[StarCluster] StarCluster AMI for New GPU instance?

[Syoyo Fujita]

Hello StarCluster-es,

I am interesting to build a cluster(MPI + CUDA) with StarCluster for
EC2's new GPU instance(g2.2xlarge), but this instance type requires
specific AMI to run.
Existing StarCluster AMI simply failed to run. Confirmed so far is

- Amazon Linux AMI (HVM) 2013.09.1 - ami-d1bfe4b8
- Amazon Linux AMI with NVIDIA GRID GPU Driver
https://aws.amazon.com/marketplace/pp/B00FYCDDTE

So I'd like to contribute public StarCluster AMI based on this, but
how can create a StartCluster-ready AMI?

I've granted document but couldn't find specific procedures to build
AMI for StarCluster.

Is it OK to just install libraries described in this page?

http://star.mit.edu/cluster/docs/latest/overview.html#starcluster-machine-images-
amis


Thanks in advance,

Syoyo