When will Faster be stable?
Wright now, there is only one developer involved in the project, so we cant predict it.
Will Faster have a Python wrapper?
We plan on it. Although, we are not sure about the performance implications of running user functions in Python, as it is required by most of dataset functions.
Will Faster support CUDA or OpenCL?
We plan to do so, although it is not a priority and it is not clear how it would go.
Will Faster support Apache Mesos or Hadoop Yarn?
Eventually yes, although we can not promise that it will. In order to support them a great part of the framework would have to be rewritten. Because of that, it would have to come later on.
In the About Section you mention heterogeneous environments. Do you mean exactly?
For now, we mean multiple machines with heterogeneous performance and the same architecture. Although it does not mean that we will not support it in the future. Right now it is too much work to be done in order to even support a single architecture.
Since Mesos uses CMake, will it be multi-platform?
Faster being multi-platform is not our main concern. We will focus on Linux servers, Ubuntu LTS in particular. Though it doesn’t mean that it cannot work on other platforms. Notice that Faster uses OpenMP and advanced C++11 features in its core, therefore it needs a modern compiler that support those functions like GCC. You are welcome to try using it on other platforms. Let us know if you find out anything interesting.
Are there any scientific papers about Faster?
Not Yet. As soon as the core functions are implemented and we are able to implement K-means, SVN or graph community detections algorithms efficiently we will publish our findings.
Are there any high level libraries/tools, for Graph processing or Machine Learning implemented using Faster?
Not yet, but you are welcome to implement one! :D Right now we are focusing in the framework basic infrastructure implementation. Soon we will have at least some ML algorithms implemented as examples.
