If you post the associated "Show me how my Representative Voted" link, please consider sponsoring the post that it shows up on your friend's feeds. The reason is that often these articles are repeated with multiple sources leading to them not reaching the importance criteria to display on other's feeds. It is too important of an issue to allow $7 in sponsorship fees to bury the message.

Maddog has brought these devices around to several conferences (I believe an AMD Geode?). It is a few kilowatts an hour for a mesh network. As Jon says, the idea is to take a few individuals in the area and train them to install the devices and ensure the backhaul experience is good. The subscribers in the area pay a small maintenance fee to the individual that provides the service, which is aiming to provide a network where the infrastructure isn't available. In other words, you are describing an Internet Connection per a single apartment building. Maddog is describing a single Internet Connection shared amongst hundreds with the available bandwidth pretty low for each user. It will allow for individuals to get to websites (designed for 56k) and some locally hosted instructional material that is on the mesh network. It isn't for Hulu or Netflix. Many of the users will only have cell phones with wifi for access. Jon is really passionate about this topic and he has done his legwork. It would not suprise me at all to see this project a success.

There is a healthy discussion here: Build the Enterprise Discussion

Essentially the crew would spend a majority of their time in a smaller shielded section of the craft including sleeping pods that are heavily shielded.

In OpenCL you need to copy items from the system memory to the GPU's memory and then load the kernel on the GPU to start execution. Then you must copy the data back from the GPU's memory at the end after execution. AMD is saying that you can instead pass a pointer to the data in the main memory instead of actually making copies of the data.

This should reduce some of the memory shifting on the system and speed up OpenCL execution. It will also eliminate some of the memory constraints on OpenCL regarding what you can do on the GPU. On a larger scale it will open up some opportunities for optimizing work.