Solving the Lack of Cache Memory in Logic Processors with TopoLogic

TopoLogic is a Japan-based startup that focuses on developing high-performance semiconductor memory and high-speed heat sensing technology using unique physics of topological material. As one of the winners of the IC Taiwan Grand Challenge Batch 4 in the AI Core Technologies and Chips category, TopoLogic’s TL-RAM technology matches the speed of SRAM, with DRAM-like capacity.

AI Growth is Rapidly Outpacing Current Memory Technology's Capabilities

AI's explosive growth has spurred a massive increase in demand for both memory and computing power. While existing memory technology was sufficient to fulfill the requirements of AI development and deployment, gaps started to appear between what it can do and what it needs to do. Conventional solution to this gap has been to increase the hardware memory, either in capacity or in quantity. However, the limited supply of memory and storage hardware alike will mean this stopgap solution will not be viable for the long term and will be increasingly untenable for most hardware manufacturers or users, regardless of whether they engage in AI, potentially stalling innovation and worsening data-center energy use. TopoLogic aims to solve this issue with their TL-RAM technology, addressing the lack of cache memory in logic processors which allows both sub-nanosecond speeds and GB level capacity.

Introducing IC Taiwan Grand Challenge

To strengthen Taiwan's position as a global semiconductor powerhouse, the National Science and Technology Council (NSTC) organizes the IC Taiwan Grand Challenge. Started in 2024, the competition originally focused purely on innovative IC designs and advanced application solutions. Today, ICTGC also honors teams in the fields of AI Core Technologies and Chips, Smart Mobility, Smart Manufacturing, Smart Medtech, and Sustainability, which can support and contribute to the semiconductor industry and Taiwan’s industry in general. TopoLogic was selected as one of the winners of ICTGC Batch 4, featuring their TL-RAM technology in the AI Core Technologies and Chips category.

Topological Material to Solve Technological Bottlenecks

TopoLogic was founded in 2021, focusing on the development of high-performance semiconductor memory, and high-speed heat sensing technology using unique physics of topological material. They developed TL-RAM, a semiconductor memory technology that drastically reduces power consumption by leveraging the unique properties of topological materials, including its unique band structure that results in unique effects in the field of electronics and spintronics, with major discoveries since 2015 and with ongoing research to discover more benefits from the technology.
TL-RAM utilizes non-collinear topological antiferromagnetic materials into conventional MRAM structures to greatly enhance the write speed performance and write current requirements compared to conventional MRAM. The spin of non-collinear antiferromagnets can be manipulated with pulses less than 0.5ns, allowing for sub nanosecond write operations, with less than 1/10 write current requirements compared to conventional MRAM based on CoFeB.
TL-RAM solves the issue of lack of cache memory in logic processors, with memory cell footprint that is 10x smaller than conventional cache memory SRAM cells, due to circuit simplicity and less transistors per cell. This allows a significant increase in the memory capacity that can be added to the same area in a logic processor. When TL-RAM is incorporated to cache memory, processor chips will be able to include several hundred MB to GB level cache memory capacity on chip, significantly reducing the memory bottleneck of conventional processors.