DEV Community

Cover image for Why are there no 128-bit CPUs?
Justice-hub
Justice-hub

Posted on

Why are there no 128-bit CPUs?

Why does it seem like this evolution has stopped there? What do these bits represent? Will there be 128-bit processors in the future?

What are bits?

In the simplest terms, bits are the basic unit of information in computing. They can be either on or off, and they're what make up everything from individual letters to entire websites. Most digital devices these days use binary code, which is a series of ones and zeroes that represent those on and off states. So when you see a "1" or "0" online, that's a bit in action.

Computers use large amounts of these little guys to do all sorts of things - like store data, run programs, or communicate over the internet. And because there are so many different ways to string them together (or not), that means there can be an almost infinite number of combinations for creating new words, pictures and ideas. So the next time you're online and see something cool (or weird), remember that it was created one bit at a time!

Why are there no 128-bit processors on the market?

There are many reasons why CPUs do not come in 128-bit varieties. The most obvious reason is that the benefits of a wider bus width would be outweighed by the increased cost and complexity of manufacturing such a chip. Additionally, there is no real demand for such a product in the market; 64-bit processors offer more than enough performance for most users. Finally, even if there were demand for 128-bit CPUs, current software would not be able to take advantage of their capabilities.

In closing, i want to remind you that today there are solutions beyond 64 bits, but they are not classic CPUs, but specialized components to perform specific activities and which work on particular data sets. In addition, there are logical solutions that allow you to work on 128-bit or higher data simply by exploiting modern architectures to create multi-bit virtual systems. This confirms that physical chips with over 64 bits are not really needed to meet modern technological challenges.

Top comments (2)

Collapse
 
armousness profile image
Sean Williams

The only real reason to have 128-bit registers and buses is for SIMD, so there are processors with 128-bit support (like the IBM Cell), but I think they're only really used in scientific applications. After all, if you're not in computer graphics, when was the last time you did a vector-add?

Collapse
 
phlash profile image
Phil Ashby • Edited

A number of complex processors (x86 included) will operate with 128+ bits internally, especially if you count the parallel ALUs, out-of-order instruction dispatch, cache widths... but the instruction set architecture (ISA) usually sticks to 64bit registers apart from SIMD (as noted by Sean hereabouts) since there is little demand for wider compute capability and the knock-on impact to software tooling would be expensive in time/effort for a very small consumer base. There is then the packaging challenge of that many bits in parallel @ GHz speeds... external CPU buses are using more serial technologies these days (like PCIe).

[edited to add] Transmeta had an x86 compatible CPU that actually translated to 128bit VLIW instructions: en.wikipedia.org/wiki/Transmeta