Bringing SOT-MRAM Tech Closer to Cache Memory
By Farrukh Yasin, Van Dai Nguyen, Siddharth Rao and Gouri Sankar Kar (imec)
EETimes (December 12, 2024)
For many decades, ultrafast and volatile SRAM has been used as embedded cache memory in high-performance compute architectures, where it resides very close to the processor in a multi-level (L1, L2, L3) hierarchical system. Its role is to store frequently used data and instructions for quick retrieval, with L1 being the fastest of all cache memories. SRAM bit density scaling has been slowing down for some time, and bit cells increasingly suffer from standby power issues.
The spin-orbit torque (SOT)-MRAM memory solution has several advantages, such as low standby power consumption, GHz-level switching or write speeds, negligible leakage, practically unlimited endurance, high reliability, and scalability. For these reasons, the industry is increasingly evaluating SOT-MRAM as a promising alternative to SRAM in embedded last-level, cache-memory applications.
To read the full article, click here
Related Semiconductor IP
- 100-10Gbps Ethernet MAC Controller
- SD Card Controller
- AXI4 Memory Controller IP Core
- AXI-Lite I2C Controller Core
- LVDS Transceiver IP core
Related Articles
- How to accelerate memory bandwidth by 50% with ZeroPoint technology
- Multi Chip, Multi Environment Simulation, Bringing Software Closer to Hardware and Saving Money
- Using scheduled cache modeling to reduce memory latencies in multicore DSP designs
- Which DDR SDRAM Memory to Use and When
Latest Articles
- LLM Agent-Based Design of RISC-V CPUs with FPGA-in-the-Loop Evaluation
- Automated Multi-Line Bug Repair Using Model Checker Aided LLMs: A Case Study on RISC-V Processors
- Cryogenic Recharacterization of a Commercial 22 nm FD-SOI Foundry Standard-Cell Library
- A 105 MS/s 4.18 nJ/inference 0.765 mm² Mixed-Signal Streaming ML Engine for Wireless Baseband in 12LP+ FinFET
- Agentic AI for Heterogeneous System Design Across Chip, Scale-Up, and Scale-Out Domains