Rambus officially announced a few days ago that they have settled their patent dispute and have agreed to a license agreement. This agreement closes the final dispute between Rambus and another technology company. In the past, Rambus has been responsible for years and years of litigation and has been generally considered across the IT industry as one of the worst patent trolls in existence. This is primarily because a lot of the patents that they held and hold have never been productized and their existence was purely to force other companies to pay them licensing fees. Personally, if a company wants to be able to force competitors to license their patents, they have to make a reasonable effort to sell the technology itself.

The only two commercial technologies that Rambus has ever made that actually made it into final products were RDRAM and XDR, but both technologies had incredibly short shelf lives and were ultimately buried due to high costs. Even so, Rambus successfully sued many of their competitors and eventually won settlement agreements with companies like Nvidia, Hynix, Nanya, Elpida and virtually any company on earth that deals with memory in one way or another. However, Micron has held out and refused to settle with Rambus. The details of the deal are that Micron will license Rambus’ technology for 7 years at a cost of $280 million and will settle the patent dispute between the two companies, closing a chapter in both companies’ histories. Furthermore, this agreement will also extend to Elpida, the memory manufacturer that Micron recently purchased.

In a recent conversation with some Rambus representatives at ARM’s TechCon we were able to get a better understanding of the company’s new direction and their goal to not be seen as a patent troll anymore. Rambus wants to become more of technology enablement company that also assists the industry in creating new products and technologies rather than forcing them to license a certain tech. This new direction came directly from the company’s CEO who believes that the company needs to move away from their patent troll image and to work towards becoming a more respected technology institution. And I can say that this is certainly a good thing, considering the general consensus about Rambus in the past with jokes about how the company supposedly employs more lawyers than engineers and the sort.

We got to take a look at some of Rambus’ newest technologies, which include their Brazos chip (yes, the same name as AMD’s) which apparently doesn’t exist anywhere on the internet other than in this very article (comes as a shock to me). It was explained to me that Rambus’ Brazo chip is part of a solution that Rambus wants to offer to memory manufacturers to help them qualify their memory and to improve performance and latency to degrees that enable much higher frequencies and better stability. You can see it pictured below with some Elpida memory soldered into the board. This solution is squarely targeted at improving lower power memories like LPDDR3 and is unsurprisingly called R+ LPDDR3 DRAM and is designed to improve upon already existing memory to deliver better power and performance.

They also have some new memory technologies that they are working on that are designed to be complementary to existing standards and are designed to improve the performance of memory overall. This technology was specifically targeted at DDR4 and beyond and is designed to show us what we could expect from future memory standards before the current future ones even land on the market. This is yet another R+ technology and can be called something like R+ DDR4 and lowers the voltages to near ground voltage, which becomes a pretty significant problem as DDR memory continues to lower in voltage. 

They are also working on image sensor technology, in order to allow for multiple exposure photography (HDR) to occur at much better performance levels and to increase the overall dynamic range of image sensors. While I don’t know how they will compete with companies like Sony, it will still be interesting to see if they can get any design wins with their new image sensor, clearly targeted for mobile. They call this technology their Binary Pixel technology and claim that it solves a lot of the problems mentioned above.

They also showed some of their latest tools in regards to DPA (Differential Power Analysis) countermeasures, which is basically a series of tools and services that Rambus’ security division offers in order to help companies, governments and individuals assess where they could potentially be vulnerable through a signal analysis. While their demonstration was certainly an interesting one, there is very little beyond their own documentation to prove that this is a widespread problem and can always be cracked using this method. Nonetheless, the idea that someone could crack into your phone without ever actually connecting to it directly is certainly a scary one. Below, they had their tools showing a Samsung phone running a cryptography program running RSA keys and how they were able to analyze and break it down.