Taking advantage of opportunities in the cloud is not easy for small companies and young startups. New systems arrive to level the playing field.
For customers who haven’t gotten around to installing that massive server farm, Penguin Computing has introduced Penguin on Demand, or POD. POD is built using Linux clusters with specialized hardware interconnects and software configurations for High Performance Computing (HPC).
Designed for end customers and service providers, POD provides a high performance, scalable, on-demand HPC computing. POD can be accessed from a workstation or it can be used as an extension of an existing HPC cluster. POD gives customers direct access to compute cores rather than relying on machine virtualization, which is used by other companies offering pay-as-you-go cloud computing.
The idea is that customers can share time on the POD. Jobs can be submitted and customers can see the progress on their job. (It’s interesting, if totally obvious, how this harks back to the days when time on mainframe computers was leased. At least everything is much faster and the computers more dependable.) In situations where customers require their own, always available, and secure capabilities, Penguin’s POD offers dedicated compute nodes. These nodes are reserved — they sit idle when not being used by the customer. Dedicated cores are provided on a monthly basis.
In addition, POD has GPU compute nodes for GP GPU jobs. The GPU Compute Nodes include a dual Xeon X5540 4-core processor (2.53 GHz), 24GB of memory (3GB per core) and 3 Nvidia Tesla C1060 Computing Processors. Penguin charges for GPU Compute Nodes on the basis of per server hours and not on a per core hour.
So, one of the things customers might want to do with HPC is render and access huge amounts of 3D data, right? Most recently Penguin and Mental Images have announced the availability of RealityServer 3D on POD. The RealityServer takes advantage of Nvidia’s Tesla GPUs and includes mental images’ 3D Web technology to enable visualization and collaboration. As a result, customers can access, navigate, interact with, and adjust the lighting for large models and environments.