Frequently Asked Questions

Distributed computing is not simple. That is why DeepSci exists. Hopefully these questions and answers will further your understanding.

1. What is BOINC?
   
   BOINC is open source software from the University of California at Berkeley. It stands for Berkeley Open Infrastructure for Network Computing. It provides the basic infrastructure for hosting a project. DeepSci has created software to further optimize and control the environment to extend scalability and eliminate bottlenecks.
   
   
2. Why use DeepSci to run open source software?  
   
   You can certainly host your own BOINC server, many projects have. The learning curve to do so is rather high, depending on your existing familiarity with Linux, shell scripts, PHP, CGI, C++, MySQL, Apache, networking, authentication, and etc. At DeepSci, we believe a researcher shouldn't be required to have all of these skillsets available on their research team just to do distributed computing. Also, DeepSci uses additional components that assure reliability, and scalability. These do not exist in BOINC and often present problems as projects attract a growing user base.
   
   
3. How much volunteer computing time should I expect?
   
   The one thing about volunteers is that they are not entirely predictable. But many projects find that if they adequately explain their research objectives, the interest in volunteering idle computer time exceeds the work they have to be done. Indeed, this is how BOINC was born. SETI@home had more volunteer computing time then they could effectively utilize. By creating BOINC and the subsequent creation of other projects, they found a way to share this valuable resource to other research teams. In the BOINC community, 1,000 hosts is considered a small project. There are over 500,000 active hosts at any given time working on publicly known projects.
   
   
4. What is the maximum TFLOPs when using DeepSci?
    
   There is no limit, and we constantly work to stay ahead of demand to assure that growth is as seamless as possible to everyone involved. Many BOINC projects are driving more then 30 Tera FLOPS on a sustained basis. The costs to run the project will depend on the scale of required bandwidth, and server resources required, not the scale of the total compute time utilized. This is part of why it is difficult to make assumptions and prices for a standardized package.
   
   
5. What requirements are there on the science application?
   
   One of the advantages of using DeepSci's approach is that the application is written in a way that runs on personal computers rather then tightly coupled, proprietary systems such as clusters or supercomputers. So, you can run the program more universally without concern that a rewrite will be required to take your project's scale to the next level. You can support any or all of Windows, Mac, Linux, and CUDA environments.
   
   
6. Does DeepSci write the application for me?
   
   Generally no. You are the expert in the field required to write the application. So most researchers bring an existing application along with their need for computing time. But we invite the challenge if this is the service you require to be successful.
   
   
7. What if my application cannot be run in parallel?
   
   In general, there are ways to break out a problem to take advantage of a large array of computers. Sometimes you have to change how you are thinking about the problem to be solved. It is often possible to apply math principles or creative thinking to reinvent the problem definition in a way that affords application of distributed computing.
   
   One familiar example is that of climate prediction. How can you predict the climate for the year 2108 if you have not yet completed the prediction for 2107? One solution is to have each of the thousands of computers proceed on historical data with a unique set of assumptions and then see which assumptions yield models that do not boil or freeze the oceans. So rather then attempting to distribute each year of prediction, you assign each machine the entire task, with the specific set of assumptions to use in the model.
   
   
8. My research is valuable intellectual property, won't I be giving that away using DeepSci?
   
   Any time your modelling/reduring/simulation application reaches the public, there is a risk of someone taking the time to reverse engineer it, infer your research approach, and complete your research before you do. But most likely by the time they have completed the reverse engineering, you are on to a better version and left them in the dust. However, if privacy is required, DeepSci offers creation and management of private clusters as well.
   
   
9. BOINC is free, how can DeepSci charge for using it?
   
   DeepSci does not charge for using BOINC. It charges for hosting and managing the servers and bandwidth that run your project, and for customizations to the existing open source that are required to properly process your research.
   
   
10. How do I get started?
    
    We start the process with a free consultation. This assures we understand everything you require to be successful and for us to provide the level of service that will make you proud to recommend us to your respective research communities and colleagues. It starts with a click to Contact Us.