From standing start to peak performance
Accelerate your HPC strategy
Brought to you by
Computing surveyed 76
IT decision-makers using High Performance Computing (HPC)
at their organisation.
High performance computing (HPC) workloads are amongst the most sensitive when it comes to the hardware that underpins them.
Thread counts, memory capacity and bandwidth, and storage performance are all crucial. Dedicated accelerators also play a vital role in boosting throughput.
All of this not only has implications for CapEx on-prem, or OpEx in the cloud, but also energy consumption. Efficiency levels must be maximised for HPC workloads often running near capacity.
This report, featuring dedicated research findings from HPC end user organisations, reveals how best to accelerate your HPC strategy.
It answers key questions around whether to opt for on-premises or cloud solutions, the technology and business considerations that underpin a sound strategy, common use cases, and the main challenges and benefits experienced along the way.
Executive summary
24
of organisations use all on-premises HPC.
%
The most common issues facing organisations are ageing infrastructure, ongoing operations or management costs, and high energy consumption.
ICON
%
Key Findings
ICON
80 per cent of IT leaders agree their organisation is in the early stages of AI adoption.
%
The main factors shaping AI strategy are cost reduction, decision-making speed and confidence, and customer insights.
ICON
%
The greatest gains of HPC are recognised as performance and scalability improvements, business agility, and high-capacity memory.
%
ICON
HPC is being predominantly used for data analysis as well as simulating and modelling test scenarios.
%
Demand is surging for HPC as organisations become increasingly aware of the benefits, throughput, and competitive advantage available. HPC is all about enabling high speed insights and performance at scale, making vast numbers of complex calculations rapidly.
With this comes large cost savings too as HPC systems process data faster, applications run faster, and responses are generated faster. Over 70 per cent of organisations are using HPC predominantly for its data analysis capabilities.
Real benefits in real-time
HPC has been at the forefront of industry innovation for the past few decades, playing a core and critical part in analysing vast volumes of data at speed. HPC systems remove the task of physical testing by allowing simulations or modelling to be carried out in place of expensive, repetitive physical enactments. For roughly half of the organisations surveyed, HPC is being used in simulations and modelling. Similarly, a quarter are using HPC in computational fluid dynamics.
Because of this, HPC is a key asset in academic research in allowing simulations for complex scenarios such as airplane wing airflow, analysing drug interactions or DNA sequencing and molecular modelling, eliminating the need for physical testing.
Cost is further benefited as less time and resources are wasted thanks to high performance at ever-improving efficiency levels. This is exemplified in industries such as computer animation that use rendering to achieve special effects. Conventional processors would typically take days or weeks to achieve the same calculated outputs as HPC.
Importance of endpoint device decision makers
Given the widespread shift to cloud computing amidst a competitive landscape, organisations are tasked with choosing the best cloud provider for them. The top three considerations for IT leaders when it comes to committing to a provider include cost, whether they are existing customers of the provider, and the technical superiority of the solution
Considerations and challenges
A distributed workforce makes a Zero Trust security strategy even more important
The importance of providing secure remote access to data, applications, and networks to ongoing organisational success
Remote working presents a serious risk to our network security
In the past, adopting HPC meant owning or borrowing supercomputers. Today, utilising accelerators that are already built into the CPU gives organisations immediate and reliable performance while saving on additional costs associated with new hardware or leasing infrastructure.
The features that enable optimum, efficient HPC are heavily dependent on choosing the right hardware. Memory capacity, bandwidth, storage, and throughput are all important hardware decisions, meaning organisations must concentrate on maximising their investment.
Accelerators and hardware
“Improved monitoring and understanding of spikes and usage patterns”
“Better performance”
“Server-class features and thread performance with cutting-edge low power design”
“Substantial improvement in response time”
Biggest endpoint device buying challenges
Executive summary
Real benefits in real-time
On-prem v cloud
Considerations and challenges
Accelerators and hardware
55%
14%
49%
44%
Supply chain issues
Hybrid/remote working
Budget constraints
User expectations
Security concerns
Strongly disagree
Somewhat disagree
Neither agree nor disagree
Somewhat agree
Strongly agree
3%
9%
25%
24%
39%
Intel (Nasdaq: INTC) is an industry leader, creating world-changing technology that enables global progress and enriches lives. Inspired by Moore’s Law, we continuously work to advance the design and manufacturing of semiconductors to help address our customers’ greatest challenges. By embedding intelligence in the cloud, network, edge and every kind of computing device, we unleash the potential of data to transform business and society for the better.
To learn more about Intel’s innovations, go to newsroom.intel.com and intel.com.
Whether it’s powering the next great scientific breakthrough or generating the data that just might save an endangered species, Intel hardware is giving power to innovation, tackling today’s biggest challenges with built-in AI accelerators.
Learn more
Learn more
About the sponsor – Intel
24%
26%
30%
15%
74%
RETURN TO HUB
Introduction
Key findings
Challenging landscape
People problems
The importance of capable hardware
Conclusion
Sponsor info
Using devices past their life cycle could impact reliability, productivity and security
Keeping these stakeholders on board, while also enabling them to work productively and securely may require a refresh rethink.
Securing endpoints is difficult due to a dispersed workforce
1 = not at all important, 10 = extremely important
1 2 3 4 5 6 7 8 9 10
1% 1% 1% 1% 2% 2% 7% 21% 26% 39%
How does Zero Trust fit into your endpoint management processes?
It’s a key requirement
It’s a key part of our strategy
It is becoming integral
It doesn’t fit into our processes at the moment
It is not something we have explored
It requires a change in philosophy to an ‘always verify, never trust’ model
Without it, we would struggle to manage all our resources effectively
Under consideration
Something we need to adopt in the future
Currently planning our approach to this in our cybersecurity strategy
It’s hard to say, we’re not sure how yet
We are getting there. It will hopefully become a core element
Currently not included but considered necessary going forward
The impact of accelerators on HPC workloads
“Huge impact on relevant workloads”
19%
16%
3%
49%
18%
13%
Strongly disagree
Somewhat disagree
Neither agree nor disagree
Somewhat agree
Strongly agree
0%
32%
12%
5%
51%
We surveyed 130 IT decision-makers involved in managing, testing, evaluating, specifying, recommending, or procuring endpoint estate products at their organisations. The majority were IT managers, IT directors and
C-suite professionals.
25%
When developing their endpoint refresh strategy, organisations should evaluate their existing endpoint needs, how these could change in the future, how new devices will fit with current ones and how best to optimise new endpoints.
Out-of-sequence refreshes may be required to ensure those across your organisation are equipped with the necessary technology to do their jobs correctly.
1st
2nd
3rd
4th
IT teams
End-users
Financial decision-makers
Procurement personnel
End users are more vocal and more knowledgeable in what they want and need
OS updates
29%
Challenge today
More challenging over next two years
64%
53%
49%
35%
34%
16%
Just over half say hardware specification is vital for effective AI and analytic workflows.
ICON
%
15%
13%
17%
32%
24%
14%
HPC infrastructure strategies
All in the cloud
An even split between cloud and on-premises
Mainly in the cloud but with some on-premises
Mainly on-premises with occasional cloud “bursts”
All on-premises
OS updates
Data analysis
Simulating/
modelling test scenarios
Fraud detection
Computational
fluid dynamics
AI inferencing
Designing new products
Sponsor info
32
%
have exclusively operated HPC in the cloud.
60
%
of organisations agree cloud computing would alleviate their current HPC challenges.
97
%
expect their use of cloud HPC workloads to increase or at least remain the same in the coming years
20
%
of HPC users are knowingly employing accelerators – leaving room for many more to benefit from untapped performance gains.
94
%
of respondents agree the use of accelerators has already enabled new HPC use cases for their organisation or they expect it will very shortly.
Respondents’ use of HPC
13%
12%
Medical applications
Financial risk assessment
Improved performance
55%
Improved scalability
51%
Business agility
25%
High-capacity memory
25%
Reduction of operating costs/management costs
22%
Improved security
20%
High-capacity networking
20%
Reduced complexity
18%
Reducing need for physical testing
15%
Important factors for HPC workloads (3 maximum)
Cost
50%
We were already a user of their other cloud services
47%
Technical superiority of solution
47%
Technical issues relating to your specific HPC need
24%
Compatible with on-premises HPC
22%
Top considerations when choosing a HPC cloud provider (3 maximum)
As workloads, budgets, and priorities change, the choice of where to house HPC capabilities can be challenging. While on-premises HPC can run the risk of becoming underutilised or outmoded, with large capital expenditure required to upgrade, it can make most sense for those with predictable workloads, niche requirements, or strict security and compliance obligations. Nonetheless, ongoing operating and management costs are a key problem for those using on-premises HPC environments and scalability can be an issue.
Many have made the leap to the cloud for more efficient HPC spending, allowing the flexibility to scale up or down to meet the needs and demands of their organisation. Cloud environments also offer the ability to adapt data collection.
Sixty per cent of respondents agree the use of cloud computing could help alleviate their current HPC challenges. As your cloud provider iterates on its performance, reliability, and power credentials, your HPC workloads will similarly see enhanced capabilities. In contrast, a third neither agree nor disagree that this is the case, emphasising how different setups will suit different organisations.
For some, a hybrid approach will make the most sense. Around a third of surveyed organisations are operating HPC in a split between cloud and on-premises in a best-of-breed strategy, while 15 per cent are using the occasional cloud ‘burst’ in their approach to computing.
Approaches to HPC vary – roughly a quarter of organisations are fully on-premises when it comes to HPC, while around a third choose to be entirely in the cloud.
More challenging over next two years
64%
53%
49%
35%
34%
16%
Challenge today
On-prem v cloud
13%
47%
34%
5%
0%
More challenging over next two years
64%
53%
49%
35%
34%
16%
Challenge today
“Cloud computing could help alleviate my organisation’s current HPC challenges”
Neither agree nor disagree
Somewhat disagree
Strongly disagree
Strongly agree
OS updates
Somewhat agree
CLICK HERE
Learn more
To learn more about on prem v cloud, explore this digital report
Unsurprisingly, costs are top of mind today for IT leaders. Economic uncertainty and budget constraints mean teams must be discerning when implementing new approaches or maintaining existing ones.
The capital costs are considerable: setting up an HPC environment can come with a price tag north of $100m and – as hardware must be regularly refreshed – the initial outlay is far from the total cost.
Ultimately, the benefits to performance, scalability, and agility (as noted in the following data) will outweigh and justify initial expenditure when it comes to HPC. Running HPC greatly reduces operating and management costs, and providers will be able to fine tune capabilities to specific organisations. For example, the amount of memory needed, the number of processors, as well as what kinds of networks should be used.
Technical superiority is also a key consideration when choosing cloud providers. Organisations want to employ the latest generation of processors to ensure they are maximising performance and energy efficiency, as well as putting accelerators to use. This will permit the desired capacity and performance during peak times and allow multiple simulations or models to be carried out in tandem. Having this capability is especially important for organisations prioritising consistent results from data and computations such as simulations for aerodynamics, chemistry, automobiles, and more.
The creation of new business models
8%
The need for advanced hardware capabilities resonates with respondents, with around 1-in-2 IT leaders admitting ageing infrastructure is a chief issue when it comes to HPC. HPC workloads involve huge amounts of complex data, demanding the best CPU and GPU performance for accurate and timely results.
Roughly a third of organisations identify high energy consumption as a key problem in using HPC, but there is hope.
As semiconductor manufacturing process nodes shrink, so too do can the energy demands of the resulting processors. Architectural improvements, such as support of specific instruction sets can further improve efficiency. Approaches such as liquid cooling also help to reduce operating temperatures, again benefiting energy usage, and reducing failure rates.
However, 1-in-4 organisations say they have not experienced any of the below issues, demonstrating the success of HPC.
Ongoing operation/management costs
28%
High energy consumption
27%
Job queue is too long for business needs
21%
HPC has not kept pace with our business needs
14%
Difficulties keeping up with frequent upgrades
13%
High utilisation rate causing problems
11%
My organisation has not experienced any of these issues
24%
HPC issues (3 maximum)
25
%
Unsure
32
%
0 to 25 per cent
13
%
76 to 100 per cent
15
%
51 to 75 per cent
16
%
26 to 50 per cent
HPC utilisation
While a quarter of respondents are unsure of their current HPC utilisation rate, organisations are typically operating between 26 to 100 per cent. Just thirteen per cent are utilising HPC at over three-quarters of total capacity, highlighting that just a minor proportion are making the most of their computing resources.
When asked how they expect their use of HPC to change in the coming years, greater utilisation is the dominant sentiment. Sixteen per cent of IT leaders expect their use of HPC to increase greatly in the next two years, while the majority, at 53 per cent, expect it to increase somewhat.
Overwhelmingly, 97 per cent of organisations expect their usage to increase or stay the same, suggesting widespread initial success is motivating continued or greater use in the future.
Increase greatly
Increase somewhat
Stay the same
Decrease greatly
Decrease somewhat
16%
0%
53%
28%
4%
Expected changes in use of HPC workloads
over the next two years
For some, differing workloads demands will require specific hardware, however, in many instances built-in accelerators overcome this issue. HPC accelerators aid workload performance and tackle advance computational tasks at high speed. In contrast to standard computing systems, HPC leverages massive parallel computing power through clusters and other high-performance components.
A fifth of organisations are currently using accelerators to speed up HPC calculations. However, 40 per cent are unsure on their implementation for their enterprise, suggesting it is very likely adoption is more widespread.
20
%
Yes
Does your organisation use HPC accelerators?
Organisations using HPC accelerators overwhelmingly agree they have improved the efficiency of their workloads.
They also clearly recognise the success of HPC accelerators, with over 90 per cent of respondents agreeing its use has already enabled new use cases for their organisation or they expect it will very shortly.
27%
40%
27%
7%
0%
“Accelerators have greatly improved the efficiency of our HPC workloads”
Somewhat agree
Neither agree nor disagree
Somewhat disagree
Strongly disagree
Strongly agree
47
%
Yes
7
%
No
47
%
Not currently but predict they will soon
Have accelerators enabled new HPC use cases for your organisation?
The benefits of accelerators is clear to most of those surveyed. When asked how usage has affected their organisation, respondents voiced significant improvements to performance, monitoring, and response times.
HPC solutions will continue to evolve with powerful, scalable tools and technology. Improvements and growth will rely on underlying hardware and software.
Overall, respondents are evidently impressed with their organisation’s use of HPC. For 44 per cent of IT leaders, HPC has had a transformative impact on their company. While issues may have been encountered, most, if not all, expect their use of HPC to rise in the coming years.
11%
33%
49%
7%
1%
“HPC has had a transformative
effect on our organisation”
Somewhat agree
Neither agree nor disagree
Somewhat disagree
Strongly disagree
Strongly agree
Discover how Intel is supporting HPC in organisations
Ageing infrastructure
47%
41
%
No
40
%
Unsure
From standing start to peak performance
Accelerate your HPC strategy
Brought to you by
Computing surveyed 76 IT decision-makers using High Performance Computing (HPC)
at their organisation.
High performance computing (HPC) workloads are amongst the most sensitive when it comes to the hardware that underpins them.
Thread counts, memory capacity and bandwidth, and storage performance are all crucial. Dedicated accelerators also play a vital role in boosting throughput.
All of this not only has implications for CapEx on-prem, or OpEx in the cloud, but also energy consumption. Efficiency levels must be maximised for HPC workloads often running near capacity.
This report, featuring dedicated research findings from HPC end user organisations, reveals how best to accelerate your HPC strategy.
It answers key questions around whether to opt for on-premises or cloud solutions, the technology and business considerations that underpin a sound strategy, common use cases, and the main challenges and benefits experienced along the way.
Executive summary
Key Findings
The greatest gains of HPC are recognised as performance and scalability improvements, business agility, and high-capacity memory.
%
HPC is being predominantly used for data analysis as well as simulating and modelling test scenarios.
HPC has been at the forefront of industry innovation for the past few decades, playing a core and critical part in analysing vast volumes of data at speed. HPC systems remove the task of physical testing by allowing simulations or modelling to be carried out in place of expensive, repetitive physical enactments. For roughly half of the organisations surveyed, HPC is being used in simulations and modelling. Similarly, a quarter are using HPC in computational fluid dynamics.
Because of this, HPC is a key asset in academic research in allowing simulations for complex scenarios such as airplane wing airflow, analysing drug interactions or DNA sequencing and molecular modelling, eliminating the need for physical testing.
Cost is further benefited as less time and resources are wasted thanks to high performance at ever-improving efficiency levels. This is exemplified in industries such as computer animation that use rendering to achieve special effects. Conventional processors would typically take days or weeks to achieve the same calculated outputs as HPC.
Data analysis
Simulating/
modelling test scenarios
Fraud detection
Computational
fluid dynamics
AI inferencing
Designing new products
Demand is surging for HPC as organisations become increasingly aware of the benefits, throughput, and competitive advantage available. HPC is all about enabling high speed insights and performance at scale, making vast numbers of complex calculations rapidly.
With this comes large cost savings too as HPC systems process data faster, applications run faster, and responses are generated faster. Over 70 per cent of organisations are using HPC predominantly for its data analysis capabilities.
24%
32%
17%
13%
15%
HPC infrastructure strategies
All in the cloud
An even split between cloud and on-premises
Mainly in the cloud but with some on-premises
Mainly on-premises with occasional cloud “bursts”
All on-premises
Considerations and challenges
RETURN TO HUB
RETURN TO HUB