Near-Memory Computing
- (FitzRandolph Gate, Princeton University - Kimberly Chen)
- Overview
In-memory computing (IMC) is a technique that runs computer calculations in computer memory, such as RAM. IMC uses physical processes like current summation and charge collection to perform numerical operations. This can accelerate common computing tasks like matrix-vector multiplication.
IMC is based on two main principles:
- Data storage: The way data is stored
- Scalability: The ability of a system, network, or process to handle growing amounts of data
IMC solutions can provide real-time application performance and scalability by creating a copy of data stored in disk-based databases in RAM. When data is stored in RAM, applications can run up to 1,000 times faster because data doesn't need to be retrieved from disk and moved into RAM before processing.
IMC can improve complex event processing and deliver faster reporting, product releases, and quicker decisions. It can also improve the user experience and customer satisfaction.
Some use cases that need high-speed processing and benefit from IMC include: payment processing, fraud detection, predictive maintenance, algorithmic trading, self-driving cars.
Some disadvantages of IMC include:
- Volatility: If the system crashes or loses power, all data in memory can be lost
- Cost: Memory is more expensive than disk storage
