TechieGroups

At AMD’s Fusion Developer Summit 2011 AMD announced Graphics Core Next, their next-generation GPU architecture. GCN would be AMD’s Fermi moment, where AMD got serious about GPU computing and finally built an architecture that would serve as both a graphics workhorse and a computing workhorse. With the ever increasing costs of high-end GPU development it’s not enough to merely develop graphics GPUs, GPU developers must expand into GPU computing in order to capture the market share they need to live well into the future.

At the same time, by canceling their 32nm process TSMC has directed a lot of hype about future GPU development onto the 28nm process, where the next generation of GPUs would be developed. In an industry accustomed to rapid change and even more rapid improvement never before have GPU developers and their buyers had to wait a full 2 years for a new fabrication process to come online.

All of this has lead to a perfect storm of anticipation for what has become the Radeon HD 7970: not only is it the first video card based on a 28nm GPU, but it’s the first member of the Southern Islands and by extension the first video card to implement GCN. As a result the Radeon HD 7970 has a tough job to fill, as a gaming card it not only needs to deliver the next-generation performance gamers expect, but as the first GCN part it needs to prove that AMD’s GCN architecture is going to make them a competitor in the GPU computing space. Can the 7970 do all of these things and live up to the anticipation? Let’s find out…

TIN BOX FLOGGER Dell has been caught offering misleading advice on the image quality produced by ‘standard’ and ‘high-end’ graphics cards.

As part of Dell’s ‘Help Me Choose” website on graphics cards, the firm shows two seemingly identical monitors displaying the same image. The image on one of the monitors connected to a “standard graphics card” appears to have muted colours and be out of focus, while the second monitor connected to a “high-end graphics card” has a sharper image with higher colour saturation.

While the drastic image difference is woefully misleading, curiously Dell’s textual description of integrated and discrete graphics, albeit simplistic, is on the mark.

Dell’s summary of integrated graphics, which the firm claims is a “basic solution”, says, “While integrated graphics are capable of processing some 3D images, the shared resources can sometimes prevent them from delivering an optimal experience.”

As for Dell’s summary of discrete graphics cards, which the firm labels as a “Mid-range to high-end solution”, it said, “Because this solution does not burden the CPU and system memory, overall system performance is likely to be significantly better with discrete graphics when running applications that use complex images.”

Although many computer enthusiasts might be unhappy at the lack of detail, there’s nothing fundamentally wrong with Dell’s summary of integrated and discrete graphics cards. The problem is with the graphical representation, which the firm knows is what consumers will remember.

Assuming Dell is showing two identical monitors connected by the same interface using the same colour profile, then it is incredibly unlikely that a ‘standard graphics card’ – which Dell does not define – could produce such low image quality compared to a ‘high-end graphics card’ – another term that Dell does not define.

Image qualities from GPU vendors such as AMD, Intel and Nvidia do vary but are usually only discernable in games with high-levels of anti-aliasing and anisotropic filtering being applied. Neither would apply in the image of a Windows desktop that is displayed on both monitors in Dell’s images.

Dell was unable to provide us with comment regarding its “Help me choose” graphics card advice by press time. µ

TAIWANESE HARDWARE MAKER Asus has launched its limited edition Republic of Gamers (ROG) Mars II graphics card for hardcore gamers.

Asus has produced a giant of a graphics card in the Mars II, which is apparently 22 per cent faster than a Geforce GTX590. The firm claims its dual GPU card and 3GB of GDDR5 RAM make it capable of running DirectX 11 games in higher than HD 1080p resolution on full detail settings.

Asus said, “The limited-edition card is built on twin NVIDIA GeForce GTX 580 GPUs and a mammoth 3GB of GDDR5 video memory, with a proprietary ROG design that combines extreme cooling and extensive tuning facilities that cater to the most demanding power user.”

The monster sized card houses two 120mm fans for a 600 per cent increase in air flow and has a DirectCU II copper design that results in a 20 per cent drop in temperature compared to a GTX590. There’s a Turbo Fan button on the card that sets the fans to full speed for those extra demanding gaming sessions.

Asus has used an “extra stable” 21-phase “super alloy” design to ensure constant wattage and prevent power loss. It claims this also yields a two and a half times longer lifespan and 35 degree cooler operation temperature compared to a Geforce GTX590.

The card has a “Tweak utility” that allows users to modify parameters such as clock speed, fan speed and voltages from a user interface for overclocking. Just so you know it’s special, the collectors’ edition will have a laser-carved, sequentially-numbered aluminium plate to advertise its status.

The card will need to be powered by three (3) 8-pin PCI-Express power connectors, so you’ll need make sure your power supply unit can handle it. It will offer two HDMI and two DVI ports.

We don’t have an exact price yet for the Asus ROG Mars II graphics card, but we expect it to retail for £700 or more. A Geforce GTX580 graphics card costs around £400. µ

GRAPHICS CARD VENDOR HIS launched its 6770 IceQ X and 6770 IceQ X Turbo cards.

These HIS 6770 IceQ cards are based on AMD’s reference Radeon HD6770 with 1GB GDDR5 memory, but they feature the firm’s trademark IceQ heatsink and fan, which has two heatpipes, and the firm claims it is quieter than the heatsink fan on reference boards. The considerable cooling also helps HIS ship its graphics cards factory overclocked.

The HIS 6770 IceQ has the GPU is clocked at 850MHz and the memory at an effective 4.8GHz. On the HIS 6770 IceQ Turbo the GPU runs at 880MHz with the 1GB GDDR5 memory running at 5GHz.

Both HIS 6770 IceQ boards have two dual-link DVI outputs as well as Displayport and HDMI outputs. Like all AMD Radeon HD6770 boards, the HIS 6770 IceQ units support Microsoft’s DirectX 11 and AMD’s Eyefinity multi-display software.

Although HIS didn’t announce pricing for its 6770 IceQ graphics cards, similar cards from other vendors such as Gigabyte, MSI, Sapphire and XFX all retail from £100 and go up to £120, though Gigabyte’s most expensive board in that price range has the AMD Radeon HD6770 GPU running at 900MHz.

It is likely that HIS will price its two boards somewhere in the middle of that pack, at around £105, more or less. µ

CHIP DESIGNER VIA has closed a deal with smartphone vendor HTC to sell off its ailing S3 Graphics division.

S3 Graphics, which has been the eternal underdog in the 3D graphics arena, had all but ceased to exist in retail and outside of Asia. Its financial performance was lacking, to say the least, and even with a much-needed cash injection in 2005 from its investment partner, WTI Investment International, it failed to re-launch its graphics business.

VIA has announced that the deal will cost HTC a round 0 million (US), 7 million of which will line VIA Technologies’ coffers while 3 million will go to WTI for its share of S3. In exchange, VIA will retain a permanent license to current S3 graphics intellectual property that will allow it to continue marketing its chipsets without having to shell out to HTC for licensing.

This is Taiwan, however, and there is usually a twist.

WTI Investment International is owned by Cher Wang, who coincidentally is also chair of the board at both VIA and HTC. So, regulatory approval aside, things are being kept in the family. Wang pays Wang for the IP owned by Wang through a different investment portfolio. Needless to say the transaction has already been approved by the boards of directors at all three companies.

An era has ended, we’re afraid to say. The INQUIRER can still remember when Creative Labs snuck a handful of journalists to a private meeting just so it could show off the wonders of S3 Texture Compression. That was 11 years ago, when S3 was still rather competitive.

Actually S3 Graphics’ single most valuable cash cow thus far has been S3 Texture Compression, which it has licensed to just about every graphics vendor. However, considering how the smartphone business is measuring itself against HTC, the company must be interested in taking the fast lane to designing an S3 Graphics core that will make it independent of other chip vendors’ SoC designs.

Smartphone designers don’t have many options available to them if they want to compete with the smartphone and tablet manufacturers that source chips from Nvidia (Tegra), Qualcomm (Snapdragon and Adreno), Imagination Technologies (PowerVR) or ARM (Mali).

It seems like a win-win situation for Wang and her investment portfolio. We’re just surprised it didn’t happen earlier. µ

If desktop graphics hardware can be more than a little confusing, deciphering performance of mobile graphics parts can be (and has historically been) an absolute nightmare. Way back in the day it was at least fairly easy to figure out which desktop chip was hiding in which mobile kit, but both AMD and NVIDIA largely severed ties between mobile and desktop branding. They may not want to readily admit that, and in the case of certain models they still pretty heavily rely on the cachet associated with their desktop hardware, but it's by and large true. So to help you make sense of mobile graphics, we present to you the first in what will hopefully be a regular series of guides.

We’ve just returned from sunny Bellevue, Washington, where AMD held their first Fusion Developer Summit (AFDS). As with other technical conferences of this nature such as NVIDIA’s GTC and Intel’s IDF, AFDS is a chance for AMD to reach out to developers to prepare them for future products and to receive feedback in turn. While AMD can make powerful hardware it’s ultimately the software that runs on it that drives sales, so it’s important for them to reach out to developers to ensure that such software is being made.

While AFDS serves many purposes, the final purpose – and what is going to be most interesting to most outside observers – was to prepare developers for what’s coming down the pipe. AMD has big plans for the future and it’s important to get developers involved as soon as is reasonably possible so that they’re ready to use AMD’s future technologies when they launch. Over the next few days we’ll talk about a couple of different things AMD is working on, and today we’ll start with the first and most exciting project: AMD Graphics Core Next, AMD's next generation GPU architecture.

CHINESE HARDWARE MAKER Hightech Information System (HIS) has announced its upgraded HD6950 graphics card range will be available from 14 April.

Its graphics card lineup consists of three cards and comes with 1GB or 2GB memory options. The range starts with the HD6950 Iceq X, which has a core clock speed of 800Mhz and a 5GHz memory clock speed.

The middle of the three is the HD6950 Iceq X Turbo with a core clock speed of 840Mhz and a memory clock speed of 5.12GHz. The top of the line is the HD6950 Iceq X Turbo X. This card has the fastest clock speeds of 880MHz and 5.2GHz, with HIS claiming it’s “the world’s fastest HD6950″.

The cards feature high definition technology that can display 3D content at HD 1080p, provided you have the appropriate accompanying hardware. This can support 3D gaming, 3D Blu-ray and third party stereo 3D middleware software with its AMD Radeon HD6800 Series GPU.

The AMD Radeon PCI-Express HD6950 range supports DirectX11 and comes with dual DVI ports, a single HDMI port and two mini display ports. The INQUIRER has requested prices for the cards, but to get an idea, the HIS HD6950 is currently available in the UK for around £200. µ

TABLET AND NOTEBOOK popularity is expected to help shrink discrete GPU sales by more than 10 per cent by 2014 and drive the development of graphics enabled CPUs.

According to market research firm Isuppli, sales of tablets and notebooks will depress discrete graphics device shipments from 73 million in 2009 to 62 million in 2014. Their rise will push takeup of new graphics enabled processors including CPUs to replace the old style GPUs.

Isuppli predicts that a doubling of ultraportable notebook shipments by 2014 will mean that those products will lead the way. As well as higher shipments Isuppli foresees 82.9 per cent of notebook PCs using graphics crunching microchips by 2014, compared to 39 per cent today. If Isuppli can predict such a precise figure as 82.9 per cent four years ahead of time, could they please contact this journalist with next week’s lottery numbers?

The Isuppli prediction would also seem to be good news for Intel, because of the graphics shoveling chips not the lottery. Isuppli says Chipzilla has already cornered the market in them with its Core I series. For AMD, the research company expects it “to launch graphics-enabled microprocessors in the fourth quarter of 2010 and 2011 time frame.”

However AMD has pushed ahead with its Vision branding that has an emphasis on consumers use of graphics and video and it has already announced its Fusion graphics Accelerated Processing Unit product line. The INQUIRER hears from reliable sources that Fusion will tip up in this fourth quarter. Perhaps this helps explain Isuppli’s fourth quarter graphics-enabled microprocessors prediction?

The focus on small form factor devices bodes equally well for ARM, which could win big in the tablet market after years of success in the smartphone arena. ARM’s new A10 Eagle chip is being licensed by Texas Instruments for mobile computing devices. µ

THE CURRENT reference design single card graphics performance leader is the AMD ATI Radeon HD5970. Basically a combination of two HD5870 1GB GPU blocks slowed down by some 20 per cent to accomodate the heat and power limits of the PCIe card specification, and connected via an on-board PLX PCIe bridge, the HD5970 has led the market for nearly all of the past year. Now, prior to the arrival of the AMD ATI Radeon HD6000 ‘Southern Islands’ GPU line in October, there is a kind of unofficial refresh going on at the high end.

Basically, the key vendors like Asus, Gigabyte, XFX and Sapphire are offering sped-up top end graphics cards that would have otherwise qualified to be called, say, HD5890 and HD5990, but since it’s not a full new product SKU rollout, they are considered the accelerated factory pre-overclocked units. At the very top of the pack is Asus’ Ares.

The card has the same architecture as the normal AMD ATI Radeon HD5970 dual GPU setup, and even the PCB dimensions are about the same. However, the GPUs on board run at the full HD5870 individual speed of 850MHz GPU and 4.8GHz GDDR5 memory, and, on top of that, each GPU has 2GB of RAM for a total of 4GB on the card. Wonderful! But, the changes required a brand new cooling system barely fitting the two slot width and, of course, much more power. Here you have two 8-pin plus one 6-pin graphics power connector on the card. Put two of those cards in a QuadFire parallel GPU setup on a, say, Intel Core i7-980X six core platform, and you’ll exhaust a 1000W PSU.

The card comes in an ultra large carton box, larger than even server mainboard packaging. Inside it is a black suitcase, James Bond style, which when open reveals the card and its accessories.

At the first look, the Asus Ares card is big and beautiful, a statement that fits this monstrous card just perfectly. It is impressive looking and, in a defensive situation, with its combination of weight and sharp edges, it could be a deadly weapon.

It took a bit of extra care to insert the card into our initial test platform, the Asus Rampage III Extreme mainboard using the Intel Core i7-980X six core CPU running at the default 3.33GHz clock. The 6GB of Geil Black Dragon DDR3-1600 RAM and an Intel X25-M 160GB SSD, as well as the Thermaltake 1000W PSU in our trusty Xigmatek Midgard-S chasis, which by now has survived three board swaps without a scratch, rounded out the test bed system configuration. The card was surprisingly silent, even during the benchmark runs at full load.

In this initial test, before we try to overclock the card further and use its Asus Smartdoctor and GamerOSD utilities, I ran the usual Windows 7 64-bit platform with the 3Dmark Vantage DX 10 and Unigine Heaven 2 DX 11 tests, as well as Sandra synthetic GPU performance benchmarks. Here they are:

3Dmark Vantage on Ares:

And on the generic Asus HD5970 on the same system:

Unigine Heaven:

Sandra GPU render:

Sandra GPGPU compute, over 5 TFLOPs single precision and 1.2 TFLOPs double precision floating point on a single card, with plenty of local RAM for the job:

And GPU memory too:

Impressive! This is by far the fastest GPU setup in a single slot I’ve ever seen. I wonder how it’d scale in a quad GPU dual card configuration, but what I can say is that Asus has, with this extra bit of engineering, created a true multi GPU performance monster, without sacrificing single GPU performance or memory capacity.

Note yet another configuration opportunity here. If you’re using the card for GPGPU compute applications, where the extra 2GB of memory per GPU chip helps a lot, you are not bound by the Crossfire limits. In fact, in a mainboard like the EVGA SR2 or the Gigabyte X58A-UD9, you could insert four of these ARES cards, each with its own PCIe X16 link, and have eight GPUs for over 20 TFLOPs single precision and over 4 TFLOPs double precision floating point capability, in a single box. All that, of course, assuming that your compute routines are happy with relying on the AMD Stream or OpenCL programming approaches.

In Short
Asus set another record here with the Ares Limited Edition ATI Radeon HD5970 graphics card, at least until and if it releases the rumoured Mars 2 dual Nvidia GTX480 GPU Limited Edition on a single card.

Maybe Nvidia will be nice enough to provide Asus with those rare, kept aside, full 512-shader bins of the GF100 chip to make a thousand core dual GPU card. That thing would need three 8-pin power plugs, even more than the Ares. Compared to the dual Nvidia GF100 card’s projected 600W power draw, this dual ATI Radeon HD5970 Ares card will look positively power saving.

In the meantime, we’ll look more closely at the Asus Ares card’s graphics performance against other GPUs in more environments, as well as how well it overclocks. Watch this space. µ

The Good
Top performance, reasonably compact design, large memory, limited edition card.

The Bad
Since AMD’s next-generation ‘Southern Islands’ HD6000 line of GPUs is just around the corner, there might have to be an Ares 2 soon.

The Ugly
Nothing.

Bartender’s Score
9/10