a CPU organization and a GPU team into one focused organization. Will the company pull it off? Or will we look back on AMD two years from now and wonder what happened?

I’m hoping AMD will pull it off, because I want multiple companies competing in both the CPU and GPU space. It’s competition that will drive creative and cool products, while consumers will benefit from reasonable price structures. But the stars are not necessarily aligned, and AMD’s competitors on both sides of its house are behemoth by comparison. Success, if it comes at all, will be a tough slog.

No BS Podcast #124: Surviving Friday the 13th

The Maximum PC Staff — Mon, 16 Nov 2009 14:30:15 -0600

In this episode, the gang discusses the implications of Intel's $1.25 billion anti-trust settlement with AMD, the release of the Motorola Droid, and the Modern Warfare 2's record-breaking launch. We also answer a few doctor questions, and Gordon fumes about replacable laptop batteries, headshots, and games that play themselves in his most hilarious rant of the week yet.

Do you have a tech question? A comment? A tale of technological triumph? Just need to get something off your chest? A secret to share? Email us at maximumpcpodcast@gmail.com or call our 24-hour No BS Podcast hotline at 877.404.1337 x1337--operators are standing by.

Subscribe: http://feeds.feedburner.com/maximumpc/1337

Cray Takes Back Supercomputer Crown

Paul Lilly — Mon, 16 Nov 2009 11:45:19 -0600

IBM's Roadrunner system at the U.S. Department of Energy's Los Alamos National Laboratory in New Mexico is no longer the planet's most powerful supercomputer. That distinction now belongs to a Cray supercomputer named "Jaguar" at Oak Ridge National Laboratory, which regained the performance crown over the weekend, ComputerWorld reports.

Jaguar, which benefited from a few recent upgrades, is now capable of 1,759 petaflops per second courtesy of 224,162 processor cores. That's enough to jump ahead of IBM's Roadrunner, which dropped to 1,042 petaflops per second after it was repartitioned.

Number three on the list of supercomputers is Kraken at the National Institute for Computational Sciences at the University of Tennessee. Kraken is capable of churning out 832 teraflops per second and was ranked No. 6 in June.

One of the more interesting supercomputers belongs to China. The hybrid Intel-AMD Tianhe-1 in the city of Tianjin pushes out 563 teraflops per second, putting it in fifth place. China's supercomputer combines Intel's Xeon processors with AMD-brand GPUs as accelerators. Each node contains two Xeon chips attached to two AMD GPUs.

Image Credit: knoxnews.com