With the increasing demand of AI computing power, the bottleneck of computing power and power consumption is difficult to solve. Compute-In-Memory（CIM）technology, a non-von Neumann architecture, is regarded as an effective method to make a huge break-through of the chip “Memory Wall” and performance bottleneck.
First generation chip
SRAM-CIM based AI chip with ultra-high computing power and ultra-high energy efficiency ratio
Based on SRAM-CIM technology, the next generation chip chip can provide 10 times power efficiency than existing products.
MRAM/RRAM-CIM based AI chip with ultra-high computing power and ultra-high energy efficiency ratio
Based on advanced MRAM/RRAM-CIM, the next generation chip can achieve a computing power of up to 1000T+ by continuously expanding the model capacity.