Tab navigation
- Overview
- Virtualization
- Database
- ERP
- HPC- selected tab,
Systems optimized for high performance computing
Extreme computing performance without the extreme cost
GPU expansion options take advantage of existing BladeCenter and iDataPlex capabilities
Certain problems place extraordinary demands on computing resources. Engineering problems like fluid dynamics and crash simulation; scientific and medical applications such as seismic processing or image comparison; and financial applications like complex pricing models demand extreme computational capabilities.
But how do you acquire and manage these resources without installing specialized compute clusters?
IBM offers expansion options for existing systems built around NVIDIA’s next generation Graphics Processing Unit (GPU) Computing Modules. GPUs are now available for intensive, highly parallel, floating point processing tasks. These tasks – similar to pure graphics processing – can take full advantage of the compute architecture of the GPU. And because these options can be added to your existing systems, you can protect your current investment and lower your TCO while getting the additional performance you need for your demanding applications.
IBM BladeCenter GPU Expansion Blade
The IBM BladeCenter GPU Expansion Blade (BGE) can help you address your parallel compute needs while maintaining the reliability, flexibility and cost advantages of your BladeCenter infrastructure. Built around an NVIDIA Graphics Processing Unit (GPU), the BGE can improve performance of targeted applications by up to 150 times. And because the BGE attaches to a BladeCenter compute blade within a BladeCenter chassis, it takes advantage of the basic I/O and power resources of the BladeCenter system. You can stack up to four BGE blades with a single BladeCenter HS22 blade, to provide over 2000 gigaflops of processing power for your double precision applications.
IBM iDataPlex dx360 M3
The iDataPlex dx360 M3 features support for NVIDIA GPUs offering ten times the compute capacity of previous generation iDataPlex servers with 72 percent less power consumption. Taking advantage of the density, power and cooling technology of the iDataPlex system, the capability to add two GPUs per node offers extraordinary performance gains without compromising the iDataPlex advantages. And you can significantly increase the density of your compute solution by getting 1 TeraFlop of performance in a single node and 44 TeraFlops in a rack. The result is a fast, cool, flexible system that offers extraordinary performance and extraordinary total cost of ownership.
IBM Intelligent Cluster
IBM's Intelligent Cluster solutions embrace these new GPU offerings by simplifying and expediting the deployment of HPC clusters, while providing clients the benefit of a single point-of-contact for the entire cluster. Our integrated solutions are an outstanding choice for any organization that recognizes the economic advantages of reduced time to deployment but has concerns about the time and technical resources required for the end-to-end implementation.
