The Golem Project is a project based on blockchain technology whose objective is to create a “distributed supercomputer”, which takes advantage of the computing power of all those who are part of its network.
The idea behind the project is nothing new, but it is very revolutionary, and instead of relying on a centralized server farm, Golem distributes the computational load to a crowd, which is called “Suppliers”. This means that anyone in the network can rent their computing power and receive a payment in Golem tokens (GNT).
Seen one way is a round business, since most of the time computers have their CPUs in complete leisure, a computing power that can be directed to other tasks and for which, the owner of said CPU can receive an income , exploiting in a good way a resource for which payment was made. This is how Golem is making use of idle computing time, in a business that has multiple applications in the world.
The project was initiated by Julian Zawistowski, Piotr Janiuk and Andrzej Regulski in November 2016, when they presented their whitepaper to the world, with the aim of creating a hybrid platform IaaS (Infraestucture as a Service) and PaaS (Platform as a Service).
How does Golem work?
Contrary to what they might think when they see the term blockchain, Golem does not do mining, but its software orchestrates all the machines in its network, to perform tasks that need a huge computing potential, dividing that task into all the ” Suppliers “with which account, hence the name of distributed supercomputer, with which the project is identified, whose main use cases range from realistic simulations of physical and chemical phenomena, weather forecasting, photorealistic rendering, machine learning and artificial intelligence.
All this technological framework is possible thanks to the Ethereum network, on which Golem built its bases, and whose token is only part of the many ERC-20 tokens that exist. The development team behind Golem, chose Ethereum for the following reason:
We have evaluated alternatives, but we have decided to host our Transaction Framework, the future API, among other things, in Ethereum, because it provides all the technical characteristics and flexibility we need; so there is no need to change this.
Additionally, the Golem team presents information on its process of updating, developing and making decisions in a very transparent manner, indicating that throughout the ecosystem there is a shortage of developers that forces them to prioritize the features that will be presented in new versions.
System rewards will be handled separately from the token to ensure the integrity of the token and transaction payments. This will work as follows: All transactions in the Golem network will be subject to a fee, approximately 5% of the total value of the transaction. This will include both transactions, between users and transactions between users and software developers. The resource that will be handled in this network will then be: computing power.
Transaction Framewok – Golem’s biggest innovation
The developers of Golem from the beginning understood that the platform offered endless possibilities to other developers and companies that wanted to use the Golem network. In view of this, they decided to create the Transaction Framework, a payment platform built on Ethereum to facilitate the implementation of transaction schemes according to the needs of the developers of apps that use the infrastructure of Golem and Ethereum. This system will allow the developer to have the following unique characteristics:
Schemes of micropayments and batching. Payments in off-chain channels to accelerate transactions. Payments to software developers directly. Payment per unit use, that is, the ability to pay per hour, per node, per core, among others.
Current status of the System
At this time, there are many more providers than applicants who use Golem. The network has an impressive dashboard where they can check the status of the network in real time, showing that they currently have 2019 cores, more than 3 TB of RAM and more than 3 TB of hard disk. To have a clearer idea of the enormous amount of resources available to the Golem network, taking into account that the most powerful supercomputer in the world so far is Summit, located at the OAK Ridge National Laboratory in the United States. The following specs:
600 GB of Ram (Coherent and DDR4)
800 GB of Ram located in the GPU support subsystem (use GPU to improve the computing power of the system)
9216 CPU cores located in 22 POWER9 packages designed by IBM
Undoubtedly Summit has some impressive specs, but Golem is not far behind and gives us a clear vision of what can be developed with this technology.
Continued development
The development of Golem has not stopped, and its official GitHub, has a great development activity. The last stable version of Golem is the version, 0.18.3 called Brass Golem, whose main features are:
Schema definition of basic tasks. Registration of basic applications, since it is the first version of the Application Registry based on Ethereum that allows to save defined tasks with a definition scheme of basic tasks. Integration of IPFS to coordinate the delivery of data and content of tasks, for example, deliver the necessary files to calculate a task, return the results to the applicant.
Docker environment with images provided by Golem for calculations sandbox, facilitating service deployment. Local verification: a probabilistic verification system based on the calculation of a fraction of the task in the applicant’s machine. Basic UI and CLI. Basic reputation system Implementation of Blender and LuxRender tasks.
According to your roadmap, there are three major versions yet to be developed, which are Clay Golem, Stone Golem and Iron Golem, which greatly expand the functionality of the system, and although there are no defined launch dates for these major versions, the team of Development is building the necessary foundations for these advances.