Could Intel stop Nvidia from producing chips for Nehalem?
Last July, Nvidia announced it would make its chipsets and SLI multi-GPU technology interoperable with Intel's latest generation of CPUs, with their highly-advanced Nehalem architecture. Monday, Intel said no to that in court.
In a court filing Monday in Delaware which remains under seal, CPU maker Intel asked for a declaratory judgment against GPU and chipset producer Nvidia, stating that the terms of the companies' existing x86 technology license do not extend to the Nehalem generation. This according to Intel spokesperson Chuck Mulloy, who confirmed the filing with Betanews this morning.
Nehalem is the first architecture from Intel to use an embedded memory controller, which it calls QuickPath Interconnect (QPI); AMD has had a similar technology called HyperTransport for several years. That QPI radically changes the way graphics processors and motherboard chipsets -- the two key products Nvidia makes -- communicate with the CPU. In order for Nvidia to utilize some of its own key technologies, including its multi-GPU system called SLI, system builders believe Nvidia would need to make -- or to have already made -- radical changes in its own designs.
In a July announcement, according to The Register, Nvidia indicated that it had already done so, and that Nehalem support was on its way. Intel's response has been negative toward that ever since, and it has just formalized that response in court.
"The suit seeks to have the court declare that Nvidia is not licensed to produce chipsets that are compatible with any Intel processor that has integrated memory controller functionality, such as Intel's Nehalem microprocessors," Mulloy told Betanews, "and that Nvidia has breached the agreement with Intel by falsely claiming that it is licensed."
Mulloy went on to say his company has been negotiating with Nvidia over a licensing agreement for more than a year, without success. One possible outcome of the court matter could be a resolution which does lead to a license for Nvidia.
For its part, Nvidia took its grievances public this morning, with a statement from President and CEO Jen-Hsun Huang that asks its readers to visualize an epic struggle between the CPU and the GPU for validity in the personal computer, and alleging that the CPU as an architecture in and of itself is becoming outmoded.
"We are confident that our license, as negotiated, applies," remarked Huang. "At the heart of this issue is that the CPU has run its course and the soul of the PC is shifting quickly to the GPU. This is clearly an attempt to stifle innovation to protect a decaying CPU business."
The company's statement went on to cite its numerous design wins with leading PC builders, including Apple, HP, and Lenovo, as an indication that sentiment about what chip actually defines a computer is shifting toward the GPU. But it concedes that all these design wins were made possible through the existence of its current technology license with Intel, and agrees with Mulloy that the companies have been in talks for over a year.
For its part, Intel has advanced its own theories that the "soul," if you will, of graphics rendering may move away from the GPU and toward the CPU, its own home territory. For a few years now, Intel has been building its own graphics processing architecture, code-named "Larrabee," around a many-core theory in which software provides the pipelining methodology that GPU makers such as Nvidia and ATI encode into their hardware.
"This paper describes a highly parallel architecture that makes the rendering pipeline completely programmable," reads the prelude to an Intel white paper on Larrabee produced last August (PDF available here). "The Larrabee architecture is based on in-order CPU cores that run an extended version of the x86 instruction set, including wide vector processing operations and some specialized scalar instructions."
Nvidia's early reaction to Larrabee was, to put it mildly, childish. In a public statement last April, without divulging much specifics, Huang promised his company would "open a can of whoop-ass" on Larrabee. Since that time, Nvidia has taken public positions on Intel literally blaming it for many of the ills facing the technology industry at large. For example, just last week, Huang blamed Intel for the backslide in PC sales by having created a netbook processor, Atom, that enabled customers to settle -- in his opinion -- for cheaper PCs with lesser graphics quality.
With 60% lower revenue than the previous year, and with its arch-rival ATI now a fully integrated component of AMD, Nvidia's last lifeline -- should it come to that -- would probably have to come from Intel. Huang may bluster, but it's Intel holding the high cards.
"It is our hope that this dispute will not impact other areas of our companies' working relationship," Mulloy told Betanews this morning. Right now, there may not be enough substance to that relationship for anyone to notice an impact.
6:10 pm ET February 18, 2009 - The seal has been lifted on Intel's suit against Nvidia filed Monday, and now we know more about the extent of the CPU maker's complaint -- some of which deals with the nature of Nvidia announcements as recently as last week.
"Intel...contends that Nvidia has not complied with its contractual obligations to refrain from making false or misleading statements to third parties about the parties' licensing agreements," the suit reads. "Nvidia has unequivocally told third parties, including customers of Intel and the trade press, that its agreements with Intel license it to market the Disputed Nvidia MCPs [media and communications processor products]. Nvidia's statements are false and misleading, and therefore in breach of the parties' licensing agreements, because, inter alia, they fail to acknowledge that Intel vigorously contests Nvidia's claim to be licensed.
"Nvidia has failed to comply with Intel's requests that it retract or correct statements declaring unequivocally that Disputed Nvidia MCPs are licensed. Nvidia has not accepted Intel's invitation to issue a joint statement providing complete and accurate information. Intel therefore seeks an injunction preventing Nvidia from continuing to make false and misleading claims to third parties that Disputed Nvidia MCPs are licensed. Intel further requests an order requiring Nvidia to provide complete and accurate information to third parties to correct Nvidia's prior misrepresentations."