A new game-changing memory tech that could supercharge AI is on the cusp of going mainstream — rivals assemble at major event to discuss IGZO DRAM as in-memory computing inches towards reality

A representative abstraction of artificial intelligence
(Image credit: Shutterstock / vs148)

Researchers are experimenting with a number of different forms of memory which might be better suited for AI, with indium-gallium-zinc-oxide (IGZO) the latest to grab attention.

IGZO-based 2-transistor 1-capacitor (2T1C) technology is typically found in display screens, but research organization imec has identified its potential for Analogue In-Memory Computing (AIMC). 

The AIMC approach addresses the limitations of traditional digital computing, specifically speed and energy efficiency, by executing computing tasks within the memory itself using analogue technology. This minimizes power consumption and accelerates computational speed.

Denser memory array

The key advantage lies in the parallel processing and storage of data in analogue format within the memory, which offers a quicker, more efficient, and energy-conserving mode of computation. In essence, the memory itself becomes part of the computation process, negating the need for data transfers between separate units.

IGZO DRAM cells hold tremendous promise for analogue in-memory computing due to their significantly reduced standby power consumption. Additionally, IGZO transistors can be processed in the chip’s back-end-of-line (BEOL), allowing for placement on top of the peripheral circuit located in the front-end-of-line (FEOL). This results in a denser memory array with no FEOL footprint. 

At the recent 2023 International Memory Workshop (IMW) imec teams addressed some of the remaining challenges, strategies for optimizing the retention time of the gain cell, and demonstrations of successful MAC operation in an array configuration.

It wasn't the only firm to discuss the technology either, as Samsung also shared its research there.

You can read more about the subject on the imec site, but the researchers conclude that that IGZO-based 2T1C and 2T0C (a variant without a capacitor) gain cells show exceptional properties for AIMC. Compared to traditional SRAM-based technology, they offer superior energy efficiency and computational density for machine learning applications, particularly during the inference phase. The 2T0C cells excel even further in area efficiency.

More from TechRadar Pro

Wayne Williams

Wayne Williams is a freelancer writing news for TechRadar Pro. He has been writing about computers, technology, and the web for 30 years. In that time he wrote for most of the UK’s PC magazines, and launched, edited and published a number of them too.