Mass Grid was born out of the need for high-performance computing from one of the founders, Huang Xiang. Working on a Virtual Reality project that requires the use of over 60 high-speed cameras that work simultaneously in order to recreate 3D models of the subject or situation, the need for a high-performance computer was evident. Each camera was linked to one computer, that was in turned interconnected to each other to create a pyramidal parallel computing network with three levels of processing to compress the data acquired by the high-speed cameras. To create such an extensive network, we discovered that the cost of it all was extremely high, especially for a start-up company. However, after finding funding and partnering with a research institute at the University of Zhejiang in China, the project is well underway and we are expecting it to come to fruition in the next year or so with the pace of research we are achieving. And it is from this situation, that the wheels were set in motion on how to reduce the cost for such an endeavor and make high-performance computing more accessible.
First, let’s take a look at what is high-performance computing. Everybody is familiar with what a computer is and if you are reading this chances are that you are the owner or have access to a computer (mobile phone included). Now, even if you have the latest MacBook and you think it is a fantastic computer, it is not considered supercomputer. To give you a graphic idea, the most powerful supercomputer in the world, the Sunway TaihuLight, approximately has the computational power of 55,000 PlayStation 4s. (Interesting fact, with a little research you can find out that in 2010, the US Air Force built its own supercomputer using 1,760 PlayStation 3s.) The main idea being that the cores must be interconnected in a massive parallel network to be able to achieve the incredibly fast computing power. It is easy to see the reason why they are called supercomputers. All that computing power is put to use on various tasks such as physical simulations, molecular modeling and climate research to name just a few of its applications.
And if you are familiar with us and our High-Performance Computing network project, the parallels between the virtual reality project, the Mass Grid project and HPC are getting quite evident. But still why would we go forward with this project?
With the emergence of big data, AI and more complex modeling and simulations, the need for computers that are more and more powerful is continuously growing. By creating, a decentralized high-performance computing network, the network would offer endless possibilities in terms of performance. It would also be more efficient than building one supercomputer as not all the research uses the maximum capacity of the computer. Hence, we will be able to assess the power needed for different projects and provide it via the network. And by doing so, we will be able to increase the efficiency in terms of exploitation of the network.
As the HPC market is set to grow to a multi-billion dollar industry in the next 5 years, it is also logical to give a better access to it. If you look at where supercomputers are located in the world, you can see that about 80% of the supercomputers are located in the USA, China, Japan and Western Europe. By liberalizing the access to HPC via our decentralized high-performance computing network, we will be giving HPC access to a lot more people at a fraction of the cost as they will not be obligated to purchase the hardware but only run time on the network. This could help advance research in several fields that the researchers don’t necessarily have the access or financing for a supercomputer.
At the moment, there are two types of high performance cloud computing platforms. One are the large corporations like Amazon and Microsoft professional services provided by professional companies. And the second type is more of an Uber, small cloud computing companies sharing hardware to one another. And that is where we are going with our project, we are building our network, to become an Uber easily accessible on demand high performance cloud computing. Once, the network is operational anyone with a computer will be able to connect to the network and get paid for usage of his computing power or submit an project to be performed. And there is computing power available out there, for example in 2017 there were 269.72 millions shipped worldwide and since the launch of the PS4 there were 70 million sold, that is a lot of computing power that is available.
We are hoping that this network will help bring a new era of decentralization and a larger segment of the population will be able to take advantage of the fourth industrial revolution.