I see that you will be providing GPU based computing, this makes sense, but I'd like to understand what constraints this puts on the language used. My preference would be to use Pascal as the language. FPC, the Free Pascal Compiler, is supported on Linux, and it supports OpenCL, which can use GPUs. Will the environment support this? Alternatively, what is the preferred language / framework / API for this exercise? Or, to put it a different way, is there a list of supported computing options?

Created by Peter Brooks fustbariclation
Hi Peter, You can following the instructions on the page [Submitting Models](https://www.synapse.org/#!Synapse:syn4219927/wiki/231827) to start building your own Docker container. Please let us know if there are instructions that require more detailed information and we will update them.
Thank you Thomas. That's very useful - I look forward to the webinar next week to learn how to set it all up - unless there's an 'idiot's guide' for me to follow. I've downloaded the test images, but, presumably, they are available in a directory in docker.
Welcome to the DM DREAM Challenge Peter! Your docker container has access to one NVIDIA Tesla K80 GPU that appears to the system as two GPUs (NVIDIA Tesla K80 = 2x Kepler GK210). Your docker container is like a Virtual Machine with the OS of your choice (Ubuntu, CentOS, etc.). You can install all the software that you want in your container as you would do on your computer. From there, you just have to make sure that the framework that you select can communicate with the above NVIDIA graphic cards.

Language options - API & GPU preferences page is loading…