← 返回 JSSC 论文列表JSSC 2014第11期Power Management130nm
A 0.79 pJ/K-Gate, 83% Ef ficient Uni fied Core and V oltage Regulator Architecture for Sub/Near-Threshold Operation in 130 nm CMOS
提出一种嵌入计算子系统的电压调节架构,显著提升超低功耗平台能效。
83%效率(0.52-0.6V输出范围),系统级能耗降低44.8%
电压调节器近阈值计算能量效率域堆叠FIR滤波器
▸创新点1:多电压域堆叠技术(电路创新) - 通过动态堆叠多个电压域,实现在近/亚阈值区域的高效能量转换,系统能效提升44.8%,输出电压范围0.52V至0.6V时效率达79%-83%。
▸创新点2:核心交换机制(系统创新) - 采用动态核心交换策略,优化计算负载分配,显著降低系统级能耗,验证了能量模型的准确性。
▸创新点3:计算-电压调节器一体化设计(架构创新) - 将信息处理子系统嵌入能量传输子系统,实现超低功耗平台的高效协同,相比传统SC-VRM系统节省能源。
▸创新点4:能量模型与仿真验证(方法创新) - 提出并验证了C-VRM的能量模型,通过实测数据与仿真结果匹配,证实了设计的可行性和准确性。
Abstract
This paper presents the compute voltage regulator module (C-VRM), an architecture that embeds the information processing subsystem into the energy delivery subsystem for ultra-low power (ULP) platforms. The C-VRM employs multiple voltage domain stacking and core swapping to achieve high total system energy ef ficiency in near/sub-threshold region. Energy models for the C-VRM are derived, and employed in system sim- ulations to compare the energy ef ficiency bene fits of the C-VRM over a switched capacitor VRM (SC-VRM). A prototype IC incorporating a C-VRM and a SC-VRM supplying energy to an 8-tap fully folded FIR filter core is implemented in a 1.2 V , 130 nm CMOS process. Measured results indicate that the C-VRM has up to 44.8% savings in system-level energy per operation compared to the SC-VRM system, and an ef ficiency ranging from 79% to 83% over an output voltage range of 0.52 V to 0.6 V. Measured values of the and match those predicted by system simulations thereby validating the energy models.