The Next Limiting Factor

It’s an interesting time in the semiconductor industry. Nodes continue to shrink, we’re on the verge of adopting a new type of transistor (finFET), and there’s also a shift away from planar CMOS – to name a few things on the horizon. What’s also extremely interesting is how design automation and semiconductor manufacturing technology continues to keep the pace with it all.

However, in both pre- and post-manufacturing another factor threatens to slow things down: the size of data files that the tools across the design and manufacturing chain must deal with. Case in point: in my article, “Dealing With The Data Glut” an EDA vendor noted an emulation database file it had received from a customer: the file was 500-gigabytes and represented a single frame of video. A single frame! While it is understandable considering what could be happening within a single frame of video (especially in a video game, for instance), the number is still mind boggling.

On the post-manufacturing front, scanning electron microscopy (SEM) files are just as larger or even larger. Used to study optical and electronic properties of semiconductor materials, the file sizes here can run into the terabytes.

So not only are tool makers all working to account for new process nodes, new techniques, and complexity – but they also must deal with the incredibly large file sizes that either slow the tools down to a crawl, or flat out won’t run.

EDA vendors are constantly working to make their tools run more efficiently on multicore machines and use advanced load balancing. Algorithms are also being developed to choose the right data to feed into tools to reduce the file sizes, but this is still not a ubiquitous practice. There is equivalent work happening on the post-manufacturing front.

The burning question is, will all of this work be enough? Can we keep up with the explosion in complexity and data sizes or will this be the next limiting factor to getting a chip into the consumers hands?

~Ann Steffora Mutscher