The GPU was invented by NVIDIA back in 1999 as a way to quickly render computer graphics by offloading the computational burden from the CPU. A great deal has happened since then—GPUs are now enablers for leading edge deep learning, scientific research, design, and “fast data” querying startups that have ambitions of changing the world.
That’s because GPUs are very efficient at manipulating computer graphics, image processing, and other computationally intensive high performance computing (HPC) applications. Their highly parallel structure makes them more effective than general purpose CPUs for algorithms where the processing of large blocks of data is done in parallel. GPUs, capable of handling multiple calculations at the same time, also have a major performance advantage. This is the reason SoftLayer (now part of IBM Cloud) has brought these capabilities to a broader audience.
We support the NVIDIA Tesla Accelerated Computing Platform, which makes HPC capabilities more accessible to, and affordable for, everyone. Companies like Artomatix and MapD are using our NVIDIA GPU offerings to achieve unprecedented speed and performance, traditionally only achievable by building or renting an HPC lab.
By provisioning SoftLayer bare metal servers with cutting-edge NVIDIA GPU accelerators, any business can harness the processing power needed for HPC. This enables businesses to manage the most complex, compute-intensive workloads—from deep learning and big data analytics to video effects—using affordable, on-demand computing infrastructure.
Take a look at some of the groundbreaking results companies like MapD are experiencing using GPU-enabled technology running on IBM Cloud. They’re making big data exploration visually interactive and insightful by using NVIDIA Tesla K80 GPU accelerators running on SoftLayer bare metal servers.
SoftLayer has also added the NVIDIA Tesla M60 GPU to our arsenal. This GPU technology enables clients to deploy fewer, more powerful servers on our cloud while being able to churn through more jobs. Specifically, running server simulations are cut down from weeks or days to hours when compared to using a CPU-only based server—think of performance running tools and applications like Amber for molecular dynamics, Terachem for quantum chemistry, and Echelon for oil and gas.
The Tesla M60 also speeds up virtualized desktop applications. There is widespread support for running virtualized applications such as AutoCAD to Siemens NX from a GPU server. This allows clients to centralize their infrastructure while providing access to the application, regardless of location. There are endless use cases with GPUs.
With this arsenal, we are one step closer to offering real supercomputing performance on a pay-as-you-go basis, which makes this new approach to tackling big data problems accessible to customers of all sizes. We are at an interesting inflection point in our industry, where GPU technology is opening the door for the next wave of breakthroughs across multiple industries.