The Problem With Clocks


The synchronous digital design paradigm has enabled us to design circuits that are well controlled, but that is only true if the clocks themselves are well controlled. While overdesign techniques ensured that to be the case in early ASIC development, designs today cannot afford such luxuries. As we strive for lower power and higher operating frequencies, the clock has become a critical desig... » read more

Managing Voltage Drop At 10/7nm


Power integrity is becoming a bigger problem at 10/7nm because existing tools such as static analysis no longer are sufficient. Power integrity is a function of static and dynamic voltage drop in the power delivery network. And until recently, static analysis did an effective job in measuring the overall robustness of PDN connectivity. As such, it is a proxy for PDN strength. The problem is ... » read more

Intelligent Power Allocation


The modern System-on-Chip (SoC) has higher thermal dissipation than its previous generations, because of the following factors: Increasing processor frequencies. Decreasing SoC package and device sizes. Higher levels of integration. Static power consumption trends with the most advanced SoC fabrication[1]. Faster frequencies mean faster switching, which means more power consumption... » read more

Pushing Batteries Too Far?


Reports of battery fires in consumer devices are not abating. The culprit in almost all cases is the lithium-ion battery. In some cases, this is a manufacturing issue, where predictable intervals of failure can point to a breach in the membrane separating the anode and cathode or a metal particle contaminant that causes a short circuit. Those kinds of flaws are well understood, based upon ho... » read more

Self-Driving Cars Rattle Supply Chain


Automotive compute workloads are consolidating as carmakers push toward autonomous vehicles, but the changes necessary to make this all work are causing huge disruptions in an industry that has fine-tuned its supply chain over more than a century. Consolidation is essential for a variety of reasons, including efficiency of the computations, complexity management, and lower deployment costs. ... » read more

HBM Upstages DDR In Bandwidth, Power


For graphics, networking, and high performance computing, the latest iteration of high-bandwidth memory (HBM) continues to rise up as a viable contender against conventional DDR, GDDR designs, and other advanced memory architectures such as the Hybrid Memory Cube. [getkc id="276" kc_name="HBM"] enables lower power consumption per I/O and higher bandwidth memory access with a more condensed f... » read more

TFETs And/Or MOSFETs For Low-Power Design


As discussed in Reducing Subthreshold Swing With TFETs, papers at December’s IEEE Electron Device Meeting examined a variety of potential designs for tunneling transistors (TFETs). That focus continued at the recent CS International Conference. In particular, Nadine Collaert discussed IMEC’s work on InGaAs homo-junction devices. Many compound semiconductor devices depend on heterojunctio... » read more

TFETs Cut Sub-Threshold Swing


One of the main obstacles to continued transistor scaling is power consumption. As gate length decreases, the sub-threshold swing (SS) — the gate voltage required to change the drain current by one order of magnitude — increases. As Qin Zhang, Wei Zhao, and Alan Seabaugh of Notre Dame explained in 2006, SS faces a theoretical minimum of 60 mV/decade at room temperature in conventional MO... » read more

The Ultimate Shift Left


Floorplanning is becoming much more difficult due to a combination of factors—increased complexity of the power delivery network, lengthening of clock trees, rising levels of communication, and greater connectedness of [getkc id="81" kc_name="SoC"]s coupled with highly constrained routing resources. The goal of floorplanning is to determine optimal placement of blocks on a die. But connect... » read more

The Return Of Time Sharing


As early as the 1960s, it wasn't uncommon to hear that transistors would be free. Those were pretty bold statements at the time, considering most computers in those days cost $1 million, required special rooms, and budding computer scientists usually had to sign up to use mainframe computers for one-hour time slots—often in the middle of the night or on weekends. Still, those predictions ... » read more

← Older posts