Heterogeneous Computing + Quantum Engineering - EEs Talk Tech Electrical Engineering Podcast #17

Learn about parallel computing, GPUs, and quantum computers! Subscribe! ► http://bit.ly/Scopes_Sub ◄ Electrical Engineering Podcast: https://eestalktech.com

Agenda:

0:40 When processor speeds stopped increasing, microprocessor companies started to increase the quantity of cores on an IC

1:44 Now, the increased use of special processing cores is more popular

GPUs are examples of this. A GPU is different from an x86 or ARM processor.

GPUs handle vector math and matrix math. Originally, they were for processing pixels. But, they use floating point math to calculate pixel values.

A GPU is very useful if you have a number of identical operations you have to calculate at the same time.

4:00 GPUs were daughter cards, but recently GPU manufacturers have released low power parts for embedded applications. They have several traditional cores and a GPU. 

Now you can build embedded systems to take advantage of machine learning algorithms that would have normally require too much processing power thermal power. 

4:50 These are called heterogeneous processors or heterogeneous processors. A heterogeneous processor has different types of cores, and a software architect figures out how to distribute tasks.

6:00 We will start to see heterogeneous processors with multiple types of cores very soon.

Traditional processors are best for algorithms on integer + floating point operations where there isn't a benefit to doing more than one thing at a time.

A GPU is good for multiple parallel computations, so it's useful when there aren't tight dependency chains.

Neither is good at real-time processing. If you have real-time constraints there is a lot of computing required. So, a new type of hardware is necessary. Right now, ASICs and FPGAs fill that gap like we discussed in our other electrical engineering podcast episodes.

9:50 Quantum cores could be on processor boards eventually. Quantum computers that can beat traditional computers will be introduced within the next 50 years, and as soon as the next 10 years.

11:50 Quantum computing is reinventing processing. Quantum computers are going to be best at handling new and different types of problems.

12:50 There is a catalog on the web of problems that would be better for a quantum on a computer than a traditional computer.

13:30 People are creating algorithms for computers that don't exist!

The Economist estimated that the total spend on quantum computing research is over $1B per year globally. A large portion of that is generated by the potential of these algorithms and papers.

Quantum computers won't completely replace traditional processors.

15:00 Lee - the quantum computing industry is still speculative, but the potential is so great that neither large computing companies nor industrialized coun ... https://www.youtube.com/watch?v=p0TwhacA0AI