This paper describes a 0.25-/spl mu/m, 64 K/spl times/36 bit pipelined burst SRAM using a 6.156-/spl mu/m/sup 2/ cell. It realizes over 500-MHz operation using a lower cost double metal process, Internal 16 K/spl times/144 organization by T-shaped bit line array reduces 20% of latency, 20% of active power, and 8.5% of die size. The low power also enables us to use lower cost thin quad flat type packages. Our solution to the soft error problem, a shallow triple well structure and four-transistor cell with stacked capacitor, improved soft error rate by 3.5 orders of magnitude compared with the conventional SRAM cell.
No takes yet. Share an insight, caveat, or question.
Sato et al. (1999) studied this question.
Synapse has enriched 5 related papers. Consider them for comparative context: