On Tuesday, IBM Research announced that its scientists have developed the first “in-memory computing” or “computational memory” computer system architecture.
For those unaware, In-Memory Computing refers to the storage of information in the RAM of dedicated servers rather than relational databases which operate on slower disk drives. It is an emerging concept that aims to replace the conventional “von Neumann” computer architecture, used in standard desktop computers, laptops, and cell phones.
IBM scientists claim new In-Memory computing architecture speeds up the computers more than 200 times. It is expected to yield 200x improvements in computer speed and energy efficiency — enabling ultra-dense, low-power, massively parallel computing systems for AI (Artificial Intelligence) applications.
Their concept is to use one device, i.e., PCM (phase change memory, a type of computer RAM that stores data by changing the state of the matter) for both storing and processing information, unlike in traditional von Neumann computer architecture, which divides the computation and memory into two different devices and hence requires moving data back and forth between memory (RAM) and the computing unit (CPU), making them slower and less energy-efficient.
IBM has announced that it has created an unsupervised machine-learning algorithm that runs on one million PCM devices. “The result of the computation is also stored in the memory devices, and in this sense, the concept is loosely inspired by how the brain computes,” said Dr. Abu Sebastian, a scientist, and IBM Research.
The further details on IBM’s current efforts in-memory computing are explained in a paper published in the scientific journal Nature Communications.