I've been on a Billy Joel kick of late, and this song is apropos to this post, too! (Please visit the site to view this video) Okay, now that I have given you a nice ear worm, we'll go on with today's post. “Water links us to our neighbor in a way more profound and complex than any other.” —John Thorson As we established in my recent post about power and neighborhood resource infrastructure , we need to design our neighborhoods with an eye to efficient energy flow. Now let’s talk about water. Picture this: You’re taking your morning shower, you’re all soaped up, and shampoo is in danger of getting into your eyes… and someone in your house flushes the toilet. Aieee! The pressure drops, and you get scalded! Now, look at a neighborhood. What if you live next to a water hog, say, a power generator, and there’s not enough water pressure for you to water your garden—or take a bath? This is a problem. To ensure this doesn’t happen in the first place, city planners might make a model of the neighborhood before building it, to test not just electricity flow, but all the resources required to make a neighborhood function properly. It’s a lot easier to fix a model than it is to bulldoze the entire neighborhood to start over again. Using a model, the city planners might identify the water aggressors, optimize placement of areas just for the farmers and the millers and the power generators (Google it—these are the top “thirstiest” industries). This would be separate from the lower-usage areas, such as residential or businesses. Using an implementation and analysis tool, such as the Voltus IC Power Integrity Solution and the Innovus Implementation Solution together, you can do the same thing on a piece of silicon, as I wrote about last week. Designing chips is also fraught with challenges in optimizing a design that doesn’t exist yet. It’s awfully expensive to make a chip only to discover the problems of power distribution—problems with power aggressors and victims, signal integrity, bugs, hot spots of all kinds—and end up having to start all over again. The Cadence Voltus and Innovus solutions work together to analyze, debug and optimize the design while verifying it fits its specifications, all before the design is signed off to go to fab. So how do you deal with the residential home that has a hot tub? And what happens in the unlikely event that everyone in the neighborhood flushes their toilet at the same time? Unlikely, perhaps, but if everyone is watching the Super Bowl and takes a break at halftime, it may be more likely than you think. And what happens when there is a drought (Californians know what I’m talking about) or a fire (some Californians also know what this is like) or a flood (hmm, Californians know about this, too) (God, why does anyone live in California?). Chipoppolis Let’s look at a neighborhood; let’s call it Chippopolis. After the Chippopolis city planners have their neighborhoods zoned as they wish them (they have used Voltus to analyze their initial design to discover the hot spots, and Innovus has placed and re-placed the buildings for optimum flow), the planners may simulate the conditions of stress: they want to measure how the droughts, floods, and other times of stress affect Chippopolis. In this image, the water main is coming from the left to the right. The water aggressor is the cluster of buildings in the lavender colored background, creating a dearth of water pressure in the apartment buildings below it, in the amber block. The lavender and amber blocks form a hot spot. If all the buildings in the hotspot cannot be moved (which would have been discovered using the Voltus analysis), there are three possible solutions to this water aggressor problem: 1. Increase the water flow to everyone from the reservoir. Pros : Everyone has enough water with enough water pressure. End of story. Cons : This may waste a lot of water on low-usage days, and may encourage leaks and inefficient usage. Also—you can’t always rely on a plentiful source (said the Californian). In semiconductor world, Voltus + Innovus analyzes power consumption on each block and gives designers a way to adjust power usage. This is referred to as “low-power design”—that is, enough to power what needs to be powered, but no power is wasted. 2. Make giant switches to turn the water off to the aggressors during low-flow days. For example, keep the water flow high during the week and lower it during the weekends. Pros : This saves water in total, and saves water when it is not needed. Cons : When the water is turned back on Monday mornings, it takes a lot of water to “wake up” the system. There is a huge “in-rush” of water, flooding the entire system, and the infrastructure must work extra hard to accommodate that ramp-up. This may take time and may take just as much water to wake the system up as it did to turn it off for the weekend. In semiconductor world, Innovus installs the switches, and Voltus analyzes the ramp-up and in-rush of current while optimizing the locations of those switches. This is called using different “power domains”. You can shut down a domain to cut power to a block (“sleep mode”), thus saving electricity. When a trigger wakes up the block, the in-rush current and wakeup time is analyzed (Voltus) and the location of the switches in a power domain may become perfected (Innovus). 3. Build water towers, or localized water sources, to support the victims of the water aggressor. These water towers store water from the reservoir and are placed near to the water victims. This is a local solution that is triggered when the water pressure drops under a certain threshold. The water is replaced in the tower when the water needs are low. Pros : This is an excellent solution for locations that need constant water, such as hospitals (every neighborhood needs a hospital, and it can’t be placed in the industrial zone across town). Cons : Water towers may fix the problem, but they are expensive, offer too many opportunities for leaks, and may prevent the rest of a neighborhood to operate at peak capacity. In semiconductor world, Voltus determines the best places and the proper sizes for the capacitors (that is, the water towers)—also known as de-coupling capacitance optimization—and analyzes the system for leaks and performs the cost/benefit analysis; Innovus makes the adjustments based on Voltus’ recommendations. The Goal of Optimization Each of these solutions has their reasons to be used. The point is to optimize the power flow and power usage using whatever means are available: Optimizing placement of each cell in each block and each block in each neighborhood Re-calculating the optimum flow of water to the entire neighborhood (low-power design) Optimizing the switches to manage the flow of water efficiently (power switching) Optimizing the placement of any water towers (decoupling) These solutions result in a system’s power usage to operate at peak capacity, just like our well-managed and well-maintained Chippopolis. When the city planners get the simulation of the system to a point that all the houses and buildings in all the blocks have enough resources to function properly, they can send their plans to the builders, who can break ground on the new neighborhood—and hope that no one decides to build an Olympic-sized pool in their backyard. And if they do anyway? At least the neighborhood is prepared, thanks to the partnership of the Voltus IC Power Integrity Solution and the Innovus Implementation Solution ! —Meera![]()
↧