Amazon Web Services has released a new service called the The Cluster GPU Instance, a new kind of instance for the processing and rendering of the most complex actions such as those required in financial services and the sciences.
Almost every computer now has a GPU. It provides the capability to offload graphics from the central microprocessor and onto a specialized processor.
As Amazon's Jeff Bar points out points out, "the instruction set and memory architecture of a GPU are designed to handle the types of operations needed to display complex graphics at high speed. The instruction sets typically include instructions for manipulating points in 2D or 3D space and for performing advanced types of calculations."
The Amazon service provides those same capabilities in the cloud. The Cluster GPU Instance will work with Hadoop so it also shows how exisitng technologies will benefit from this new instance as well.
In the AWS blog post, Jeff Barr calls it a nuclear-powered bulldozer that's 1000 feet wide and costs $2.10 per hour to drive. Descriptive, huh?
Since the Cluster GPU Instance is in the cloud, we know inherent capital costs can be reduced.
The cost to do GPU clusters on site is quite expensive. It depends on creating a virtualized infrastructure. It's a cluster of machines delegated to do computations.
With Amazon Cluster GPU Instance, the customer receives the benefits of doing massive computations that are charged according to a unit of the GPU.
According to Barr's post on the Amazon Web Services blog, The Amazon GPU is similar to the Cluster Compute Instance. It has the following specs:
- A pair of NVIDIA Tesla M2050 "Fermi" GPUs.
- A pair of quad-core Intel "Nehalem" X5570 processors offering 33.5 ECUs (EC2 Compute Units).
- 22 GB of RAM.
- 1690 GB of local instance storage.
- 10 Gbps Ethernet, with the ability to create low latency, full bisection bandwidth HPC clusters.
According to AWS:
"Each of the Tesla M2050s contains 448 cores and 3 GB of ECC RAM and are designed to deliver up to 515 gigaflops of double-precision performance when pushed to the limit. Since each instance contains a pair of these processors, you can get slightly more than a trillion FLOPS per Cluster GPU instance. With the ability to cluster these instances over 10Gbps Ethernet, the compute power delivered for highly data parallel HPC, rendering, and media processing applications is staggering. I like to think of it as a nuclear-powered bulldozer that's about 1000 feet wide that you can use for just $2.10 per hour!ach of the Tesla M2050s contains 448 cores and 3 gigabytes of ECC RAM and are designed to deliver up to 515 gigaflops of double-precision performance when pushed to the limit."
This is Amazon's 11th instance type.
The AWS Cluster GPU Instance opens new frontiers for cloud computing. It raises questions about the model for developing GPU custom infrastructures within data centers when the real potential is to use the distributed nature of cloud computing and pay on a usage model.