Objectives of this Handbook
Computer-based simulations have become standard tools for industry and academia. While the power of standard PCs or workstations with multiprocessors/cores is tremendous nowadays, they still often don’t fulfill the requirements for raw computational speed, storage and main memory that repeated detailed simulations may pose. Here, High- Performance Computing (HPC) can come to the rescue.
Supercomputer Centres operate large installations of HPC systems and have acquired considerable expertise about HPC Technologies. They offer services ranging from compute time over HPC software development to HPC technology consulting.
This handbook will explain the economic benefits of using HPC for people at managerial positions and will provide a path for attaining these benefits. Apart from the purely economic reasons we consider the improvements in innovation and marketing and the competitive advantages stemming from the use of HPC to provide justification for taking the managerial decision to explore the use of HPC.
Furthermore, the technical aspects of HPC are explained in a simplified way, oriented towards newcomers in this field, offering a first introduction of common technical terms and formal aspects in HPC as well as a short practical examples of how to connect to and start an application on one of the European HPC centers.
At the end, the reader will find some representative industrial use cases.
Origin of this document
The idea and main parts of this document come from the SESAME Net European project. It has then been elaborated by the Greek National Competence Center team.
What is HPC?
The term HPC is occasionally used as a synonym for supercomputing, although usually only the fastest HPC systems currently operated world-wide are referred to as Supercomputers. The term High Performance Computing (HPC) usually refers to the application of state-of-the-art computing systems for efficiently running advanced applications such as computer simulations, often for engineering or natural sciences purposes.
While in the past the term was mostly used for the work with very specialized and often highly expensive machines, the evolution of computer and server systems has led to a commoditization of HPC systems and to a mainstreaming of its concepts and technologies. Nowadays desktop computers and workstations potentially offer such high capacities, while at the same time HPC systems have converged so much with off-the-shelf hardware, that the term also includes achieving the most performance on mainstream systems.
The term HPC is used for nearly all aspects of managing and using high-performance systems and the software running on them. Major topics in the area include activities like:
- Performance assessment of existing applications,
- Optimization of applications for high performance,
- Specialized software development for HPC systems, and of course
- Actual execution of applications, with eventual automation of workflows or interactive visualization of large result sets.
Why use HPC?
The vast increase of computational power in the last decades has created exciting opportunities. High-Performance Computing (HPC) has become indispensable tool for industry and academia to innovate in such domains as Computer aided engineering, simulations, renewable energy, financial services, satellite, earth observation, advanced image analysis, data science and precision agriculture. But it does not stop there: supercomputing can help your business too. Supercomputing can help spark your business innovations. It can improve design turnaround time for new products, reduce time to market and increase your overall competitiveness.
The most obvious motivation for using HPC is, as mentioned before, a lack of computing power and storage on your enterprise infrastructure to run simulations or other demanding computer applications. Broadly speaking, HPC technology may be useful in all cases where large amounts of computations are necessary or large amounts of data need to be processed. And even when compute infrastructure can easily be extended, it will be difficult to reach significant performance capacity for your applications such as provided by HPC centres.
However, most of the advantages of HPC are related to the kinds of computations that cannot be accomplished on regular infrastructure. Due to the advancements in applied sciences, accurate simulations and modeling of new products and processes increasingly require HPC-type equipment that is usually not available on-site, even in medium or large scale enterprise.
Another reason to investigate HPC could be a pressure to realize shorter development cycles, e.g., because of competition on the market. Typically, new products are nowadays developed with a combination of physical experiments and simulations. Having large computing power for large and complex simulations opens possibilities for new ideas, innovation and new products in pharmacy, production, automotive components, etc. By affording to run multiple simulations with different input parameters, HPC systems for example offer the possibility to adapt the simulation closer to the experiments or to optimize products through simulations.
Large industrial enterprises, e.g. from the automotive industry, operate their own HPC clusters since a decade. HPC centers can enable small and medium enterprises to access such powerful resources.
HPC can add tremendous value to any SME that develops products:
- Creation of large, high-fidelity models that yield accurate and detailed insight into a design’s performance.
- Extremely accurate simulations to predict real-world conditions to convince a customer of your product.
- In engineering HPC enables high mesh densities for improved accuracy, numerous geometric details, or sophisticated treatment of physical phenomena.
Aspects and benefits of using HPC technology, specifically by SMEs
Many studies establish how the use of HPC opens new markets or allows for expansion of the existing markets, through improvements in innovation and enabling of new technologies products.
- Economic reasons
- Innovative reasons
- Marketing reasons
- Competitive advantage reasons
Economic reasons: explaining how HPC will reduce costs and/or improve benefits.
In this section we are going to focus on the economic reasons to integrate HPC as a key aspect in the enterprise value chain. For example, reducing designing and prototyping costs: detecting design errors in early stages of product development. It’s important to always focus on the CEO-like argument, so in detecting design errors earlier it should be mentioned the positive economic impacts that this would have: early detected design errors are cheaper to overcome. Another example, if prototyping has a major role in the SME’s chain value then HPC saves the costs of building many expensive physical prototypes which now can be built virtually, minimizing the number of physical prototypes to be built.
The three key economic reasons to consider to integrate HPC in SMEs (and also in big enterprises) are:
- Cost reduction: Prototyping (wind tunnel, physical testing) and design costs;
- Reducing design times, reducing time to market, (translating this time reduction in money);
- New product/service development that is not possible to develop without HPC.
Each of these aspects will be analyzed and introduced with more detail.
HPC can help the SME to save costs, for example in prototyping and design procedures, as well as in operational costs. Physical prototyping is to create a physical prototype (an early sample, model), or release of a product build to test a concept or process. This process is very time and cost consuming, because of the time needed to produce an item outside of an assembly line and the cost of prime matters, and this only gets worse when multiple prototypes are needed, especially in case of destructive physical testing when prototypes can only be used once. Physical prototypes can be substituted by virtual prototypes that are cheaper to develop and can be used again and again without destroying them.
Virtual prototypes, once developed, they can be used to simulate the physical behavior or interactions under many circumstances: aero and hydrodynamics, substituting the need of complexity and expensiveness of wind tunnels or fluid tests, useful in: Aerospatiale, automotive, naval and constructive; Physical tests, virtual prototypes can be used to perform multiple virtual test using simulation, no matter if these tests simulate physical damage or full destruction of the prototype.
Design costs can be reduced using HPC, since HPC brings us the power to design prototypes of products without the need of any physical prototype and begin to get knowledge about it in the very early stages of design. This also improves the design process allowing detecting design errors or misconceptions in these early stages of design, where they can be corrected easily and cheaper than in later stages. Operational costs of the SMEs can also be saved by accurate and extensive simulation, modeling and optimization of certain kinds of operations. For example, HPC can be used in optimization of public transport or logistics. It is widely used in the search of oil and gas or in energy production problems.
Reducing design times
In some sectors like aeronautics, engineering or energy, where the interaction of fluids with products like: aircrafts, shipbuilding, turbines, wind or offshore energy generators is a key factor of efficiency, HPC can be a key to success because it can reduce design times drastically.
The times needed to construct a virtual prototype and discover the full interactions with their surrounding or inner contained fluids are much shorter than the time needed to get this knowledge from physical ones. This is an uncontestable advantage that can make the difference between a successful and a ruinous project, as well as enable a project that otherwise is never started because HPC possibilities were not on mind or simply were unknown when tight schedules were present and standard design times are not possible.
As an example of the prior two reasons, under the umbrella of the EU funded project FORTISSIMO, the Italian SME EnginSoft reduced the time needed for full design and optimization of a single pump turbine from 2-3 years to 6 months and this fact allowed to recover the full SME investment in HPC Cloud based simulation and external expertise in less than six months.
Products impossible to produce without HPC
Another example of economic benefit from HPC is the possibility to produce or study products that with standard technology limitations are impossible or too costly to design and produce.
The study of hazardous substances (explosive, toxic or mutagenic) has heavy limitations due to their inherent properties and security measures needed to manipulate them. The physical measures needed to overcome these limitations are prohibitive in terms of cost, so the physical experimentation with these substances will have a bad ROI in most cases and involve high risks in all of them. HPC can overcome these limitations and simulate the behavior of these substances with virtual models getting knowledge about them in a non- risky environment.
Limiting the number of physical tests only to experiments with high percentages of success
Another FORTISSIMO experiment shows how a Swiss SME, Lonza from the chemical sector, suppliers to the Pharma Biotech and specially ingredients market, explored with success the possibility to determine physical properties of compounds with desired precision using HPC. The use of HPC clusters reduced the amount of time needed to perform calculations from a six-months-long calculation in a 16- core cluster, to less than one day on an HPC system.
Economic reasons to use HPC Center services:
- Reducing costs using a Cloud-based HPC system on a pay-per-use basis at approximately half the price of owning and maintaining a sufficiently powerful in- house system.
- Running big processes that need very powerful HPC systems not affordable for SMEs.
Supercomputer centers possess state-of-the-art HPC facilities and substantial expertise in development and use of HPC applications for solving diverse practical problems. The use of these resources provide opportunities to SMEs to evaluate new technology at low financial risk, so that they can build the business case for larger investment in HPC. They have the flexibility to buy application services instead of licenses and computing systems.
Additionally, some SMEs who are service or software providers may also expand significantly their portfolio if they incorporate the use of HPC technology as part of their offering. By co-designing with supercomputing centers, SMEs can address new business cases and markets.
Innovation reasons: explaining how HPC will improve possibilities to develop innovative products and services.
Most of the advantages of using HPC technology can be obtained by SMEs that have substantial R&D department, since the access to HPC resources can improve the possibilities to develop innovative products and services. By collaborating with an established supercomputer center or a National Competence Center (e.g. eurocc_greece.gr), the SME will not necessarily need costly on premises infrastructure. By usines flexible contracts, SMEs can have a low-rist, low-effort, pay as you go access to HPC and leverage state of the art software and tools to speed up their development process.
Apart from the cost-related reasons, there are some main advantages of using HPC:
- Obtaining results much faster by employing high amount of compute power;
- Achieving much more accurate representation of the product and its working environment;
- Exploring much wider search space for the parameters of the envisaged product.
By collaborating with a Supercomputer Center, the personnel at the SME’s Research & Development department acquire precious expertise that is also useful in other, more standard situations. For example, the increased use of multi-core devices in all kinds of products makes knowledge of parallel computing paradigms an important asset. The need to achieve a given result with as little use of energy as possible stresses the importance of using optimal algorithms and scalable programming approaches.
Overall, the use of HPC for completing simulations that would normally run for days or weeks on a desktop workstation, in a matter of hours enables much faster time-to-market and thus first-mover advantage for the SME.
It should not be underestimated that being in close contact with the academia and HPC experts in industry, the SME may have advanced knowledge of new developments in HPC technology or applied sciences that will enable cross-fertilization and rapid incorporate of the new technologies in their product portfolio.
Best product build with best technology:
In our society, most of the products get an improved perception by clients when they are built with the latest technology. Even if the use of this technology only involves a very low percentage of improvement, there are clients that always want to own the “state-of-art” product with the last improvements, while others are more concerned about environment and want to have “green” products, with more efficiency and low power consumption. HPC technology can deliver all these properties, since it is an innovative technology with relatively low use by SMEs because it is not generalized yet. Innovative technologies can lead to innovative products, and for SMEs willing to improve their products it is a competitive advantage. HPC can be applied successfully to improve products in order to make them more “green”. Under the umbrella of FORTISSIMO there are some examples of products optimized using HPC that gained special characteristics that can be valuable for some products or differentiate them from the product ́s competence.
In the aerodynamics sector, Pipistrel, a Slovenian SME, used HPC to improve the design of their aircraft and KEW optimize the aircraft wiring design not only to reduce costs saving prime matters but less prime matters is less weight, and less weight is less fuel consumption and over the 30 to 50 years of an aircraft life this has at last economic relevance. In the automotive sector AVL used HPC in the cloud to perform studies about CO2 emissions reduction. In the civil construction sector, the Scottish SME IES performs building energy efficiency studies but making the jump to HPC enabled them to scale their studies from buildings to cities.
Marketing reasons: present your SME as a cutting-edge technological firm.
An important marketing reason for the use of HPC is to present the enterprise as a cutting-edge technological firm. This has a positive effect in its market position and on the perception of its products and services by customers. And of course, it results in the subsequent positive economic impacts, e.g. more sales, price premiums, and so forth.
HPC can be the difference between one leading brand and their competitors that don’t use this technology. To associate your brand with innovation and differentiate against competitors’ products, the use of HPC for product improvement is a good point which, if correctly highlighted, can increase the product brand ́s sales. Final customers always want to buy best cost-value products and including HPC in the chain value of the SME can be a technology to achieve these added value products. This consideration is even more visible in provisioning of services, where the use of powerful HPC equipment can lead to increased consumer trust.
In all those cases it is important to be able to substantiate the marketing claims with scientific and technical arguments.
Competitive advantage reasons: Technological capable firms have inherent advantages compared to other ones.
Being able to bring new products to the market, thanks to the HPC edge on research and development process, allows the SME to gain a competitive advantage over their competitors. This first-mover advantage, thanks to the technological leadership, has several positive impacts in the firm. For example brand recognition, this is the automatic association in customers’ minds of the new type of product to its initial manufacturer, e.g. the copying machines and Xerox, the tissue paper and Kleenex. Another example of first-mover advantage is the preemption of scarce assets, which allows the first-mover firm which has superior information to purchase assets at market prices below those that will prevail later in the evolution of the market.
Additionally, HPC capable firms are able to reduce the product cycle time, this is the period for a manufacturer to complete development and production of a new, or modified, product. It does not only reduce costs, but an improvement in the time to market response allows for a bigger maneuverability to adapt to the changing conditions of modern markets.
The use of HPC enables the design of a more accomplished product, with less defects or covering a wider spectrum of consumer needs.
In today’s economy it is increasingly more important to find the consumer niche and to engage with the potential clients. With HPC technology the SME can employ advanced algorithms and processes that will increase their market share and reach a wider consumer audience.
By establishing collaboration, possibly through a pilot project, not only with HPC resource providers, but also with Independent Software Vendors and research software providers, application experts and consultants, the SME is immersed in a network with high innovation potential and thus will have increased flexibility and high potential for sustainable development of innovative products.
Main parts of this document derive from work accomplished in the framework of the SESAME NET European project”