Xen + NV-CUDA: a new possible way?

Xen + NV-CUDA: a new possible way?

[OneSoul]

Hello all!

I'm a Xen user and I'm very satisfied by its power features.

Recently, reading some technical articles over internet, I have 
discoverd the big potentialities of the NV-CUDA framework in relation to 
the scientific and graphic computing that takes strong advantage from 
the most recent GPUs. Someone has used it for password recovery, 
realtime rendering, etc, with great results.

It would be possible to use this technology in the Xen project to 
achieve better performance? (NOT for VGA paravirtualization...)
It could be a significative step for the develop in virtualization 
technology?

Someone, in experimental way, has (re)wrote, for example, the md-raid 
kernel modules using the CUDA framework to accelerate some features... 
and it seems that works fine.
Why not for Xen, or related technologies, including dynamic translation?
If yes, could be a good manner to achieve more performance or a way to 
drammatically off-load the CPU taking advantage from the impressive 
parallel computing capabilities of GPUs.
Think at xen-kernel that execute separetly some kinds of task between 
CPU and GPU, like a super-cluster that separates the pure computing from 
the I/O computing.

What do you think about this draft idea?

Stephen Spector said that "this is a great idea"


Any feedback or discussion is welcome...

Xen + NV-CUDA: a new possible way?

[OneSoul]

Hello all!

I'm a Xen user and I'm very satisfied by its power features.

Recently, reading some technical articles over internet, I have 
discoverd the big potentialities of the NV-CUDA framework in relation to 
the scientific and graphic computing that takes strong advantage from 
the most recent GPUs. Someone has used it for password recovery, 
realtime rendering, etc, with great results.

It would be possible to use this technology in the Xen project to 
achieve better performance? (NOT for VGA paravirtualization...)
It could be a significative step for the develop in virtualization 
technology?

Someone, in experimental way, has (re)wrote, for example, the md-raid 
kernel modules using the CUDA framework to accelerate some features... 
and it seems that works fine.
Why not for Xen, or related technologies, including dynamic translation?
If yes, could be a good manner to achieve more performance or a way to 
drammatically off-load the CPU taking advantage from the impressive 
parallel computing capabilities of GPUs.
Think at xen-kernel that execute separetly some kinds of task between 
CPU and GPU, like a super-cluster that separates the pure computing from 
the I/O computing.

What do you think about this draft idea?

Any feedback is welcome...