What can Golem do for you?
A review of several obvious (and less obvious) use cases for decentralized distributed computing
Decentralized distributed computing is a new field that Golem is pioneering, for the first time providing everyone access to distributed computing resources traditionally only available at great expense. Here are a few ideas about what you can do with this new power at your fingertips.
Make money from unused or under-utilized PCs:
This is the simplest and most common Golem use case is being a provider to the network itself. It’s amazing to realize that nearly all of the time that a given computer is switched on, the majority of its available compute cycles are not being fully utilized. Multiply that by every computer on the planet, and you can see why receiving money for this unused power is a nice opportunity. Check out how Julian describes his laptop usage for more information about this.
Avoid pulling all-nighters:
A student of internal design may create 3D models and visualization on a powerful laptop, but always she is faced with the same problem: When the final exhibitions are closed, her professors requires dozens of amendments. Even though she technically made all of the changes on time, the final images may take tens of hours to finish.
Golem may be a fantastic solution for her and her friends. She can accumulate funds during the year, by renting her laptop when she doesn’t use it. Then, during last month, spend them to rapidly accelerate the computations she needs for her projects. This applies not only to the students of art, but also architecture, computer graphics, informatics and computational science.
Change months in the lab into minutes on a laptop:
Chemical synthesis is a purposeful reaction where substrates are changed into an expected product. Retrosynthetic analysis is a technique that can be used to find and plan an optimized path of synthesis, hopefully leading to the creation of the desired product in the cheapest and most effective way.
For years, such work was done by a group of experts, laboriously searching chemical reaction databases, trying to find the optimal solution with trial-and error methods. New developments in retrosynthesis algorithms however make this problem much more computationally effective, often allowing the most optimal reactions to be discovered within a few hours. With Golem, the whole process can be even quicker, allowing the best path to be found in minutes! This can significantly shorten the time needed to begin the large-scale production of some substances.
Help create a material that can be the next biofuel:
In-silico design of new materials may be done with different algorithms, but often is simple brute-force that does the trick. Here all known and predicted structures are subjected to physical simulation, in order to identify the one with the expected properties, or to find relations between a given structure and property. Harvard’s Clean Energy Project is using a similar approach with advanced commercial tools, to find the next generation of plastic solar cell materials. There are also open-source projects that provide useful molecular applications such as RASPA2. Such applications and tools may be used to create a new porous material — very important in petroleum engineering. Golem again massively accelerates these processes.
Make your game AI invincible:
Creating the “perfect” AI for a game requires a surprising amount of effort in the form of changes to the algorithm, learning, and tuning parameters. A perfect example is Stockfish, one of the strongest chess engines in the world, and winner of Top Chess Engine Championship Season 6. It features a distributed testing framework called Fishtest, in which people may volunteer their computing processing power to test new versions of the code. The new algorithms plays against the old algorithm, and is accepted only if it is proven better to a statistically significant degree. As of today, Fishtestnet has noted over 300 million played games that took nearly 450 years of CPU time. These are obviously not numbers that can be achieved on a single machine, and nd of course, chess is only one example. Just take a look at all all the games and simulations that you can train with the OpenAI toolkit: There’s Go, Pong, Pac-Man, and DOOM, just to name a few.
Identify what people love or hate:
Text analysis with natural language processing (NLP) can work extremely well on Golem, especially due to the fact that text is smaller than other types of data. For example, the entirety of Wikipedia in its compressed form is only 12 GB. Not too bad for all of humanity’s knowledge… So, go ahead and feed Golem with your text data from articles, social media comments, and blog posts. Using sentiment analysis methods you can find out what people really think about your products: Do they write about them in positive, negative, or neutral ways? What exactly do they like and what are the problems that their describe? What are the trends? You don’t need an army of workers analysing this data by hand — just use proper algorithms that can be run on Golem!
Create your own mining easy-to-use mining pool:
Mining pools are the way that miners of “proof-of-work” cryptocurrency protocols unite their computing power in order to share in the profits. This case is tricky because it’s realtime task where miliseconds may be important. Therefore computation may optionally take place outside of the container (or virtual machine), removing an additional layer. This is one of the flexibilities of Golem, although not all types of tasks may be available to be done outside of the VM. In any case, the net result is that you don’t have to join external mining pool with Golem: you can have your own and people may take part in it — perhaps even those that would never mine cryptocurrencies themselves.
Solve a few mysteries of your DNA code:
Computational biology is a rapidly growing branch of science, and one of the most exciting aspects of it is the exploration of our DNA code. Yes, the human genome has been almost fully sequenced and the most important genes have been identified, but still there are long parts of it with purposes we don’t fully understand.
Golem’s lead engineer Ola once spent whole summer playing with specific parts of DNA and RNA, using different machine learning algorithms trying to create a solution to identify them. This kept three of her machines loud and busy through many nights. With a solution like Golem, she could have used several dozen machines instead of just three, and perhaps discovered the answers more quickly.
Manage inventory in your company:
But what if you’re not a scientist, student, developer, or artist? If you’re an entrepreneur for example, you can use Golem decentralized distributed computing to improve your business.
Perhaps you have heard about monte carlo methods: simple, effective and greatly applicable in the field of economics… Monte carlo is easy to distribute: just try different random samples on different machines, and sum up the results. Or how about option-pricing risk analysis before running new project? Inventory management in your bathroom tiles company? Why not to try to solve these problems inexpensively with new technology?
Help you with other fun and profitable ideas:
Maybe you want to create an art in a distributed way? See the Electric Sheep project, a distributed computing screensaver that generates new abstract art animations. Those animations can be used as a screensaver and the best one are sold as paintings or clothing patterns.
Another fun project was osmarender, run in a distrubuted way, thanks to tiles@home. This was a tool that allowed users to render specific maps using OpenStreetMap data, to create for example a simplified city map highlighting the distribution of postboxes, or perhaps marking houses that are available for rent.
Source: https://blog.golemproject.net