It all started with mainframes in the first wave, moving to the personal computing and software era with the wave two and later to the internet with wave three.
The world has changed in the last few years and computing has got everywhere from mobile phones - now a crucial computing device - all the way to many IoT devices sitting on the edge.
Data, 5G and processing
Simon believes that generating data, transporting data via 5G and processing data will drive the fifth wave of computing but interpreting the data actually makes sense to users. This is where Machine learning and Artificial Intelligence come into play.
Computing is everywhere today. Not so long ago, computing device were on everyone's desks, then recently moved to everyone's pocket. Today a computer is in everything with big emphasis on IoT devices all around us. Smart thermometers have computing power, as well as wearables, and many other tiny connected devices.
Simon pointed out that the power efficiency of ARM powered devices made customers rethink where hardware lives and what software does.
Raw compute power today is not the only predictable thing. Today it is not just about gigaflops, or GHz as the market looks at the collective computing power of systems.
There is always a need for more compute power but there is a need to think to end to end systems, he said. Every phone or IoT edge of the network device needs connectivity and connectivity gets that device to the cloud when a customer opens a picture. This is just one example of everyday end to end compute.
Simon pointed out that we generate a lot of data but the future belongs to those who make sense of data. One alarmingly large number surfaced in his keynote. Future self driving cars can generate 20TB per hour and it won't be viable to send all that data to the cloud for processing. Some on device processing must find a place in future computing.
5G created around things to connect
5G is transformational technology and services will start early next year, in selected places. 5G is not only how to move data faster, it won't be just about higher resolution videos, it will also enable new services. The 5G network was designed based on the number of humans using it, and it resolved some design constraints of previous 4G networks that didn’t plan mass of IoT devices.
The 5G design paradigm relies on the foundationial idea of how many things we need to connect, and the answer is a very large number indeed.
Big opportunities are expected with 5G, not just better smartphones but also lower latency and better efficiency for many connected devices.
The network evolved over time and 3G brought a music stream and download ro businesses while 4G introduced us to faster speeds and enabled video streaming and communication business. Just think about what the future for, Uber, car sharing, Facebook, and many other applications that you use everyday out of your pocket.
It is hard to predict the next big 5G business opportunity but every G showed a huge transformation.
Data that all of these devices generate will have to be processed and this is why everyone - ARM included - is investing in AI and ML. You need to upload and sent only the parts of data that are crucial, otherwise you would clog even the large capacities of 5G. This is the processing part of the equation. It can be one of many parts of the cores, as we got to learn from ARM’s Jem Davis CTO VP, Fellow and GM, Machine Learning Group that even a tiny M0 core can do some basic AI and be used for voice activation or basic image processing.
ARM expects that its customers will create the unexpected.