The designation ‘supercomputer’ has changed. No longer does it strictly define the highly customised petaflop powerhouses that cost a small fortune to run and are used in ultra-specialist fields such as molecular simulation and interstellar calculations.
Instead, supercomputing-scale capabilities have expanded beyond the rarefied fields of scientific research and compute-intensive academic projects. Why? Because compute-intensive workloads are now a new normal in the enterprise data centre.
Many commercial sectors now want supercomputing-scale data processing and data management to drive the process system operations that deliver competitive objectives.
In verticals such as finance, for example, some of the highest-performing compute power available from Lenovo, such as such as System x3550 M5/System x3650 M5 and NeXtScale System M5 servers, are now platforms of choice for applications like high-speed algorithmic trading and low-latency trading.
Such trading platforms have to process data at the highest possible speeds, and then analyse it for insights that the system is empowered to act upon. Other decision-support tools that involve human intervention must operate just as super-fast, with latencies among the lowest benchmarkable. Technologies that can trim milliseconds off trade completion times make the difference between big profits and big losses.
In manufacturing, high-performance computing (HPC)-enabled systems shorten development cycles so products can get to market ahead of the competition. Lenovo customer the STFC Hartree Centre, for example, applies prototyping solutions that may have previously existed principally only in academic research. Instead of mocking up new products and testing them as physical entities, the centre runs advanced modelling simulations on its Lenovo HPC platforms. This allows manufactures to complete a simulation in hours rather than days.
All these real-world transformations point to a trend that Lenovo is well positioned to leverage for multiple reasons. For a start, its reach into supercomputing is unmatched by other enterprise IT solutions providers. Lenovo has established partnerships with centres of supercomputing excellence in leading academic and research institutions around the world. It is now the major technology provider in the design and delivery of pre-eminent supercomputer facilities across Europe: examples include Leibniz Supercomputing Centre (Germany), CINECA (Italy), Barcelona Supercomputing Center (Spain) and the University of Birmingham (UK).
These are more than conventional supplier–customer relationships. As opportunities arise – and they are many – Lenovo partners with its customers to develop further innovative spin-off solutions that will benefit its corporate customers. The University of Birmingham, for instance, is deploying supercomputer cooling technology co-developed by Lenovo and the university’s Research Computing Team. This breakthrough system is engineered to cut cooling-energy costs by up to 83 per cent and allow denser server configurations.
Each of these supercomputing partners is engaged upon a range of world-changing projects aimed to solve some of humanity’s greatest challenges. Lenovo technology helps accelerate the progress they make toward those ends. So it’s not surprising that the enterprise world is very interested in what HPC can also do to solve some of the greatest challenges faced by businesses.
There’s considerable crossover in high-end computing terms between commercial imperatives and academic research. Both need IT to meet hard deadlines and demanding workloads in environments that call for constant change and review. The shared resources model that most educational institutions have in place (with origins in the grid computing concept as it evolved in the 2000s) anticipated today’s enterprise cloud services model.
Lenovo’s work with CINECA – a consortium that includes 70 Italian universities – has special resonance in this context. The research institution’s MARCONI supercomputer is built on Lenovo NeXtScale System M5 servers, powered by Intel Xeon Phi and Xeon E5-2600 v4 processor families.
This supercomputer leverages the Intel Omni-Path Architecture (OPA), which provides the high-performance interconnectivity required to optimally scale MARCONI’s thousands of servers. Lenovo has been an enthusiastic supporter of Intel’s Scalable System Framework, which unites compute, memory, storage, fabric and software technologies for the next wave of HPC, and of which OPA is a key component. A high-performance Lenovo GSS storage subsystem, that integrates the IBM Spectrum Scale (GPFS) file system, is connected to the Intel Omni-Path Fabric and provides data storage capacity.
Some of Italy’s most important industrial names (think Eni, Centro Ricerche Fiat and Dompé Farmaceutici) use CINECA’s HPC facilities. Meanwhile, CINECA is also a core HPC partner for Fortissimo. The collaborative initiative enables European SMEs to be more competitive in overseas markets by providing access to tools and services running on cloud-based HPC.
Supercomputers have been somewhat synonymous with expensive technology unlike anything found in the enterprise IT mainstream. The difference here is twofold:
Lenovo has made a further major commitment to HPC with its own centre of excellence. Its HPC innovation centre is a permanent R&D and application benchmarking site, serving also as a meeting place for partners to collaborate and bring the commercial benefits of HPC to a broader market.
HPC-minded data centre managers must think hard about which technology providers they partner with going forward. Supercomputing and business computing solutions providers used to address separate markets with different needs, but as Lenovo’s experience shows, that’s no longer the case.