Raising hand as one of the suspicious number of very new commenters. :p As suspicious as I look, however, I don't belong to them and it's not my first post either.
This is quite interesting, though keep in mind that the RPi is talking to another machine to do its GPU computations. There are actually combination Nvidia GPU & ARM board like SECO's CARMA DevKit (http://shop.seco.com/carma-devkit.html) Which will actually run CUDA locally with out virtualization
So.. it's just a publicity stunt? If it's not actually using the graphics hardware on-chip then there's nothing special about it that's relevant to the raspberry pi. Apparently there exists software that can give you virtual gpu's, that's cool enough news for me, a bit weird that they market it through the pi.
But it's too expensive.
And as far as I know, the GPU virtualization breaks the barrier of the number of GPU cards inserted on the board. That is, if the library allows, we can dispatch the workload to as many as GPUs as possible.
Hey guys, you're low on memory on your php app. Now I can't check out your cloud gaming solution, maybe you should have hosted it in a cloud ;) (with more ram..)
Even full speed GPUs will be nearly worthless for bitcoin mining within weeks.
EDIT:
"Difficulty" determines the amount of computation required to create a bitcoin block. Hash rates have greatly increased lately and the difficulty automatically increases to maintain a constant block generation rate. Check out the charts here:
http://bitcoindifficulty.com/
This doesn't run anything important on the Pi. The only thing running there is a client that dispatches calculation jobs to some server with nvidia GPUs somewhere is the network.
You're right. GPU on any ARM SoC will never run full-fledge CUDA code (except Maxwell I guess), but the benefit of dynamic GPU resource allocation is still useful if we believe computing resource should be as flexible as storage in cloud environment