Historically, circuits and advances in lithography have been used in every generation as the approach to enhance SOC memory performance. Memory performance has evolved essentially as a function of memory clocks speeds, and as such has not improved significantly over the past five years. In contrast, processors continue to improve about 30 percent each year. Memory performance has not kept pace, and remains a bottleneck leading to the classic 'processor memory' performance gap. Furthermore, SOC designs need memories that are area efficient and consume less power.


To meet the needs of today's SOCs, for high performance memories, with lower power and area requirements, Memoir Systems, a provider of breakthrough memory technology, has pioneered a completely new approach - Algorithmic Memory.


Algorithmic Memories use the power of algorithms to increase the performance of existing embedded memory macros by up to 10X more Memory Operations Per Second (MOPS), while reducing area and power consumption. These memories provide deterministic performance for truly random access memory operations. For applications that have identifiable patterns of memory accesses, Memoir's pattern-aware technology (built on top of Algorithmic Memory) creates highly optimized memories with up to 16X performance, while offering very pragmatic area and power footprint.


"The company's Renaissance family of products provides drop-in replacements for existing embedded memories. The technology used is optimized for a particular process, node, and foundry and integrates seamlessly into any existing SOC design flow," says Sundar Iyer, CEO, Memoir Systems.