'Network-on-a-chip' : A note on its Genesis

We have been observing the boom of the "e-era", where the number of transistors per unit area and the corresponding speeds is increasing.

At the same time, we have to accept that we're gonna experience a maximum limit of speed with the materials we have now..

But hey, don't disappoint yet, we got a good news for you !

Well, the solution is not the discovery of a new material, but is inspired from the 'reliability' concept of our 'large-world' communication networks..

In short, our networks work like this: They don't want a 100% error-free communication; They use a network of switches and routers, our signals takes many different paths, and experience random losses..

Designers allowed for errors to occur; In turn, they set up error-correcting circuits, that can quickly correct them to a level that we cannot notice their effect.. (it's all about 'reliability' my friend..)

Now, the main problem of our chips is "Interconnects"..

The chips, that have those micro-level circuits(yeah, now they're making it nano-level) are a combination of different circuits that have a set of fixed paths in-between, referred to as the interconnects.

At higher frequencies, transmission line effects and other losses cause the logic level to degrade on its way so that logic 1 reached as logic 0 and vice versa..

If you guessed the solution, you're right..!!

Up to now, the design focus is all about ensuring "perfect communication" among the chips..!

Yeah, so we simply ditch that idea..

Similar to our large-scale systems, we will not "fix" the paths; instead we build a whole network of routers, switches and corresponding error-correction circuits; So, now, can we increase frequency of operation ?

This concept is referred to as "Network-on-a-chip"..

(Are you thinking of increase in size due to this set-up? C'mon, with our nano-scaled devices, the net difference we observe in size would be negligible..)

Thanks for reading, feel at home to express your thoughts on this post, don't forget to let us know in case of any mistakes...

If you also got something exciting to share, do express it in comments so that you can help us reach the 'Eternal Thing'..

Source: Digital Integrated Circuits - A Design Perspective (You can download it here)

Comments

Top 5 take aways from "Linux 101 Hacks" - The Bloggers League 2022

Author: Rama Subrahmanyam Hello, how are you doing..? We all know that Windows is a super cute-looking OS, but software people will connect to Linux; It offers much flexibility with file management, and plus, it's open-source too..! So, the book - Linux 101 Hacks, is a nice intro for beginners, by Ramesh Natarajan . Having said that, we look at the top 5 hacks from the book. It is a free e-book, and you can download it here . 1. 'alias' for most used commands:- There will be commands that are used repeatedly, for which we can have a shortcut using an alias. For example, for cd commands, we can have alias cd1="cd .." alias cd2="cd ../.." alias cd3="cd ../../.." alias cd4="cd ../../../.." alias cd5="cd ../../.....

Bye-bye strict timetables...! - The Bloggers League 2022

Author: Rama Subrahmanyam (ramasubrahmanyam.m@gmail.com) When we are planning for multi-tasking, we divide our time and allocate it for each task. That is a great thing, as it gives a chance for incremental growth, eventually getting compounded... Scheduling makes us punctual; sometimes forces us to be punctual if someone is counting on us, say an examination, office meeting, so on... Besides strict things, some activities should happen at planned times, like sports, gym, etc. But, we still have things, that can be done in our free time - say reading books, solving puzzles, or learning something of our passion. Should we also maintain the timetable in those cases...? Well, may not be that necessary... We get used to a task at the same time if we have a strict timetable; In a way, can also affect our peace... Suppose due to urgent work, we missed a task - We may develop a bad feeling, ...

How can we become narrow-minded people without realising it?

In science, whenever we observe a novel event, we start sharpening our brains to decode it. We make a couple more observations related to that event, and try to get a reason for it. It was also the same with the previous generation of scientists. They observed various new phenomena and they gathered as much information as possible to get the accurate reason behind them. But there was a problem with the human mind. In the past days, when there was limited information at hand and more time was required for getting new information, scientists had to make the best possible theories (i.e., explanations) with whatever data was available. It was like a competition, where the scientist with the best theory would win. This "winner" scientist, should be able to explain the data from future experiments with the theory. Consider that, a new experiment happened later, but the results were not explained by the theory of our scientist, then the theory was to be changed, and again the compet...

The Eternal Thing

Search This Blog