by Anton Shilov
09/24/2009 | 08:44 AM
With the release of the world’s first and yet only DirectX 11 graphics chip ATI, graphics business unit of Advanced Micro Devices, is again on top of the graphics cards world. Unfortunately, AMD’s central processing units cannot outperform chips from Intel Corp. In addition, global economic slump has hit AMD very hard. Three years after ATI became a part of AMD, the question arises: would AMD survive without ATI?
The acquisition of ATI Technologies by Advanced Micro Devices started to seem controversial after ATI began to lose market share, sales volumes and failed to deliver DirectX 10 lineup on time. Moreover, AMD-ATI reconsidered the Fusion central processing unit (CPU) roadmap a number of times, which made the merge even more questionable. The year 2007 was a complete disaster for the “new” AMD. The company delayed release of its quad-core AMD Opteron “Barcelona” processor and subsequently AMD Phenom X4 “Agena” chip and when it released them it appeared that the design has an erratum, which essentially stalled shipments. Moreover, ATI Radeon HD 2000 failed to beat Nvidia GeForce 8000-series in terms of performance and then ATI Radeon HD 3000 did not manage to offer performance advantages too.
“Both AMD and ATI suffered from inadequate execution in 2007, and it showed in their results. Both got on track by the end of that year, and have delivered on their roadmaps,” said Nathan Brookwood, the principal analyst with Insight 64.
Inadequate execution took its toll: sales of discrete desktop ATI-based graphics cards slumped 2.7 times from Q3 2006 to Q2 2009, according to Jon Peddie Research market tracking firm. Sales of microprocessors also declined very considerably.
This week ATI, as a part of AMD, launched the world’s first DirectX 11 graphics chip, which is a pretty clear evidence that the graphics chip designer is back on track. ATI Radeon HD 5870 is currently the world’s highest performance graphics processing unit (GPU) and has the richest feature-set available.
But what about AMD’s primary microprocessor business? The company seems to have excellent execution at this point, but its desktop chips have consistently failed to compete with Intel Corp.’s offerings in the high-end market segment for three years now and it is unlikely that next year the situation will change considerably, unless the company’s six-core Phenom II X6 “Thuban” chip has excellent clock-speed potential. It will still take some time before AMD will be in position to regain performance leadership on the CPU market. On the other hand, the fact that AMD is capable of releasing a six-core chip is a sign that its processor business is showing signs of recovery.
Considering rather massive slip of unified CPU-GPU “Fusion” chip (which AMD calls accelerated processing unit, or APU), it is not completely clear why AMD decided to take rather huge multi-billion loan, acquire ATI Technologies, then fail to execute on both companies’ roadmaps and sell off its manufacturing facilities.
Analyst Nathan Brookwood believes that the take over actually helped AMD to survive: the company gained new revenue streams with discrete graphics processing units as well as its core-logic sets, which not only bring additional earnings, but also gave AMD technological advantage in terms of graphics processing over Intel: it is indisputable fact that ATI Radeon graphics cores are a lot better compared to Intel Graphics and Media Accelerator (GMA) cores.
“The ‘old’ AMD, with integrated manufacturing, but without graphics, would have been challenged to survive last year's global downturn and Intel's new Calpella/King’s Creek/Piketon platforms. The 'new’ AMD can survive on lower revenues, and can credibly argue that its integrated graphics platforms deliver a better visual experience than Intel's, even if the Core i3 and i5 CPUs outperform AMD's processors. It would never have been able to make such a claim without the ATI acquisition under its belt,” said Mr. Brookwood.
AMD itself reckons that it is in unique position on the market, since it is the only company that can deliver both x86 microprocessors and leading-edge graphics processors. Still, the company admits that the global economic crisis impacted it quite seriously.
“As you know well, much has happened both in the industry and in the global economy in the three years since the ATI acquisition occurred. Having said that, we remain convinced that the rationale underpinning the ATI acquisition remains sound. Only two companies in the world deliver x86 processors in volume, and only two companies deliver leading edge 3D graphics. AMD is the one and only company in the world can do both - and that's a unique, direct benefit from our ATI acquisition,” said Michael Silverman, a spokesperson for AMD.
In fact, even the massive delays of Fusion, which was once promised to become available in late-2008 or early-2009, makes some sense, claims the analyst. Since Intel continues to integrate its 5th generation GMA graphics core into its Clarkdale/Arrandale processors in 2010 and 6th generation GMA engine into Sandy Bridge/Ivy Bridge in 2011, AMD will continue to have performance advantage with ATI Radeon cores built into chipsets in terms of performance and feature-set and will also be able to offer DirectX 11 graphics with its first APU chip code-named Llano in 2011.
“I am as disappointed as anybody by the continued push-out of AMD's Fusion CPUs, but can't argue with their rationale for moving them out. Although Intel has multi-chip module integrated graphics now, and single chip integrated graphics next year with Sandy Bridge, both products use Intel's archaic GMA technology that lacks OpenCL and DirectCompute support. This means AMD's products offer better graphics features and greater functionality now, and will beat Intel to market with CPUs that offer OpenCL and Direct Compute, since Larrabee-type graphics don't get integrated onto Intel CPUs until 2013,” said the analyst.
So far the information about Llano has been quite sketchy, but in case the details are summarized, the accelerated processing unit will seem to be quite an interesting solution for entry-level market: it will feature up to four Shanghai/Phenom II-class cores; 4MB of L3 cache; PC3-12800 (DDR3 1600MHz) memory controller, possibly, with some tweaks to better serve x86 and graphics engines; DirectX 11 graphics core with third-generation universal video decoder; PCI Express 2.0 bus for external graphics cards. What is even more interesting is that Llano appears to be a monolithic chip combining both x86 and graphics cores. Of course, Phenom II will hardly deliver extreme performance in 2011, but the better graphics core will give AMD an edge over Intel’s offerings.
There is still a lot of work to do for both AMD and its graphics business unit to take maximum advantage of the merger between AMD and ATI Technologies. One of the main thing to achieve is to gain synergy between AMD and ATI engineers and execute successfully on the “x86 everywhere” strategy through the evolution of both CPU and GPU technologies.
In the mid-term future AMD will have to develop ultra low-power x86 microprocessors and integrate advanced graphics cores into them to compete not only against Intel Atom-based system-on-chips (SoC), but also against SoCs like Nvidia Tegra or chipsets like Qualcomm Snapdragon. Besides, AMD will also have to integrate advanced multimedia, input/output and communication functionality into its platforms going forward. AMD’s forthcoming code-named Ontario APU features CPU core based on low-power code-named Bobcat micro-architecture and an unknown graphics core, but the company positions the chip for netbooks, based on current roadmap.
The future of computing is very diverse: end-users and enterprises want to have advanced capabilities inside handsets, high-definition graphics and advanced security on desktops, full Internet experience on HDTVs, very long battery life along with desktop capabilities on laptops, extreme performance of servers amid low power consumption and a lot, lot more.
In order to respond to new challengers, AMD-ATI have to do a massive amount of work in the next three to five years.