#RAY
《SOL Ecosystem Prosperity----Ray Worth Attention》
RAY, another excellent target in the SOL system, is a distributed computing framework developed by the University of California, Berkeley. It has many advantages.
First of all, it has the smallest cluster configuration, can run on a smaller cluster, and has high performance, suitable for heavy computing workloads, and a single node is about 10% faster than Python standard multi-process.
It uses a common underlying framework, allowing multiple machines to run Python code in parallel, build and run various distributed applications.
It also has high scalability and is used to expand a variety of machine learning libraries, unlike the Spark framework.
It has a unique actor-based abstraction, multiple tasks work asynchronously, high utilization, and more flexible than Spark synchronous execution. It also supports a variety of technologies, perfect support for common machine learning libraries, and built-in GPU resource allocation support.
It is simple to use, Python first, pip can be installed, ray cli can be used to build a cluster, import programming in Python, and define functions in distributed programming with @ray.remote annotations for asynchronous execution. Automatic serialization and deserialization, low overhead. Automatically distribute code and dependent libraries to facilitate parameter sharing. ray.put is placed in distributed storage to avoid transmission overhead. Automatic detection and deserialization are used as parameters.
It is not suitable to chase the price up now. If it falls back to around 2.440, you can enter the first position and cover the position around 2.250. Pay attention to RAY, another bright heart in the SOL system.
