ECE 5748
Memory Technologies and Memory-Centric AI Computing
The first half of this course will cover semiconductor memory technologies including SRAM, DRAM/eDRAM, and embedded non-volatile memories (RRAM, ferroelectric RAM, etc.). The lectures will encompass bitcell designs, array architecture, peripheral circuits, scaling trend, and target applications. The second half of this course will teach memory-centric computing, including SRAM/eDRAM/eNVM-based compute-in-memory designs and DRAM-based processing-in-memory designs, which have been demonstrated in both academia and industry in recent years.
Topics
- Students will be able to explain different memory technologies and corresponding applications.
- Students will be able to design memory bitcells and arrays, with an understanding on the trade-off of density, power, noise margin, reliability, etc.
- Students will demonstrate understanding of SRAM-based compute-in-memory and DRAM-based processing-in-memory designs, which emerged as energy-efficient solutions for AI hardware.
Official roster learning outcomes
Prerequisites & course information
ECE 4740.
Sources
- Course catalogue ↗2026-2027