Co-opting GPU for CPU Tasks Advanced by NVidia

Earlier this week, engineers at nVidia put the finishing touches on version 0.8 of its Compute Unified Device Architecture system for Windows and Red Hat Linux. CUDA’s objective is to enable C programmers to utilize the high-throughput pipelining architecture of an nVidia graphics processor – pipelines that are typically reserved for high-quality 3D rendering, but which often sit unused by everyday applications – for compute-intensive tasks that may have nothing to do with graphics.

Today, the company announced its first C compiler – part of the CUDA SDK, which will enable scientific application developers for the first time to develop stand-alone libraries that are executed by the graphics processor, through function calls placed in standard applications run on the central processor.

NVidia’s objective is to exploit an untapped reservoir on users’ desktops and notebooks. While multi-core architecture has driven parallelism in computing into the mainstream, multi-pipeline architecture should theoretically catapult it into the stratosphere. But applications today are naturally written to be executed by the CPU, so any GPU-driven parallelism that’s going to happen in programming must be evangelized first.

Which is why the company has chosen now to make its next CUDA push, a few weeks prior to the Game Developers’ Conference in San Francisco. The greatest single repository of craftspersons among developers may be in the gaming field, so even though games already occupy the greater part of the GPU’s work time, it’s here where a concept such as CUDA can attract the most interest.

“The GPU is specialized for compute-intensive, highly parallel computation – exactly what graphics rendering is about,” reads nVidia’s latest CUDA programming guide (PDF available here), “and therefore is designed such that more transistors are devoted to data processing rather than data caching and flow control.”

Huge arithmetic operations may be best suited to GPU execution, nVidia engineers believe, because they don’t require the attention of all the CPU’s built-in, microcoded functions for flow control and caching. “Because the same program is executed for each data element,” reads the CUDA v. 0.8 guide, “there is a lower requirement for sophisticated flow control; and because it is executed on many data elements and has high arithmetic intensity, the memory access latency can be hidden with calculations instead of big data caches.”

For CUDA to actually work, however, a computer must be set up with an exclusive NVidia display driver; CUDA is not an intrinsic part of ForceWare, at least not yet. In addition, programs must be explicitly written to support CUDA’s libraries and custom driver; it doesn’t enable the GPU to serve as a “supercharger” for existing applications. Because the GPU is such a different machine, there’s no way for it to take a load off the CPU’s shoulder’s directly, like an old Intel 8087 or 80186 co-processor used to do.

So an application that supports CUDA thus, by definition, supports nVidia. AMD also has its own plans for co-opting GPU power, which it made immediately clear after its acquisition of ATI.

The CUDA programming guide demonstrates how developers can re-imagine a math-intense problem as being delegated to processing elements in a 2D block, like bestowing assignments upon a regiment of soldiers lined up in formation. Blocks and threads are delegated and proportioned, the way they would normally be if they were being instructed to render and shade multi-polygon objects. Memory on-board the GPU is then allocated using C-library derivatives of common functions, such as cudaMalloc() for allocating blocks of memory with the proper dimensions, and cudaMemcpy() for transferring data into those blocks. It then demonstrates how massive calculations that would require considerable thread allocation on a CPU are handled by the GPU as matrices.

“This complete development environment,” read an nVidia statement this morning, “gives developers the tools they need to solve new problems in computation-intensive applications such as product design, data analysis, technical computing, and game physics.”

[originating url]

NVidia Now a Supplier for MP3 Player

In a move that may very well have saved the assets of a once-venerable US media chip producer from being auctioned off, graphics chip maker nVidia announced this morning it is acquiring San Jose-based PortalPlayer, a producer of embedded media processing chips for devices such as SanDisk’s Sansa MP3 player, in a stock purchase plan totaling $357 million.

PortalPlayer had been struggling to regain its footing as a producer of multimedia processing chips after the customer that essentially put it on the map, Apple, dropped it last April without much warning as its key supplier for its video iPod. Up to that point, Apple had reportedly accounted for 95% of PortalPlayer’s business. Its replacement was Samsung, which apparently offered Apple a discount on flash memory to sweeten the deal; PortalPlayer is not a flash producer.

The PortalPlayer design has actually been considered quite innovative, and worthy of its presence in the iPod, were it not for Samsung’s package deal. Prior to winning the Apple contract, Samsung executives had publicly dubbed their proposed replacement “the PortalPlayer killer.”

Despite rumors of the company’s imminent death, PortalPlayer did manage, against all odds, to remain in the black. Two weeks ago, it reported net income for its fiscal third quarter 2006 at $1.5 million, up $100,000 from the previous quarter. That blank ink came at a cost, however: the layoff of 14% of its workforce in June, and the scaling back of operations and expectations.

As part of its comeback plan, PortalPlayer had staked a name for itself in a burgeoning new market for embedded components: secondary, miniature LCD displays for notebook computers. Its design, called Preface, consists of low-power displays on the outside of the clamshell, that can remain switched on even while the rest of computer is on standby. These displays can register the time, check the current box scores, present the weather forecast, and even show recent e-mails. Microsoft has vowed to support the concept behind this technology in its upcoming Windows Vista.

Preface could be a lucrative new technology for nVidia, which now knows it’s going up directly against AMD — no longer just ATI — in the production of new platform technologies for notebook computers.

There may be new momentum behind nVidia’s move. A recently released Merrill Lynch analyst’s report projects that, in its last fiscal quarter, the company’s market share in the entire graphics chip market increased a staggering eight points, to 29%.

Its share of the integrated chip market alone, the report also states, leap-frogged in size over that of its nearest competitor, Intel, although embedded graphics is generally known to be a low-margin business.

Yet the news of today’s acquisition does throw cold water on rumors that nVidia is seeking to be acquired by Intel. With an integrated graphics chipset business of its own, Intel doesn’t need nVidia the way AMD needed ATI. Meanwhile, with Intel also firmly positioned as a provider of embedded chipsets as well, it needs PortalPlayer even less.

[originating url]

PureVideo HD from NVIDIA

PureVideo HD from NVIDIA

The high definition video market is quite a jungle, and it’s absolutely not certain that your computer is capable of playing high definition video. NVIDIA has announced a a new chip called PureVideo HD that is capable of producing great high definition end results. More information inside.

Full press release from NVIDIA:

NVIDIA Enables an Outstanding HD DVD and Blu-Ray Movie Experience With Newly Released NVIDIA PureVideo(R) HD Technology

NVIDIA is First Graphics Company to Enable Users to Build or Upgrade a PC to Play HD DVD and Blu-ray Movies

SANTA CLARA, Calif., Nov. 2 — Movie lovers can now build a PC to enjoy a spectacular HD DVD or Blu-ray experience with the release of new NVIDIA ForceWare(R) drivers featuring NVIDIA PureVideo HD technology. PureVideo HD technology combines high-definition video decodeacceleration and post-processing to deliver extraordinary picture clarity, smooth video, brilliant color, and precise image scaling for high definition movies(1). The drivers are available now from

This is a milestone in PC entertainment,” said Scott Vouri, general manager of multimedia at NVIDIA. “NVIDIA is proud to be the first graphics processing company in the world to make it possible for consumers to build or upgrade a PC to play HD DVD and Blu-ray movies.”

PureVideo HD technology is included with all 7-series NVIDIA GeForce graphics cards. However, there are several requirements for building or upgrading a system to enjoy protected high definition content. To enjoy the ultimate HD DVD or Blu-ray movie experience on a PC, consumers will need:

  • A PCI Express graphics card with NVIDIA GeForce 7 Series HDCP-capable GPU, secure HDCP CryptoROM, and 256MB graphics memory (see list here).
  • WHQL-certified NVIDIA ForceWare drivers that feature PureVideo HD technology (
  • An optical disc drive that supports Blu-ray or HD DVD movie playback
    Blu-ray or HD DVD movie player software from CyberLink, InterVideo, or Nero
    An HDCP-compliant display.
  • A dual-core CPU with 1GB of RAM PureVideo HD provides hardware acceleration for decoding H.264, VC-1, WMV, and MPEG-2 movies to display crystal clear images that have up to six times the detail of standard DVD movies. In addition, the PureVideo discrete video processing core offloads the CPU and 3D engine of complex video tasks, freeing the PC to run multiple applications simultaneously. PureVideo HD delivers the ultimate high-definition movie experience on a PC.

PureVideo HD-compliant graphics cards include required content protection circuitry (HDCP) for playing the new Blu-ray and HD DVD movies at the highest quality possible, and are directly integrated with the leading HD movie software players. For more information about building or upgrading a system to support Blu-ray or HD playback, go to:

1) As HD DVD and Blu-Ray are new formats containing new technologies, certain disc, digital connection, compatibility and/or performance issues may arise, and do not constitute defects in the product. Flawless playback on all systems is not guaranteed.

Nvidia to build x86 processors, take on Intel, AMD?

NvidiaThe Inquirer says it has “confirmed” that 3D graphics giant Nvidia is hard at work building its own x86 processor with integrated graphics to compete with the offerings of Intel and AMD. Both Intel and AMD–which acquried ATI back in July–have been busy strapping integrated graphics on their own chips, it seems logical that Nvidia would want to play, too. Their efforts would be bolstered by its recent hiring of a bunch of engineers from Stexar, a now-defunct company that was founded by ex-Intel brains. The Inq says development is already underway, and we’ll be seeing a new CPU from Nvidia sometime in 2006.

[Via Engadget]

[originating url]