According to statements from Samsung Electronics, the company has developed high-bandwidth memory (HBM), which has built-in artificial intelligence.
The new processing-in-memory (PIM) adds AI engines to Samsung's HBM2 Aquabolt , which the company first launched in 2018.
The chip, called HBM-PIM, doubles the performance of artificial intelligence systems while reducing power consumption by more than 70% compared to conventional HBM2, Samsung claimed.
The South Korea-based tech giant explained that this was made possible by installing AI engines inside each memory bank, maximizing parallel processing while minimizing network data traffic. By delivering this improved performance, Samsung said it expects its new chip to accelerate large-scale processing in data centers, high-performance computing systems, and applications for AI-enabled mobile devices
Kwangil Park, Senior Vice President of Memory Product Planning at Samsung Electronics, said: “Our groundbreaking HBM-PIM is the first programmable PIM solution tailored for diverse AI-based workloads such as HPC, training, and inference. We plan to leverage this breakthroughby further collaborating with AI solution providers for even more advanced applications supporting PIM.”
Most of today's computer systems are based on the von Neumann architecture, which uses separate processor and memory units to perform millions of complex data processing tasks. This sequential processing approach requires data to be constantly moved back and forth, resulting in a slowdown of the system, especially when handling ever-increasing volumes of data.
Instead, HBM-PIM brings processing power directly to where data is stored, placing a DRAM-optimized AI engine inside each memory bank – a storage module – enabling parallel processing and minimizing data movement.
Samsung added that HBM-PIM uses the same HBM interface as previous efforts, meaning customers won't need to change hardware and software to implement the chip in their existing systems.
The chip is currently being tested in customer AI accelerators, which are scheduled to be completed within the first half of the year. Samsung is also working with customers to build an ecosystem and standardize the platform. The company's paper on the chip will be presented at the virtual International Solid-State Circuits Conference next week.
