Your browser is not supported. Please update it.

11 September 2014

Nvidia initiates GPU patent wars, sues Samsung and Qualcomm

Just when the mobile players seemed to be wearying of patent lawsuits, Nvidia has unleashed its first ever such attack – and is going for the jugular by targeting Samsung and Qualcomm.

Unlike most IPR lawsuits, this one will shake the smartphone and gaming sectors, because the potential implications affect many other players. While the immediate target is apparently Qualcomm’s Adreno graphics processing unit (GPU) technology, Nvidia – the grandfather of specialized graphics chips for PCs and gaming – also has its sights on all its competitors in the device world, including ARM and Imagination Technologies. These two UK firms dominate mobile processor IP, and on the GPU side, their respective platforms – Mali and PowerVR – are engaged in a tough battle.

Nvidia claims that two years of licensing talks with Samsung have ended in stalemate and so it has been forced to file suit against the Korean giant and Qualcomm, alleging that the latter’s chips, incorporated in a host of Samsung devices, violate seven of its patents. Its filings also cover Samsung’s use of Mali and PowerVR, not just Adreno.

Nvidia has filed its complaints in Delaware federal court and with the US International Trade Commission (ITC), a popular venue for recent patent wars since the agency has the power to issue injunctions against offending products (though rarely does in practice). It is asking the Delaware court to award unspecified damages, and the ITC to impose an import ban on the allegedly infringing products.

Its accusations center on Adreno, technology which Qualcomm acquired from AMD in order to build up its own graphics capability down to core level, rather than relying on third parties. Adreno is found in many Samsung offerings, including current flagship models like the Galaxy S5, Galaxy Note 4 and Galaxy Tab S. Qualcomm acquired AMD’s mobile graphics business when the latter purchased Nvidia’s then-rival ATI, while Samsung makes some GPUs inhouse, licensing both Mali and PowerVR for different models,  while also harnessing Adreno.

Specifically, Nvidia claims that Samsung and Qualcomm are infringing patents used in its Keppler GPU architecture, including US Patent 6198488, a graphics pipeline system, and US Patent 6690372, a “system, method and article of manufacture provided for shadow mapping while rendering a primitive in a graphics pipeline’. Others relate to programmable and unified shading and multithreaded parallel processing.

Samsung products included in the suits are all Galaxy models, including the brand new Note Edge and Note 4. Unlike some of the Apple/Samsung suits, which cited devices that were already ageing even at the start of the process, Nvidia is going for products which could still be current if a ruling is made within a year or so (though that would be unusually quick progress for a US patent suit, especially one involving Qualcomm’s famous legal teams). Other models included in the suits are the Galaxy S5, Note 3, S4,  Tab S, Note Pro and Tab 2.

The inclusion of devices that use ARM and Imagination cores (in Samsung’s own Exynos chipsets) is more complicated. Their Mali and PowerVR architectures are targeted in some of the citations against Samsung devices, but Nvidia is not suing them directly. This may relate to the terms of the contracts they have with Samsung and its suppliers (both provide IP to chipmakers, not actual processors), and a further twist must relate to Nvidia’s own relationship with ARM. Though it competes with the company in graphics cores – Nvidia now licences its GPU technology to certain third parties, rather than only keeping it inhouse – it is an ARM licensee on the CPU core side.

Patent lawsuits have several functions for large players, beyond straightforward protection of intellectual property advances. Some vendors are pursuing IPR licensing deals more aggressively as a revenue source, as mobile device margins fall. This applies to Nvidia, which set up a licensing business two years to offer its technologies to selected third parties, rather than using them only for its own products.

Others aim to hinder the progress of their rivals by spreading uncertainty among their partners and customers, and even securing bans on devices.

Another key motivation, which may be at play here, is to make a broad statement of technology leadership – a high profile PR tactic to remind the world of the firm’s significance in its core market. In its complaint, Nvidia puts heavy emphasis on its proud history in GPUs, reminding us that is was “the first company to put all functions necessary to graphics processing onto a single chip … critical steps in the development of the powerful mobile chips we have today. These GPUs enable consumers to run the same types of applications and graphics on their smartphones that they enjoy on their personal computers.”

It adds: “Qualcomm and Samsung are not GPU pioneers nor are they innovators in graphics technology”, and claims: “We have spent more than $9bn in R&D since 1993 when we began to build the best GPUs and the richest patent portfolio of graphics IP in the world. Our IP strategy is to earn an appropriate return on our investment by licensing our graphics cores and by licensing our patents.”

Nvidia dominated PC graphics and has been pushing its technology into high end platforms including supercomputers, and into next generation gaming systems, including its own Shield hardware. However, it has had less success in the mobile world, where Qualcomm processors, and the core IP of ARM and Imagination, dominate. Nvidia scored some early points in the tablet space with its Tegra mobile CPU/GPU, but it has struggled with the need, especially in smaller devices like handsets, to integrate as many components as possible – not just the graphics, but the modem too, a particular strength of Qualcomm. Nvidia acquired modem specialist Icera for this purpose, but earlier this year admitted that the resulting system-on-chip had been a commercial failure.

Nvidia CEO Jen-Hsun Huang told an analyst call: “We’ve been in discussions with [Samsung] a couple years because [their systems] include all three GPU architectures available outside ourselves. We made no progress, and they put no real offer on the table. We spent a lot of time with them and made every effort to have a negotiated outcome. When you have been using technology for free for a while it’s hard to sign a large, significant licensing agreement.”

An Nvidia representative said it expects the ITC to determine within 30 days whether to open an investigation. If it does a trial could come in mid-2015 with a decision following within a few months. The Delware court will set a date within 90 days for a trial that may not begin for two or three years, she said.