SCI Accelerator Program

Accelerators are co-processing components containing massive numbers of functional units, together with memory and control systems that can be added to computers to speed up of applications. Accelerators are similar to turbochargers in an automobile with the purpose of increasing the speed of an application with a low incremental use in power, for faster time to business, engineering, or scientific outcomes.

Now that clock speeds are improving only slowly for process technology reasons, microprocessor vendors propose to offer improved performance by increasing the number of cores per chip. This approach does not automatically enable microprocessors to increase application performance at the rates we’ve come to expect. Therefore, many vendors and users are proposing alternative technologies such as General Purpose Graphical Processing Units (GPGPUs), Field Programmable Gate Arrays (FPGAs), and custom ASICs that will deliver substantial increases in application performance on industry standard platforms.

The accelerator project is a hands-on, on-going effort to investigate various acceleration technologies for high performance computing and benchmark them to determine which ones are most beneficial for our customers. We are also collaborating with a number of software tool vendors to help ensure their products work well in the scale-out environment. To date, we have installed HP accelerated clusters in a number of customer deployments, including Oil & Gas, FSI and Federal environments.

Qualified Accelerators today

• NVIDIA CUDA programmable GPUs:

• Tesla S1070 in ProLiant DL160G5
• Quadro FX 5600 in ProLiant DL785G5
• Quadro FX 3700 in ProLiant DL160G5 and DL165G5

• Nallatech FPGAs in DL160



• ClearSpeed SIMD accelerators:

• Advance™ e620 in ProLiant DL140G3, DL380G5, DL160G5; BladeSystem BL460c and BL465c
• Advance™ e720 in BladeSystem BL460

Newly Qualified!

• XtremeData (XDI) in-socket FPGA acceleration for the rack mounted ProLiant DL165G5 and DL185G5