Multi-GPU servers highlight new technology from SC'14

Share this on social media:

With SC’14 in full swing, announcements are coming thick and fast. From new multi-GPU servers, HPC management software to CFD as a service, here is a roundup of day two at the US supercomputing conference.

Dell has released the latest addition to the Dell PowerEdge server portfolio, the Dell PowerEdge C4130. The company also announced Dell Storage for HPC with the Intel EE for Lustre.

The PowerEdge C4130 can achieve over 7.2 teraflops on a single 1U server and has a performance/watt ratio of up to 4.17 gigaflops per watt. Incorporating from Nvidia and Intel, the PowerEdge C4130 is available in five configurations that address the flexibility necessary for varying HPC workload characteristics.

The PowerEdge C4130 adds up to two Intel Xeon E5-2600v3 processors, 16 DIMMs of DDR4 memory and up to two 1.8-inch SATA SSD boot drives in just 1U of rack space. With up to four 300W Intel Xeon Phi coprocessors or Nvidia Tesla GPU accelerators, including the new Nvidia Tesla K80 dual-GPU accelerator, the C1430 is designed with intense compute workloads in mind.

Dell Storage for HPC with the Intel EE for Lustre software combines Dell’s hardware with Intel’s open-source Lustre software to bring the benefits of an HPC file system to enterprise organisations of all types. The solution delivers a higher level of computing power and throughput, making available information and insights derived from big data and compute-intensive applications – including advanced modelling, simulation, and data analysis – to a wider audience of enterprise users.

The solution is built on Dell’s PowerEdge servers and Dell Storage MD3460 arrays and MD3060e dense enclosures to provide redundant servers and failover-enabled storage. It is one of the first solutions with validation and support for Intel EE for Lustre software, providing customers with simplified installation, configuration, monitoring and overall management features that are backed by both Dell and Intel.

'The PowerEdge C4130 has really impressive features for HPC environments, including industry-leading accelerator density with up to four 300W PCIe accelerators per 1U,’ said Darren Kozlek, Western Hemisphere Information Systems Manager, TGS. ‘But I’m even more impressed with how Dell listened to our feedback and deliberately designed a flexible solution that met our technical computing needs. With the PowerEdge C4130, Dell delivered the density and flexibility we need to tackle our most demanding workloads.’

This year SysFera will showcase a new CFD-AAS (CFD-as-a-service) solution that makes use of the Romeo supercomputer, the most energy efficient supercomputer in France.

The solution is now available for small and medium-sized industrial users on ROMEO’s supercomputer with a ‘bring your own’ license model.

The company is also offering SysFera-DS high performance computing management software that is available through a conventional web browser. The use of a web-based portal makes it easier to access data remotely and includes functionality to allow visualisation of 3D applications.

David Loureiro, CEO of SysFera, commented: ‘Delivering easily accessible and efficient HPC CFD simulations as a service to the desktop for an industrial customer was a challenge that we tackled with SysFera-DS – in collaboration with Romeo and Tech-Am Ing. Easy and efficient access to simulation tools is crucial for industries; Tech-Am Ing's work reduced the duration and cost of the whole process of producing sugar. Our goal is to take this same level of resources and HPC access to all small- and medium-sized organisations.’

Tech-Am Ing, a company specialising in numerical simulation and complex CFD, has worked with the Romeo cluster since 2011, solving real challenges in the process industry, where CFD simulations involve complex interactions. CFD simulation performed with HPC is a powerful tool for mixing system providers and is key to future innovation.
 
Finally Allinea Software has announced an expansion of its tool suite that it hopes will boost developer productivity, and the launch of support for hybrid OpenMP codes and energy measurement in its performance profiling tool, Allinea MAP.

David Lecomber, CEO of Allinea Software, said: ‘Our unified tools for performance and debugging are making a real difference in time to develop and training costs for our customers. Consistently our users spoke of wanting to be able to complete more of their development process in the tools: when they debug and when they profile, they want to be able to make changes and immediately test those out on their system.

‘That is why we are now announcing Allinea Forge, we’re bringing together our market leading tools – the performance optimisation tool Allinea MAP and debugging tool Allinea DDT – with fully-integrated edit, compile, build, test, and version control support. Those steps are a vital part of the workflow – and it’s all here in one tool.

‘We are also announcing performance profiling for OpenMP and an early access program for energy profiling - significant additions to Allinea MAP,’ added Lecomber.  ‘We know hybrid programming is set to continue to grow in importance as per node core counts march upwards. Understanding the bottlenecks of OpenMP code is now vital to extracting performance on today’s architectures.’