Skip to main content

Intel and Nvidia cosy up to one another

The Intel versus Nvidia court date has been removed from the Delaware courthouse calendar and is yet to make a reappearance, reports the WSJ.

Queue the 'enemy-of-my-enemy' out-of-court dealings and corresponding soundtrack.

According to the Journal, Nvidia's court date has not only been removed, but prospects of hammering out a settlement are very high.

It appears that with Intel now battling mutual rival AMD on another front, Fusion, Nvidia could well be in a position to lend the chip behemoth a hand.

As AMD is on track to be the first company to ship APU products, something that is probably keeping the competition awake at night, Intel now seems more likely to come to an agreement with Nvidia and put pressure on AMD's upcoming APU platforms.

Nvidia has been at odds with Intel since February 2009 when the latter sought legal action to bar the company from producing chipset designs for Core i-series processors, alleging that a prior license agreement did not extend to the new chip architecture. This has left Nvidia out of the chipset game for the past two years - while the company has had to make do with pushing dated 9000-series mGPUs, ION and Tegra as their source of revenue in that segment.

Now, a mutually beneficial compromise seems to hang in the balance if, in fact, analyst predictions are correct. Speaking to Uchi Orji, from UBS Securities, the WSJ claims that Intel is inclined to license off its QuickPath Interconnect IP to Nvidia in exchange for parallel computing and GPU know-how from the Green One.

This sounds vague at best and, putting aside wishful thinking about corporate takeovers, we have to wonder just what are the implications of a cross-licensing deal between Intel and Nvidia?

For Nvidia this would have a huge impact. It would have a free hand at making chipsets capable of running Intel Core i-series processors and crack open the OEM market once again, which no matter how small a slice it can have now, it'll still be better than having no kit at all. Nvidia would stop looking like a dead-end for PC makers and the company would be back in good standing with Apple. That would hurt AMD even further.

Nvidia could still sell its 2010-version of the NF200 bridge chip, as well as develop full chipsets for Intel processors which could also have modern mGPUs based on Fermi (which should end up in Apple iMacs, for example).

On the Intel side of things, it would feed on undetermined Nvidia know-how that would help it build up its own GPGPU and parallel computing technologies. Both Intel and Nvidia have focused lately on developing parallel computing but, as we know from Intel's MIC (Many Integrated Core) architecture this is the base for future Intel CPU products as it provides greater scalability and power efficiency than current CPU designs.

Intel must have also done the maths by now, just by looking at AMD's financials. The graphics business is booming - thanks mainly to Nvidia's original problems with Fermi - and right now represents a sizeable amount of AMD's business. Compare the chipset sales that Intel was losing to Nvidia at the best of times against AMD's graphics business and AMD is still raking in money.

So, helping Nvidia hurt AMD will be the logical course of action. Of course it sounds a bit cheap and dirty, but it should work out to Intel's benefit.

Both Intel and Nvidia are yet to comment on a suggested Anti-AMD alliance.

Does such a deal going through mean Intel and Nvidia are best buds?

No, as they will still compete to shut each other out in the ARM vs. x86 market, but it would make life a whole lot easier for them. It will however push AMD into the mainstream/low-end where it has been cosy the last few years because the majority of people simply can't afford everything Intel and Nvidia want to push into the market. monitors all leading technology stories and rounds them up to help you save time hunting them down.