Visual Memory v. Nvidia – Federal Circuit – Aug. 15, 2017

In Visual Memory v. NVIDIA (Fed. Cir. 2017), Visual Memory appealed the dismissal of their infringement complaint against Nvidia.  The district court dismissed under Rule 12(b)(6) because the district court found Visual Memory’s asserted patent to cover patent ineligible subject matter under Section 101.

The patent at issue is U.S. Patent No. 5,953,740 owned by Visual Memory.  The federal circuit reversed, holding that Visual Memory’s ‘740 patent is an improvement to computer memory systems and not directed to an abstract idea. This case has excellent language for responding to Alice rejections.  Interestingly, the court is split in this case – Justice Hughes found abstractness and no inventive concept under the Alice test.

The ‘740 patent is directed to an improved memory system that supports multiple processors.  The ‘740 patent explains that systems commonly use a three-tiered memory hierarchy to enhance performance:  (1) low cost, low speed memory like magnetic disk; (2) medium speed memory that provides the main memory; and (3) costly high speed memory that provides processor cache.  The cache is typically smaller due to cost and cannot store all the required data.  During execution, data transfers from the main memory (2) to the cache (3) to provide quick access to the data. Replacement algorithms determine which data is transferred between main memory and cache.  Prior art memory systems were built around specific processors making them costly because each processor required a different memory system.  The ‘740 patent gets around this by programmatically tailoring to different processors.  The ‘740 patent discloses a main memory and a cache controller with an internal cache, a pre-fetch cache, and a write buffer cache.

When switched on, the processor type information is used to self-configure the programmable operational characteristics of the memory system. Separating the functionality for the caches and defining those functions based on the type of processor in this way, improves performance of the system and of the main memory.  The main memory of the ‘740 patent is faster than the prior art fast page mode because the memory is divided into pages containing either code or non-code data, and the system provides a bias towards code pages or non-code pages depending on the processor.
The ‘740 patent allows different types of processors to be installed with the same memory system without compromising on performance.

The district court granted Nvidia’s 12(b)(6) motion holding that under Alice step one, the ‘740 patent is directed to an “abstract idea of categorical data storage” which has been practiced for years, and found no inventive concept because the claimed components were generic and conventional and because the ‘740 patent failed to explain the mechanism for accomplishing the result.

The court stepped through the Alice analysis for patent eligible subject matter.  Step one is to “determine whether the claim is directed to a patent-ineligible concepts”.  If so, then the court proceeds to the second step of analyzing whether the claim elements, either individually or as an ordered combination, contain an “inventive concept” that “transform the nature of the claim’ into a patent-eligible application.”  Beginning with step 1, the court explained that importance of articulating with specificity what the claims are directed to.  The court cites Enfish where claims reciting self-referential table for a computer database were held patent-eligible under step one of Alice test, and the court also cites Thales Visionix where claims reciting a unique configuration of inertial sensors and the use of an equation to calculate the location and orientation of an object relative to a moving platform were held patent-eligible under step one of Alice test.  In Enfish, the claims were patentable because the claimed self-referential table achieved faster searching and more effective data storage.  In Thales Visionix, the Federal Circuit sided with the patentee because the patented system achieved greater accuracy than the prior art.

The court held that the claims of the ‘740 patent were directed to an improved computer memory system rather than to the abstract idea of data storage because a memory system with programmable operational characteristics defined by the processor connected to the memory system permits “different processors to be connected to the memory system without compromising their performance.”  The prior art involved a trade-off in processor performance.  This trade-off is side stepped in the claimed invention of the ‘740 patent.

Turning to step two of the Alice test, the court held that the claims were also directed to a technological improvement of an enhanced computer memory system.  The court then went on to distinguish this case from Contract Extraction (Fed. Cir. 2014) and TLI Communications (Fed. Cir. 2016) which did not involve improvements in computer functionality.

Finally the court rebutted Nvidia’s arguments.  In one argument, Nvidia unsuccessfully argued that the the claimed programmable operational characteristic were “nothing more than a black box.”  However, the court noted that enablement under Section 112 is different from patent eligibility under Section 101.  The court also emphasized that the use of conventional computer components in the ‘740 patent is not fatal to patent eligibility where the claims improve the functioning of a computer.

As mentioned earlier, it is important to note that Justice Hughes disagreed with the majority and argued that the ‘740 patent failed both step one (abstract idea prong) and step two (inventive concept prong) of the Alice test.

Opinion here:
Visual Memory v. NVIDIA (Fed. Cir. 2017)

Leave a Reply

Your email address will not be published. Required fields are marked *