Saturday, 1 September 2018

Memory Processing Unit Could Bring Memristors to the Masses


Memristors: Computers of the Future

Today’s world is all about doing things fast. We want our phones to work faster, our computers and even our toasters. So, scientists are continually on the lookout for the next big thing that will make computer and the like run faster. One such thing discovered by scientists is a memristor. If you’ve not heard of it, then its no wonder as the term was coined just recently or with the discovery itself.

Memristors not only make your computer or phone work faster but will also cut energy consumption,like you wouldn’t believe. Memristor is a way of arranging advance computer parts on a chip so that it performs faster and with less energy consumption.

Where will Memristors be used? 

Memristors, according to scientists, will improve performance especially in those low power environments such as your smartphone. It can also be used to make an already efficient thing even more efficient like in the case of supercomputers.

How do Memristors work? 

Semiconductors in the industry make things fast by ensuring faster processing but when it comes to receiving and sending data, that is the part that takes time as these semiconductor processors have to work with other parts to do it.

Memristors is a solution to this problem. Named as combination of memory and resistors, which you may have already figured, memristors can process and save data in memory all in the same place, which will significantly speed up calculations and such.

How are Memristors different from Traditional means: 

Traditional means use bits of 1s and 0s but in the case of memristors, they work on a continuum.
The team behind memristors use smaller blocks to break down a large mathematical problem which improves the flexibility and efficiency of the system. These smaller blocks are called “memory processing units” and can be useful in implementing machine learning and artificial intelligence algorithms.

They can also be used in areas of simulation such as in predicting the weather. Mathematical problems in the form of rows and columns are directly imposed on the grid of memristors. Operations that multiply and sum the rows and columns in the table are then done simultaneously.

Whereas in the case of a traditional processor, it would perform mathematical calculations such as the sum and multiplication of rows and column individually taking a lot of time and energy. While with memristors all this happens in one step itself.

Using Memristors in Practical problems: 

Many science and engineering problems are very difficult to solve because of their complex forms and numerous variables needed in their models. Memristors can be used to simplify such problems and model them in the correct manner to get the right answer taking up much less time and with that energy.

When it comes to partial differential problems, solving them exactly is near impossible. To get an approximate value itself is the job for supercomputers. These types of problems involve loads of data and getting a memristor to sit in a supercomputer and perform these calculations will save up a lot of time and get results much faster.

No comments:

Post a Comment

Note: only a member of this blog may post a comment.