Abstract

Non-volatile magnetic RAM (MRAM) offers high cell density and low leakage power while suffering from long write latency and high write energy, compared with conventional SRAM. The use of hybrid memories (e.g., SRAM and MRAM together) can take advantage of the best characteristics that each technology offers. In this paper, we explore the 3D-stacked SRAM/MRAM hybrid L2 cache architecture by using a design-time optimization that determined each bank capacity and a ratio between SRAM and MRAM capacities. Also, this paper proposes a runtime cache management scheme that improves the system performance. Experimental results show that the proposed method yields, on the average, 61% performance improvement in terms of instructions per second (IPS) compared to the conventional SRAM-only L2 cache or MRAM-only L2 cache.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call