Friends & Frenemies: Collaboration and Competition in HPC

By Chris Downing

May 29, 2018

The business of buying, selling and managing high-performance computing resources is for the most part an amicable one. Aside from the periodic low-grade spat between Mellanox and Intel over the benchmarking of their respective interconnects, we see little evidence of vendors bad-mouthing each other.

On the customer front, when a relationship with a vendor breaks down it usually results in a switch of supplier for the next time, and not much else. Cases where the supplier is just shifting hardware are relatively safe, but comparatively low margin – providers of more lucrative software support have a much harder time meeting expectations and maintaining relationships. Cloud vendors have a relatively easy time here once they are incumbent, but for the time being have to do far more work to win the business in the first place.

Friends

Collaboration between HPC research centres is mostly focused on user-space. While those who make use of HPC to get their job done are more than happy to build both academic and commercial partnerships, the staff who deliver the service itself do not have much of a steer towards this kind of outreach activity. Often, this is because services are under-staffed and therefore busy juggling upkeep and procurement cycles.

Medium and large centres might communicate as part of a user forum or special interest group for a particular region or technology, but these relationships tend to be more “show and tell” than “make stuff happen”.

Service teams should aim to be part of the wider HPC community, rather than sitting on the fringes. This might include chiming in on their preferred mailing list with tips, or attending the big industry-wide conferences. Aside from just being a nice way to work, this kind of collaboration gives much faster insights into new tricks which might benefit local users.

Mid-range commercial and academic communities also stand to benefit from the experiences of their “bigger fish” counterparts who run Top50 machines such as US labs and various national HPC centres. Someone from this crowd can always be found in the keynote section of any given HPC conference, so turn up and listen. Inevitably the talk will focus on their latest triumph rather than bothering with the daily struggle to keep things running smoothly, but between the probing questions and notable omissions of detail you should get a sense of what problems are being faced as these sites roll out new technology.

Another useful step worth taking is to identify some industry sales staff whose interests align with what you want to do within your service going forward, and make an effort to get together to discuss common problems and to share your views. This sort of activity is easy to remember to do when you approach a procurement, but that is exactly the wrong time – when you have money to spend, the answers you get will inevitably be “optimistic” about what is achievable in the short term. Better to wait until just after a procurement, and get the right people in to provide a more honest roadmap.

In the same vein, make use of the knowledge of your current supplier while they are supporting your system. The simplest way to do this is to hold regular meetings to discuss performance, upgrades etc. If they push back on this time commitment, remind them that they can learn from you too. Knowledge of how you work, and how you plan to work, should be something they are able to incorporate into their next tender response.

Finally, remember that a pro-active and positively engaged set of users will be your best friends. As long as they have confidence that the service is helping with their workload rather than hindering them, they can provide valuable input regarding the operating model while also offering “ammunition” for use in justifying further investment.

Enemies

Although HPC centres are ostensibly competing for scarce funding and staff, worrying about what the competition is doing is rarely a worthy concern for the service team beyond as a learning opportunity – most know to remain focused on what value they can deliver for their own users, which won’t always involve making the same sort of upgrades as the university in the next city over. Likewise, academics researching in the same field only typically “compete” in the sense that they might be bidding for the same external funds, which would only influence their access to HPC resources if successful bidders can also do a better job of extracting money from their institution.

As with any other aspect of their business, competitors in the commercial space need to balance the benefits of promoting their activities (e.g., for recruiting or marketing purposes) against giving away too many competitive advantages. This also applies to public sector to some extent, particularly where commercialisation of systems or data were part of the pitch for getting the money to fund the HPC service.

Internally, enemies to the HPC team only tend to appear if things have been going wrong for a while already; belligerent users, lawyers without technical expertise, and an unsupportive finance office during upgrades can all lead to misery. The first can be fixed (or at least triaged) by practicing some selective deafness – but managers need to get buy-in from their project board or top-level stakeholders before they start ignoring the needs of their more difficult users to make the service more tolerable for everyone else. The other problems are usually forgotten about until procurement time – at which point the cycle of misery begins afresh, eventually leaving everyone exhausted and more inclined to drag out the lifetime of their hardware beyond its natural span just to avoid going through the hassle yet again in three or four years. Breaking out of this cycle and turning HPC into a “business as usual” cost is one of the reasons we expect to see growing enthusiasm for cloud usage.

Frenemies

Sometimes, particularly when a round of upgrades is being planned or a fresh installation is due, relationships can become more complicated.

It is important to remember that in most cases, the supplier is only your friend to the extent that repeat custom is important to them. During the initial purchase you will hear lots of pleasing comments about “partnership”, but what that looks like when you are trying to run a production service is hard to say. Larger sites, particularly those with an operational responsibility such as weather forecasting, can reasonably demand that the partnership represents something tangible like on-site staff, though they will obviously be paying a premium for it.

Outside your immediate vendor relationships, industry sales people will promise the moon on a stick – and are rarely punished when it turns out that their company can’t deliver. Despite this, sales staff are a useful source of information and so are worth listening to (for a while at least).

Internally, users will find themselves at odds with each other over future system design and requirements. Depressingly often, requirements are driven by the loudest stakeholder, even if they are not the most productive – users must be encouraged to fight their corner while also providing realistic input to their share of the business case.

Look out for #1

The first thing an HPC customer can do to look out for their own interests is to have a good set of supportive stakeholders in place before starting their procurement. The last thing you need when problems arise is a group of people straining themselves to yell “I told you so!” and drowning out any discussion of putting things right.

For purchasers in the EU, strongly consider sticking with the open procurement method, rather than angling for a competitive dialogue or other more convoluted options. If you take the time to run some market analysis in advance, the removal of back-and-forth discussions actually offers a lower risk of delays and disruption.

Running a good open procurement requires knowing what you want, at least at a high level. For cases where the “right way” isn’t immediately clear, there is no harm in asking for options within the procurement and deciding based on price and vendor feedback, or simply saving some funds to perform an upgrade once the system is live and users have a clearer picture of what they are missing out on.

An open, formal procurement should also help with project management for the supplier by giving them a good idea what to expect between winning and cashing their cheque. The procurement contract should include some form of insurance against a sloppy delivery process by the supplier, usually in the form of liquidated damages. While suppliers may groan and wring their hands, they need to be made to understand that the short-ish lifetimes of HPC systems means that delays have a real impact on competitiveness – LD clauses are not there as a punishment, but as insurance against lost value. There is no good reason why a customer should be expected to take all of that risk themselves, effectively insuring their supplier against poor project management and an inadequate supply chain.

That said, procurements do not need to become adversarial when problems arise. Purchasers stand to benefit most by being accommodating when suppliers face minor problems and saving the “strongly worded letter” for cases where they appear to be taking advantage of your good nature.

If smaller HPC centres want to maintain a legitimate case for on-premises hardware, they will need to stay close to their internal friends and understand what exactly they want, and what they can realistically deliver in terms of value returned to the institution. Similarly, hardware suppliers will need to strengthen their efforts not to be seen as enemies by customers who are weary of procurement – this does not automatically mean big discounts (though clearly, that never hurts); instead, they should focus on making delivery painless and ongoing support truly collaborative. Finally, cloud providers need to continue their work on acting as customer friends by making the cost, value and functionality arguments for cloud migration as pain-free as possible. Without a willingness to engage in open and well-informed discussion about these topics, no amount of trial credits can provide a convincing argument that the cloud is the right place to be.

About the Author

Chris Downing joined Red Oak Consulting (@redoakHPC) in 2014 on completion of his PhD thesis in computational chemistry at University College London. Having performed academic research using the last two UK national supercomputing services (HECToR and ARCHER) as well as a number of smaller HPC resources, Chris is familiar with the complexities of matching both hardware and software to user requirements. His detailed knowledge of materials chemistry and solid-state physics means that he is well-placed to offer insight into emerging technologies. Chris, Principal Consultant, has a highly technical skill set working mainly in the innovation and research team providing a broad range of technical consultancy services. To find out more, visit www.redoakconsulting.co.uk.

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industy updates delivered to you every week!

Nvidia Leads Alpha MLPerf Benchmarking Round

December 12, 2018

Seven months after the launch of its AI benchmarking suite, the MLPerf consortium is releasing the first round of results based on submissions from Nvidia, Google and Intel. Of the seven benchmarks encompassed in version Read more…

By Tiffany Trader

Neural Network ‘Synapse’ Technology Showcased at IEEE Meeting

December 12, 2018

There’s nice snapshot of advancing work to develop improved neural network “synapse” technologies posted yesterday on IEEE Spectrum. Lower power, ease of use, manufacturability, and performance are all key paramete Read more…

By John Russell

IBM, Nvidia in AI Data Pipeline, Processing, Storage Union

December 11, 2018

IBM and Nvidia today announced a new turnkey AI solution that combines IBM Spectrum Scale scale-out file storage with Nvidia’s GPU-based DGX-1 AI server to provide what the companies call the “the highest performance Read more…

By Doug Black

HPE Extreme Performance Solutions

AI Can Be Scary. But Choosing the Wrong Partners Can Be Mortifying!

As you continue to dive deeper into AI, you will discover it is more than just deep learning. AI is an extremely complex set of machine learning, deep learning, reinforcement, and analytics algorithms with varying compute, storage, memory, and communications needs. Read more…

IBM Accelerated Insights

Blurring the Lines Between HPC and AI @ SC18

The dominant topic at SC18 was the convergence of HPC and Artificial Intelligence (AI) with some of the biggest research and enterprise HPC users providing perspectives on how HPC and AI are moving closer together. Read more…

Is Amazon’s Plunge into Server Chips a Watershed Moment?

December 11, 2018

For several years now the big cloud providers – Amazon, Microsoft Azure, Google, et al – have been transforming from technology consumers into technology creators in hardware and software. The most recent example bei Read more…

By John Russell

Nvidia Leads Alpha MLPerf Benchmarking Round

December 12, 2018

Seven months after the launch of its AI benchmarking suite, the MLPerf consortium is releasing the first round of results based on submissions from Nvidia, Goog Read more…

By Tiffany Trader

IBM, Nvidia in AI Data Pipeline, Processing, Storage Union

December 11, 2018

IBM and Nvidia today announced a new turnkey AI solution that combines IBM Spectrum Scale scale-out file storage with Nvidia’s GPU-based DGX-1 AI server to pr Read more…

By Doug Black

Is Amazon’s Plunge into Server Chips a Watershed Moment?

December 11, 2018

For several years now the big cloud providers – Amazon, Microsoft Azure, Google, et al – have been transforming from technology consumers into technology cr Read more…

By John Russell

Mellanox Uses Univa to Extend Silicon Design HPC Operation to Azure

December 11, 2018

Call it a corollary to Murphy’s Law: When a system is most in demand, when end users are most dependent on the system performing as required, when it’s crunch time – that’s when the system is most likely to blow up. Or make you wait in line to use it. Read more…

By Doug Black

Topology Can Help Us Find Patterns in Weather

December 6, 2018

Topology--the study of shapes--seems to be all the rage. You could even say that data has shape, and shape matters. Shapes are comfortable and familiar concepts, so it is intriguing to see that many applications are being recast to use topology. For instance, looking for weather and climate patterns. Read more…

By James Reinders

Zettascale by 2035? China Thinks So

December 6, 2018

Exascale machines (of at least a 1 exaflops peak) are anticipated to arrive by around 2020, a few years behind original predictions; and given extreme-scale performance challenges are not getting any easier, it makes sense that researchers are already looking ahead to the next big 1,000x performance goal post: zettascale computing. Read more…

By Tiffany Trader

Robust Quantum Computers Still a Decade Away, Says Nat’l Academies Report

December 5, 2018

The National Academies of Science, Engineering, and Medicine yesterday released a report – Quantum Computing: Progress and Prospects – whose optimism about Read more…

By John Russell

Revisiting the 2008 Exascale Computing Study at SC18

November 29, 2018

A report published a decade ago conveyed the results of a study aimed at determining if it were possible to achieve 1000X the computational power of the the Read more…

By Scott Gibson

Quantum Computing Will Never Work

November 27, 2018

Amid the gush of money and enthusiastic predictions being thrown at quantum computing comes a proposed cold shower in the form of an essay by physicist Mikhail Read more…

By John Russell

Cray Unveils Shasta, Lands NERSC-9 Contract

October 30, 2018

Cray revealed today the details of its next-gen supercomputing architecture, Shasta, selected to be the next flagship system at NERSC. We've known of the code-name "Shasta" since the Argonne slice of the CORAL project was announced in 2015 and although the details of that plan have changed considerably, Cray didn't slow down its timeline for Shasta. Read more…

By Tiffany Trader

IBM at Hot Chips: What’s Next for Power

August 23, 2018

With processor, memory and networking technologies all racing to fill in for an ailing Moore’s law, the era of the heterogeneous datacenter is well underway, Read more…

By Tiffany Trader

House Passes $1.275B National Quantum Initiative

September 17, 2018

Last Thursday the U.S. House of Representatives passed the National Quantum Initiative Act (NQIA) intended to accelerate quantum computing research and developm Read more…

By John Russell

Summit Supercomputer is Already Making its Mark on Science

September 20, 2018

Summit, now the fastest supercomputer in the world, is quickly making its mark in science – five of the six finalists just announced for the prestigious 2018 Read more…

By John Russell

AMD Sets Up for Epyc Epoch

November 16, 2018

It’s been a good two weeks, AMD’s Gary Silcott and Andy Parma told me on the last day of SC18 in Dallas at the restaurant where we met to discuss their show news and recent successes. Heck, it’s been a good year. Read more…

By Tiffany Trader

US Leads Supercomputing with #1, #2 Systems & Petascale Arm

November 12, 2018

The 31st Supercomputing Conference (SC) - commemorating 30 years since the first Supercomputing in 1988 - kicked off in Dallas yesterday, taking over the Kay Ba Read more…

By Tiffany Trader

CERN Project Sees Orders-of-Magnitude Speedup with AI Approach

August 14, 2018

An award-winning effort at CERN has demonstrated potential to significantly change how the physics based modeling and simulation communities view machine learni Read more…

By Rob Farber

Leading Solution Providers

SC 18 Virtual Booth Video Tour

Advania @ SC18 AMD @ SC18
ASRock Rack @ SC18
DDN Storage @ SC18
HPE @ SC18
IBM @ SC18
Lenovo @ SC18 Mellanox Technologies @ SC18
NVIDIA @ SC18
One Stop Systems @ SC18
Oracle @ SC18 Panasas @ SC18
Supermicro @ SC18 SUSE @ SC18 TYAN @ SC18
Verne Global @ SC18

TACC’s ‘Frontera’ Supercomputer Expands Horizon for Extreme-Scale Science

August 29, 2018

The National Science Foundation and the Texas Advanced Computing Center announced today that a new system, called Frontera, will overtake Stampede 2 as the fast Read more…

By Tiffany Trader

HPE No. 1, IBM Surges, in ‘Bucking Bronco’ High Performance Server Market

September 27, 2018

Riding healthy U.S. and global economies, strong demand for AI-capable hardware and other tailwind trends, the high performance computing server market jumped 28 percent in the second quarter 2018 to $3.7 billion, up from $2.9 billion for the same period last year, according to industry analyst firm Hyperion Research. Read more…

By Doug Black

Nvidia’s Jensen Huang Delivers Vision for the New HPC

November 14, 2018

For nearly two hours on Monday at SC18, Jensen Huang, CEO of Nvidia, presented his expansive view of the future of HPC (and computing in general) as only he can do. Animated. Backstopped by a stream of data charts, product photos, and even a beautiful image of supernovae... Read more…

By John Russell

Germany Celebrates Launch of Two Fastest Supercomputers

September 26, 2018

The new high-performance computer SuperMUC-NG at the Leibniz Supercomputing Center (LRZ) in Garching is the fastest computer in Germany and one of the fastest i Read more…

By Tiffany Trader

Houston to Field Massive, ‘Geophysically Configured’ Cloud Supercomputer

October 11, 2018

Based on some news stories out today, one might get the impression that the next system to crack number one on the Top500 would be an industrial oil and gas mon Read more…

By Tiffany Trader

Intel Confirms 48-Core Cascade Lake-AP for 2019

November 4, 2018

As part of the run-up to SC18, taking place in Dallas next week (Nov. 11-16), Intel is doling out info on its next-gen Cascade Lake family of Xeon processors, specifically the “Advanced Processor” version (Cascade Lake-AP), architected for high-performance computing, artificial intelligence and infrastructure-as-a-service workloads. Read more…

By Tiffany Trader

Google Releases Machine Learning “What-If” Analysis Tool

September 12, 2018

Training machine learning models has long been time-consuming process. Yesterday, Google released a “What-If Tool” for probing how data point changes affect a model’s prediction. The new tool is being launched as a new feature of the open source TensorBoard web application... Read more…

By John Russell

The Convergence of Big Data and Extreme-Scale HPC

August 31, 2018

As we are heading towards extreme-scale HPC coupled with data intensive analytics like machine learning, the necessary integration of big data and HPC is a curr Read more…

By Rob Farber

  • arrow
  • Click Here for More Headlines
  • arrow
Do NOT follow this link or you will be banned from the site!
Share This