HPC - High Performance Computing Blog

Name that supercomputer (Quiz)

2013-03-12T11:24:00.000Z

Instead of a sensible HPC blog post, how about some fun? Can you name these supercomputers?

I'm looking for actual machine names (e.g. 'Sequoia') and the host site (e.g. LLNL). Bonus points for the funding agency (e.g. DOE NNSA) and the machine type (e.g. IBM BlueGene/Q).

Submit your guesses or knowledgeable answers either through the comments field below, or to me on twitter (@hpcnotes).

For the photos, if you are stuck, you might need to use clues from my twitter stream as to where I have been recently.

Answers will be revealed once there have been enough guesses to amuse me. Have fun!

Which supercomputer are we looking underneath?

Acceptance of this leading system became a HPC news topic recently

NAG provides the Computational Science & Engineering Support Service for this one

One letter is all that’s needed to describe this supercomputer

Racing cattle powered by Greek letters

Spock was one of these

Which supercomputer does this photo show the inner rows of?

Memory with a deerstalker & pipe

Put an end to Ming (or did he)?

This plant/leaf is normally silver when used as the national symbol of this one’s host country

Predictions for 2013 in HPC

2013-01-11T17:19:00.000Z

As we stumble into the first weeks of 2013, it is the season for predictions about what the coming year will bring. In my case, following my recent review of HPC in 2012, I get to make some predictions for the world of HPC in 2013.

Buzzwords

First up, this year’s buzzword for HPC marketing and technology talks. Last year was very much the year of “Big Data” as a buzzword. As that starts to become old hat (and real work) a new buzzword will be required. Cynical? My prediction is that this year will see Big Data still present in HPC discussions and real usage but it will diminish in use as a buzzword. 2013 will probably spawn two buzzwords.

The first buzzword will be “energy-efficient computing”. We saw the use of this a little last year but I think it will become the dominant buzzword this year. Most technical talks will include some reference to energy-efficient computing (or the energy cost of the solution or etc.). All marketing departments will swing into action to brand their HPC products and services as energy efficient computing – much as they did with Big Data and before that, Cloud Computing, and so on. Yes, I’m being a tad cynical about the whole thing. I’m not suggesting that energy efficiency is not important – in fact it is essential to meet our ambitions in HPC. I’m merely noting its impending over-use as a theme. And of course, energy efficient computing is not the same as Green Computing – after all that buzzword is several years old now.

Energy efficiency will be driven by the need to find lower power solutions for exascale-era supercomputers (not just exascale systems but the small department petascale systems that will be expected at that time – not to mention consumer scale devices). It is worth noting that optimizing for power and energy may not be the same thing. The technology will also drive the debate – especially the anticipated contest between GPUs and Xeon Phi. And politically, energy efficient computing sounds better for attracting investment rather than “HPC technology research”.

The second buzz-theme that I see emerging this year, probably towards the second half of the year is “ease-of-use”. See - it’s only a buzz-theme not a buzzword because it’s not a catchy phrase yet. But someone will be inspired and a buzzword it will become.

I think that the drive for the public HPC community to engage with industry, together with the increasing recognition of the need to widen the adoption of HPC beyond high end niches, and the marketing requirements of technologies such as Xeon Phi will mean that ease-of-use will become a key aspect of the HPC debate in 2013. To have a sustainable growth and future, HPC has to move away from requiring specialist skill sets for users. For advanced users and programmers – this is OK – the technology is complex and rightly requires expertise. But for basic users who are primarily using HPC to do a job, e.g. in engineering, we need a better interface than batch scripts and a better way to extract performance than understanding the underlying technology architecture.

Technology choices

Less of a prediction, more of stating the obvious, but I’d expect the battle between GPUs and Intel MIC to continue through 2013 – with the benefits and issues of each approach being debated, demonstrated and challenged. The debate over when and how far to commit to accelerators/co-processors compared to traditional processor-only HPC systems solutions will continue. I know of several HPC procurements where accelerators/co-processors are still being avoided due to the need to ensure a solution that is optimal across a general purpose HPC service.

Many individual applications have been shown to achieve meaningful performance gains using GPU/MIC but when considering a large multi-user community, there will always be some applications that simply will not be ready for that architectural change for some years, or where a lack of parallel programming experts means that not enough of the application base can be converted yet. However, I do that think that manycore processors are an inevitable step along the HPC evolution and applications that avoid adopting manycore (whether GPU/MIC/other) now are simply delaying the day when they will have to evolve or be left behind. The harsh reality is that the hardware evolves and applications have to move with it. Rarely does the hardware ignore technology trends to swing back to application expectations. I am not sure GPU/MIC in their current forms are the final form of manycore processing for the exascale era – but I’m sure they are close enough to provide a valuable stepping stone that won’t be wasted effort and will deliver performance in the meantime.

Company moves

In terms of HPC companies, this is always a risky area to make predictions, but I’d guess two are probably well known enough that I can make predictions now and claim credit later for stating the bleedin’ obvious.

Intel spent much of the last few years swallowing parts of the HPC ecosystem to broaden its HPC capabilities. Prediction? Intel will continue to acquire key capabilities related to HPC and we will start to see their plans emerge for how they will exploit this ingested ecosystem.

Related to that, is Cray – they bought Appro towards the end of 2012 but by my sums they still have some money left over from the Intel interconnect deal – I’d expect Cray to invest some more of that money this year – perhaps in acquisitions, perhaps in technology research, perhaps in collaborations. Yes – I know I’ve covered my bases there – but predictions are awfully hard to get right if you don’t cover your bases!

I'd expect some major moves from a few other HPC companies too - but let's not be too ambitious with the predictions!

Industrial HPC

Finally, the start of 2013 saw the news stories around the break-up of Encanto. This was the supercomputer installed in New Mexico to engage with industrial users of HPC and encourage high tech industry to locate R&D in the area.

This theme, national lab / university HPC centre engaging with industry on HPC adoption, has been a long standing vision of HPC funders and centres across the globe. Only a few centres in the world have made a success of industrial partnerships, yet many new ones are able to convince their governments to invest in supercomputing facilities to bring supercomputing capabilities to industrial users and so generate positive economic impact. Why are the past failures not a barrier to this? Partly because it is like the lottery – success is hard to be sure of and most will fail - but the one that gets it right will have a huge pay-off.

If anyone does manage to make use of HPC technologies pervasive within their industrial base, then the long term economic benefits could be immense – especially if other engagement attempts continue to fail and thus a multi-sector competitive advantage opens up. But the economic and business models have to be right - not merely the technology evangelization.

Of course, bringing HPC benefits to industrial users is not purely the preserve of academic centres and national labs – there are many commercial organisations providing this support too. For example, my own company, NAG, provides services to industry to evolve and scale applications to benefit from modern HPC technology – whether multicore clusters or petascale supercomputers.

So ...

So that’s it then: energy-efficient, easy-to-use, and industrially engaged. 2013 looks fun already!

A review of 2012 in supercomputing - Part 2

2012-12-20T22:31:00.001Z

This is Part 2 of my review of the year 2012 in supercomputing and related matters.

In Part 1 of the review I re-visited the predictions I made at the start of 2012 and considered how they became real or not over the course of the year. This included cloud computing, Big Data (mandatory capitalization!), GPU, MIC, and ARM - and software innovation. You can find Part 1 here: http://www.hpcnotes.com/2012/12/a-review-of-2012-in-supercomputing-part.html.

Part 2 of the review looks at the themes and events that emerged during the year. As in Part 1, this is all thoroughly biased, of course, towards things that interested me throughout the year.

The themes that stick out in my mind from HPC/supercomputing in 2012 are:

The exascale race stalls
Petaflops become "ordinary"
HPC seeks to engage a broader user community
Assault on the Top500

The exascale race stalls

The global race towards exascale supercomputing has been a feature of the last few years. I chipped in myself at the start of 2012 with a debate on the "co-design" mantra.

Confidently tracking the Top500 trend lines, the HPC community had pinned 2018 as the inevitable arrival date of the first supercomputer with a peak performance in excess of 1 exaflops. [Note the limiting definition of the target - loosely coupled computing complexes with aggregate capacity greater than exascale will probably turn up before the HPC machines - and peak performance in FLOPS is the metric here - not application performance or any assumptions of balanced systems.]

Some more cautious folk hedged a delay into their arrival dates and talked about 2020. However, it became apparent throughout 2012 that the US government did not have the appetite (or political support) to commit to being the first to deploy an exascale supercomputer. Other regions of the world have - like the USA government - stated their ambitions to be among the leaders in exascale computing. But no government has yet stood up and committed to a timetable nor to being the first to get there. Critically, neither has anyone committed the required R&D funding needed now to develop the technologies [hardware and software] that will make exascale supercomputing viable.

The consensus at the end of 2012 seems to be towards a date of 2022 for the first exascale supercomputer - and there is no real consensus on which country will win the race to have the first exascale computer.

Perhaps we need to re-visit our communication of the benefits of more powerful supercomputers to the wider economy and society (what is the point of supercomputers?). Communicating the value to society and describing the long term investment requirements is always a fundamental need of any specialist technology but it becomes crucially essential during the testing fiscal conditions (and thus political pressures) that governments face right now.

Petaflops become "ordinary"

2012 was the year that petascale supercomputers became "ordinary" (in as much as a multi-million dollar megawatt-chewing complex technology can be "ordinary"). Certainly by the end of 2012 there were enough deployed supercomputers having a peak performance of one petaflops or more that we stopped being aware of exactly how many there were (33 according the Nov 2012 Top500 list). In fact, even the next order of magnitude (10+PF), is starting to be populated.

Another measure of petaflops moving beyond the domain of the leading national labs and academic supercomputer centers was the announcements by both BP and Total of petascale supercomputers deployed for industrial use. There may well be more supercomputers in industry that have not yet been announced.

And as a final measure of "ordinary", I perceive the HPC community now regards the technical challenges of petascale computing as a production support issue not a research topic for future technology. This, in spite of the many challenges left at petascale - only a tiny proportion of HPc users are successfully using petascale resources. But the allure of exascale is already taking the limelight in the R&D community away from the petascale reality.

HPC seeks to engage a broader user community

There was a growing realization during 2012 that the HPC community needed to take ownership of the issue of widening the user base. Although not a new theme, I did notice this crop up more often in panel discussions, remits for new supercomputer services/facilities, and conference talks. HPC has potential to a much broader audience than currently using the technology.

Leaving aside the reality that many users of HPC would not self-identify themselves as HPC people, there are three main growth directions that visionaries, funding bodies and practitioners have been pushing in 2012:

non-academic use of supercomputing (specifically industrial users);
more local and smaller HPC facilities in addition to the nationally leading systems - espeiclaly in non-trational user communities (i.e. not just chemists and physicists etc.);
future users - the skills and awareness investments needed now to prepare the future generations of scientists and engineers for a world of HPC opportunities.

I weighed in with my own thoughts on this theme in my 1000x article at HPCwire.

Whilst this theme has thankfully risen in prominence throughout 2012, I will still advocate we need to focus more of our community effort on this broadening of the HPC user base.

Assault on the Top500

Among the most debated events of the supercomputing world in 2012 was the news that NSF/NCSA Blue Waters has chosen not to submit the Blue Waters supercomputer to the Top500 list. The system would clearly have made the Top10, is one of the highest profile supercomputer projects in the world, and has been anticpated as a flagship scientific resource. So the rejection of a Top500 ranking was big news. Bill Kramer of NCSA set out the reasons in an exceptionally well thought out article listing not only the issues with the Top500 but also suggested solutions. There followed much debate on blogs, news sites and twitter - of which at least some missed the point in my opinion.

In my view, the Top500 is NOT broken - it is the use to which the Top500 data is put that is broken. The Top500 is useful because it is a set of 500 data points collected twice a year in a consistent manner for 20 years. It's value is as a data set - either of 500 points at any one collection, or as trends over time of subsets, etc.

It is not useful to base procurement, funding or architectural decisions on any single one of those data points alone. The ranking position of any one machine is meaningless - because it no longer correlates well with scientifc delivery. Note that the problem is not with the Linpack (HPL) benchmark per se - any single benchmark would have this issue of being unrepresentative of the whole workload/mission of a supercomputer.

Delivering real science and engineering with supercomputers requires far more than a good HPL result - it requires a production service wrapped around the machine - including hardware feature such as storage, software support and performance development, and skills support (training etc.). The Top500 does not measure these full service capabilities - only that a machine exists and has managed one hero run of HPL.

To me the key thing to remember is that the Top500 does have its uses - but only when used in aggregate or for trends - not when quoting a single position on a given release of the list.

Finally

This ends my (too long) review of 2012 in supercomputing - but the obvious follow on is to look towards 2013. Tune in to this blog after the Christmas season to find out my predictions for HPC/supercomputing in 2013.

In the meantime, feel free to chip in with comments and discussion on my review of 2012.

And, to all my readers (if any!) - Merry Christmas!

A review of 2012 in supercomputing - Part 1

2012-12-18T21:43:00.000Z

It's that time of year when doing a review of the last twelve months seems like a good idea for a blog topic. (To be followed soon after by a blog of predictions for the next year.)
So, here goes - my review of the year 2012 in supercomputing and related matters. Thoroughly biased, of course, towards things that interested me throughout the year.

Predictions for 2012

Towards the end of 2011 and in early 2012 I made various predictions about HPC in 2012. Here are the ones I can find or recall:

The use of "cloud computing" as the preferred marketing buzzword used for large swathes of the HPC product space would come to an end.
There would be an onslaught of "Big Data" (note the compulsory capital letters) as the marketing buzzword of choice for 2012 - to be applied to as many HPC products as possible - even if only a tenuous relevance (just like cloud computing before it - and green computing before that - and so on ...)
There would be a vigorous ongoing debate over the relative merits and likely success of GPUs (especially from NVidia) vs. Intel's MIC (now called Xeon Phi).
ARM would become a common part of the architecture debate alongside x86 and accelerators.
There would be a growth in the recognition that software and people matter just as much as the hardware.

So - how did those predictions match reality? (Or - how can I pretend I was right all along?)

Well, "cloud computing" is still with us as a label for a whole range of products and services that I think could be delivered just as effectively without that marketing label (and that already existed before "cloud" was invented). However, it is no longer being used as the first choice buzzword of HPC marketing departments. It is no longer guaranteed to turn up in any HPC panel discussion. "Cloud" has been replaced by "Big Data".

No-one can doubt the explosion of "Big Data" throughout the HPC community in 2012 - whether applied to all sorts of products that previously existed quite happily without a "Big Data" label, or mandatory mentions in panel discussions on any aspect of HPC at every HPC event, or sometimes even in a context that is actually relevant! The "must-mention-at-any-cost" aspect of "Big Data" threatened to hide that there was a real issue within - both in terms of technical challenges and business opportunities. Data (big, fast, complex, whatever) and HPC have always been natural partners but there is so much potential for more to come from this intersection.

In 2012, GPU computing eventually became normal. So much so in fact, that it has taken on the role of incumbent against the newcomer - Intel's Xeon Phi (I still think "MIC"). Anyone tracking the future of HPC either towards exascale computing or towards more energy & cost efficient individual HPC will know that some form of manycore seems to be inevitable as the base processor architecture. GPUs and MIC each have their own strengths - and neither are probably the final format that will see us through the exascale era. However, both are here now.

Much of 2012 was spent in anticipation of the launch of Xeon Phi. NVidia continued to march onwards with the announcement of Kepler products. Both sides worked the marketing and technical conversations to highlight the advantages of their manycore solution. Even AMD joined in at SC12 with the S10000 (how many zeros?). I think it is safe to say we haven't seen the end of this battle yet.

2012 also saw the emergence of ARM as a serious option in discussions around processors that could take us to both exascale computing and energy efficiency for individual-scale HPC systems. Several prototype/research projects using ARM for HPC progressed in 2012. Some vendors even announced ARM based servers for technical computing. ARM still needs to evolve in various aspects to catch up with x86 (and even GPUs) in HPC but the creeping of ARM into the HPC community has a quiet but firm momentum.

Finally

So, as you can see, I am claiming that all of my predictions for 2012 have turned out to be true! It is possible I am a visionary seeing the future of HPC. It is more likely I was stating the flippin' obvious and steering clear of anything controversial :-)

However, I have yet to address my last prediction: "a growth in the recognition that software and people matter just as much as the hardware". I think this has only partly come true.

Yes, there are people out there who have seen the light (I am biased) and have realized that, however powerful the hardware is, it requires effective software/applications and people to really deliver the best science and engineering output. (Or, as someone told me in an excellent phrase "to actually realize some impact based on having moved a lot of electrons around a big heat-generator".)

And the theme of software and people does turn up in discussion panels and opinion articles reasonably regularly. But there has still been a focus on measuring supercomputing efforts by the scale of the hardware - a supercomputer centre needs a big machine - a centre with a big HPC software team but only a small supercomputer doesn't yet carry the same impact/prestige in the community.

There has been a steady growth in the number of commercial and public sector organisations investing in HPC software innovation services and skills development/training. (Which is good - because that is what we [NAG] deliver - and business has grown nicely in 2012.)

However, I think there is still so much untapped potential in software innovation that users and providers of HPC services can exploit.

One of the strongest outcomes of NAG's HECToR Computational Science and Engineering (CSE) Support Service (the UK national supercomputing service) is a large body of evidence that has proved that investing in the application software alongside the hardware delivers much more science that investing the same total budget in supercomputers alone.

Next - the rest of 2012 and looking onwards to 2013

In Part 2 of this review, I discuss the themes and events that emerged during the year (i.e. the ones I didn't predict). Then, in the first few days of the new year, I will make my predictions for 2013 in HPC. Follow me on twitter (@hpcnotes) to catch the next blog posts - and for my general tweets on all things HPC.

HPC notes at SC12

2012-11-08T15:11:00.000Z

I'll be at SC12 next week.

I have a mostly full schedule in advance but I always leave a little time to explore the show floor, and to meet new people or old friends.

If you are at SC12 too, you might be able to find me via the NAG booth (#2431) - or walking the streets between meetings - or at one of the networking receptions.

If you are a twitter person - you can find me at @hpcnotes (but be warned I won't be tweeting most of the HPC news during the show - @HPC_Guru is much better for that).

Hope to see some of you there. And rember my tribute from last year to those not attending SC.

HPC fun for SC12

2012-11-06T14:21:00.004Z

I've previously written some light-hearted but partly serious pieces for the main supercomputing events.

I'm working on one for SC12 too - again to be published in HPC Wire - but in the meantime, here are pointers for the SC11 and ISC11 articles:

The making of “1000x” – unbalanced supercomputing

2012-10-12T16:47:00.002+01:00

I have posted a new article on the NAG blog: The making of "1000x" – unbalanced supercomputing.

This goes behind my article in HPCwire ("Chasing1000x: The future of supercomputing is unbalanced"), where I explain the pun in the title and dip into some of the technology issues affecting the next 1000x in performance.

The first mention of SC12

2012-10-02T13:43:00.000+01:00

It's that time of year again. SC has started to drift into my inbox and phone conversations with increasing regularity - here comes Supercomputing 2012 in Salt Lake City. Last year, in the run up to SC11 in Seattle, I wrote the SC11 diary - blogging every few days on my preparations and thoughts for the biggest annual event of the supercomputing world.

I'm not sure I'll do such a diary again this year (unless by popular demand - not likely!). However, I will be writing some articles for some publications (HPC Wire and others - see my previous articles) in the coming weeks which will set the scene for SC from my point of view - burning issues I hope will be debated in the community, key technology areas I will be watching, and so on.

In the meantime, if you crave SC reading material, you might amuse yourself by reading my previous fun at SC time (e.g. The top ten myths of SC - in HPC Wire for SC11) or you might even want to translate my fun from ISC (Are you an ISC veteran?) to new meanings at SC.

If you want more serious content, then browse on this blog site (e.g. tagged "events") or on the NAG Blog (e.g. tagged "HPC").

If you find nothing you like - drop me a comment below or via twitter and I'll see what I can do to address the topic you are interested in!

What is the point of supercomputers?

2012-08-02T15:30:00.001+01:00

Maybe it seems an odd question to ask on a blog dedicated to High Performance Computing (HPC). But it is good to question why we do things – hopefully leading us to a clearer justification for investing money, time and effort. Ideally, this would also enable better delivery – the “how” supporting the “why” – focusing on the best processes, technologies, etc. to achieve the goals identified in the justification.

So, again, why supercomputing? Perhaps you think the answer is obvious – supercomputing enables modelling and simulation to be done faster than with normal computers, or enables bigger problems to be solved.

First, for anyone (even seasoned HPC people) trying to explain and evangelize HPC or supercomputing to other scientists, potential new users, funding bodies, politicians – well, anyone really – I can firmly recommend John West’s talk at the HPCC 2012 conference as an excellent use of your time.

Gary Johnson wrote a great article in HPC Wire recently, discussing the case for continually investing in the biggest supercomputers in the world, and raising the possibility that perhaps the world’s largest supercomputers are becoming too far detached from the actual needs of the scientists and engineers who are used to justify these large procurements.

However, I was recently prompted to think of the reason for supercomputing as comprising two related but different opportunities.

The first opportunity is about enabling faster, bigger, better simulation and modelling – supporting scientists and engineers to be more productive (faster results, search more parameter space, etc.), and to exploit more accurate computer predictions (bigger, more complex,) in their research, design and testing processes. The business need or scientific case for bigger, better, faster computational modelling is usually readily made and, to many, that alone is enough to answer "why supercomputing?". Gary's HPC Wire article discusses this case, and the resulting different ways to deliver such a supercomputing resource.

The second type of opportunity is about enabling new or game-changing capabilities. At this year’s NCSA Private Sector Program annual meeting, I was joined by Gerry Labedz on a panel moderated by Bill Gropp, on the topic of modern software for supercomputing – see here for my summary of the panel and a link to the video. Among other things, in this panel discussion, we spoke about the idea that supercomputing is about enabling dreams.

This concept of radical capability is a critical factor when considering why supercomputing is important to a business, to the economy, to science, etc. It does not, necessarily, mean using the most powerful supercomputers in the world. For a researcher or business previously using a desktop computer for modelling/simulation, being able to exploit the power of a few hundred cores in a small cluster could be game-changing.

This happens when the new computational power opens up not just faster results but new ways of approaching a problem. In desperate search of an analogy, consider air travel. For many users, it is a faster and better way to get from A to B than driving. For other users, it enables them to do something they just couldn't do by driving. This might be reaching an overseas destination. It might be the speed of air travel makes a specific vacation possible because they don't have enough holiday entitlement to take the time get to the vacation destination by slower means of transport (car/ship). Thus, the same platform can deliver very beneficial improvements in operational performance to some users, whilst enabling entirely new experiences to other users.

And so to supercomputing. It is not the size of the supercomputer that distinguishes between the "bigger, better, faster" and "new capabilities" benefits, but what it means to the business processes or research methods of a given user. There is obviously a grey line between these two types of supercomputing opportunity but, in my view, the difference is real - and is measured in terms of the impact on the business or research agenda, not the scale of computing resources used.

Supercomputers are for dreams

2012-06-15T16:21:00.000+01:00

I was invited to the 2012 NCSA Annual Private Sector Program (PSP) meeting in May. In my few years of attending, this has always been a great meeting (attendance by invitation only), with an unusually high concentration of real HPC users and managers from industry.

NCSA have recently released streaming video recordings of the main sessions - the videos can be found as links on the Annual PSP Meeting agenda page.

Bill Gropp chaired a panel session on "Modern Software Implementation" with myself and Gerry Labedz as panellists.

The full video (~1 hour) is here but I have also prepared a breakdown of the panel discussion in this blog post below.

Bill's Panel Overview and Introduction

The core premise of Bill's panel theme was that supercomputing is "not just about the solver" - imagine we had a perfect solver - what else must we address? Bill intro takes the first ~8 minutes of the session.

Gerry's Introduction and Comments

Gerry then talks from 08:51 to 24:18, giving an introduction and overview of this thoughts on the topic. These are worth listening to, as Gerry makes some outstanding points.

Andrew's Introduction and Comments

I then speak from 24:18 to 32:34, giving an introduction and overview of my thoughts on the panel theme. I think these are worth listening to also :-)

Questions and Debate

Then follows approximately half an hour of questions and debate including:

waiting for silver bullets [32:35]
leaking vs. communicating, including HECToR Distributed CSE Support [37:23]
creativity styles in engineering - evolutions vs revolution [39:34]
tools for supercomputing - help the heroes or the masses? [45:41]
supercomputing as a possible leader of computer science [48:41]
modern programming languages (what is wrong with Fortran?) [51:28]

The panel discussion raises some good points, has a few moments of humour (e.g. running away when Fortran is mentioned), and generates some excellent quotes on supercomputing, including Gerry's brilliant encapsulation of one of the main concepts I was trying to get across: "supercomputers are for dreams" [43:13], plus Gerry's "only the heroes survive" [42:51].

Enjoy watching the panel video. Let me know what you think - via the comments page on this blog, on twitter (@hpcnotes), or in person (e.g. at ISC12 in Hamburg next week). Oh, and enjoy the other videos too - there are many more gems from this top quality event.

Some fun for ISC12

2012-06-06T14:55:00.003+01:00

I have written a guest blog post for the ISC'12 website - "Are you an ISC veteran?". The article is intended to raise a few serious observations amongst the fun.

I also wrote an earlier guest blog post for the ISC'12 website - "Is co-design for exascale computing a false hope?"

I've added these two links to my page on this site "Interviews, Quotes, Articles" (which lists my various articles, interviews, etc. in other locations around the internet).

The power of supercomputers - energy, exascale and elevators

2012-05-30T18:06:00.001+01:00

Paul Henning has written on his blog (HPC Ruminations) about the growing issue of power requirements for large scale computing. Paul's blog post - "Familiarity Breeds Complacency" - is partly in response to my article at HPCwire - "Exascale: power is not the problem" and my follow-up disucssion on here - "Supercomputers and other large science facilities".

Paul makes several good points and his post is well worth reading. He ends with an observation that I've noted before (in my own words):

One of supercomputing's biggest strengths - it's ability to help almost all areas of science and engineering - is also one of it's greatest weaknesses - because there a portfolio of cases rather than a single compelling champion to drive attention and investment.

PS - I've added Paul's new blog to my list of HPC blogs and news sites.

Looking ahead to ISC'12

2012-05-25T16:13:00.004+01:00

I have posted my preview of ISC'12 Hamburg - the summer's big international conference for the world of supercomputing over on the NAG blog. I will be attending ISC'12, along with several of my NAG colleagues. My blog post discusses these five key topics:

GPU vs MIC vs Other
What is happening with Exascale?
Top 500, Top 10,
Tens of PetaFLOPS
Finding the advantage in software
Big Data and HPC

Read more on the NAG blog ...

Co-design for exascale

2012-03-29T20:16:00.001+01:00

I wrote a blog for the ISC website on co-design for exascale.

This has also been mentioned on InsideHPC here.

I made similar comments at the panel hosted by Thomas Sterling at the HPCC conference in Newport, RI earlier this week.

The video of this panel should be posted soon at InsideHPC soon.

HPC Insiders - The Newport Gathering

2012-02-09T17:45:00.002Z

The warm up for the annual HPCC meeting in Newport RI (March 26-28) has started - Are You an HPC Industry Insider?.

"The National High Performance Computing and Communications Conference (NHPCC) will highlight several exciting changes this year. Also known as the Newport Conference, the elite gathering that started 26 years ago as a one-day event to bring vendors together with government agency personnel has expanded its focus this year to include a more global perspective."

"Another significant change this year is the emphasis on manufacturing and competitiveness."

I have a page on this blog site the main HPC events of the year. Many people have rightly remarked that the HPC community really is that - a community - and that there is still a relatively high degree of connection between the various practitioners. In other words, despite its growing size and global reach, it feels like a small community. People know each other. Consequently, networking, whether technical or commercial, goes a long way to helping your business. Whatever your scale of technical computing, from multicore workstations to multi-thousand-node supercomputers, getting involved with the active HPC community can help you with your parallel computing goals. Online resources can help, but by far the most effective way of benefiting from the wider HPC community is by participating at the right events.

I enjoy this Newport event - I think it is one of the best annual events for the HPC community - and am looking forward to great discussions and meeting the many friends in the international HPC community. See you there!

Cloud computing or HPC? Finding trends.

2012-01-19T18:42:00.000Z

I posted "Cloud computing or HPC? Finding trends." on the NAG blog today. Some extracts ...

Enable innovation and efficiency in product design and manufacture by using more powerful simulations. Apply more complex models to better understand and predict the behaviour of the world around us. Process datasets faster and with more advance analyses to extract more reliable and previously hidden insights and opportunities.

... and ...

High performance computing (HPC), supercomputing, computational science and engineering, technical computing, advanced computer modelling, advanced research computing, etc. The range of names/labels and the diversity of the audience involved mean that what is a common everyday term for many (e.g. HPC) is an unrecognised meaningless acronym to others - even though they are doing "HPC".

... and then I use some Google Trends plots to explore some ideas ...

Read the full article ...

My SC11 diary index

2011-11-08T22:59:00.000Z

Here is a handy index to the ten entries in my diary of the weeks leading up to SC11.

Enjoy.

My SC11 diary 10

2011-11-04T18:36:00.001Z

It seems I have been blogging about SC11 for a long time - but it has only been two weeks since the first SC11 diary post, and this is only the 10th SC11 diary entry. However, this will also be the final SC11 diary blog post.

I will write again before SC11 in HPC Wire (to be published around or just before the start of SC11).

And, then maybe a SC11 related blog post after SC11 has all finished.

So, what thoughts for the final pre-SC11 diary then? I'm sure you have noticed that the pre-show press coverage has started in volume now. Perhaps my preview of the SC11 battleground, what to look out for, what might emerge, ...

Let's start with processors. GPUs are no longer the new kid on the block - they have been hyped and evaluated long enough to be considered part of the HPC landscape now. ARM and DSPs have replaced GPUs as the emerging and potentially disruptive HPC technology to find out about, with promises of low power, better FLOPS/Watt, cheaper HPC, etc.

Even Intel is getting some attention in the "new" architectures arena with MIC. Intel's manycore product family (Knights Corner) is nearing availability, and has already won some deals (e.g. Stampede at TACC). NAG's own experience so far in the MIC early evaluation program is that the MIC is relatively easy to port code to.

And, of course, the traditional CPU teams, Intel Xeon (e.g. Sandy Bridge) and AMD Opteron (e.g, Interlagos) will be making their case as the HPC processors to choose. AMD's Bulldozer/Interlagos is sufficiently different from previous processors (shared FP unit, FMA4, etc) to be thought of as "new".

We have CPUx86 vs MIC vs GPU vs ARM vs DSP. So, the processor/architecture battleground looks vibrant and interesting. Let's see who emerges as the popular candidate after SC11.

What about software? Well, almost all of those architectures will require developments to your software in order to effectively exploit the new processor technology. For MIC, the porting looks OK, but vectorization is needed for good performance. For GPUs, it is not just a case of re-writing in CUDA or OpenCL, but of reconsidering the choice of algorithms used to extract much more parallelism at the node level. If that works, then performance can be very good. For ARM - who knows yet? But software (portability, performance, ease-of-performance, etc.) will be the key, no doubt.

Almost independent of your choice of processor, scale will still be an issue. Whether hundreds of GPU accelerated nodes or thousands of pure CPU nodes, software will have to scale to get the performance. Tools and expertise to support this will be critical. What software tools will emerge from SC11 as important players in the development of scalable software - from desktop to multi-petascale? Well Allinea is one example that has been making some pre-show noise, but there are others. Also worth keeping an eye out for news on the expertise - programs to develop and encourage careers in HPC, training opportunities, recruitment, etc. [Commercial message: NAG can help with expertise - provide parallel programmers to scale your applications, training, consulting, ...]

Big Data. HPC Analytics. SC11's theme is related to big data and HPC - and the pre-show noise seems to have produced plenty of content on big data and analytics using supercomputers. In my cynical days, I simply say this is merely the new buzz. Cloud is too blah now (been on the marketing agenda for too long). Same for Green. Both are still very relevant to HPC, but they aren't "fresh". So along comes HPC Analytics or Big Data. Massive and often volatile data sets generated by or processed by supercomputers is both a key challenge and an opportunity. One well worth looking out for solutions to at SC11. Products and solutions for large data will be a big part of SC11 exhibition and discussions. As will the hardware and software for processing massive data. R&D in this area will be important too. But I fear there will be some over-use of the "big data" topic, as with cloud and green, which will only work to obscure the real problems and potential.

Rumours. The fun bit of SC. What news or gossip will break during SC11? Key people moving between companies. Large deals announced. The replacement for IBM Blue Waters? The next big non-USA supercomputer. A new processor? Acquisitions? Sometimes the Top500 provides the big news and the surprises. I have heard whispers of some some things that may emerge, and there is always scope for unexpected announcements, leaks, or gossip.

What else? The "missing middle". Broadening HPC to organizations and users outside the big labs, supercomputer centers and large companies. This important issue might make headway at SC11. I will be involved in several discussions around this topic (including presenting on the Intel booth). And, let's not forget Exascale - I'm sure that will get plenty of attention at SC11 too.

What are you looking out for? What do you think will emerge as winners and losers?

Finally, if you can, make sure you catch the HPC Analyst Crossfire session Friday - I was a panel member at ISC11 and it was good fun - but more importantly, Addison Snell has pulled together an excellent panel for SC11 and it will be well worth your time.

I'll have a busy SC11 but if you want to find me to hear more about my opinions on the HPC world, berate me over a blog post, or learn more about NAG's HPC services & consulting, then please do get in touch.

See you in Seattle.

My SC11 diary 9

2011-11-02T19:14:00.001Z

I mentioned yesterday about preparation for SC11. I thought today I'd add a few miscellaneous tips I've gathered over the years. In no sensible order or grouping ...

First off, print a copy of your schedule before you travel to Seattle. Yes, I mean print - actual paper. I know we all have our schedule on the various electronic status symbols we carry around with us. But when their batteries decide to have a holiday, or when their internet connection won't play - you'll be glad of that piece of paper. I know it will be out of date within minutes of printing. Write on it - use a pen - that's a thing a bit like a smartphone stylus but ink comes out the end. Even an out of date printout is better than none when you need to check in a hurry where you should be next.

Unless you know Seattle well, print a map of the area around the convention centre - the main hotels etc. Put it in a water-resistant plastic slip (see comment yesterday about the weather). By the end of the SC11 week, you will know your way round perfectly and won't need this. But you might do earlier in the week. Don't rely on your smartphone and Google/Bing/etc maps - see paragraph above!

Hotels (for those of you who stay in them - some of us only get B&B's) - keep an eye out for bargains for SC12 hotels. Yep, as soon as you return from SC11, check out the SC12 city - if you know you will be at SC12 there may be some bargains on hotels (maybe flights too). Does't always work, but in the past I have stayed at some nice hotels at SC for a fraction of the price others paid using this method. Especially valuable in those cities where there are only a few hotels close to the convention center. In this case, securing proximity is as useful as securing a good price.

Food. Pack some snacks (or figure out where to get some quickly and easily on the move). Schedules can get so bad that lunches especially get lost unless they are booked meeting-meals. In that case, extracting your favourite chocolate bar from your pocket/laptop case/backpack can provide an essential energy boost during the day. Don't go overboard though - I am aware of one person who ensure they have a full supply of Coke Zero - even if it means wheeling a dozen bottles around all day in an executive looking travel bag (you know who you are! and PS - thanks for being so generous as to donate some to my thirst on occasions in the past!).

Party-hopping. There are many more evening parties, receptions, dinners, networking events and professional get-togethers than days in the week. So you will have a choice of several every night. Attending more than one per night can be done. In fact it is probably efficient in terms of work networking to plan more than one venue per night to maximise the number of people you meet. But don't attempt so many events per night that the time in each is too short to be worth it (either socially, professionally, or in terms of consumption).

Meetings - get local contact details (cell-phone etc.) for important meetings so you can call ahead when you lose your time-keeping, or your map, or something else comes along to knock your carefully planned schedule into chaos. Likewise, make sure people meeting you know how to call you to grovel that they are running late (or, if it's you that's late, to query where you are and why aren't you there yet).

You will not know everyone who says hello to you. We are all familiar with the experience of being "recognized" by someone, without being able to return the compliment. Because a good proportion of the HPC community will be at SC, this will happen even more at SC. In this case, you can play along, hoping for a clue from the conversation as to who exactly this person is and whether they know you (and whether you want to talk to them!). Or you can bluntly play the "I know too many people" card - "I'm sorry, I'm terrible with faces - and names - and memory - who are you?". Not the ideal approach if you can help it, but is the fastest way of getting the conversation round to meaningful content. Or, if feeling mischievous, stop a random SC11 attendee in the exhibit hall/street, "recognize" them and see how long you can keep it up :-)

Final tip - don't write blogs that are too long - so I'm stopping here!

My SC11 diary 8

2011-11-01T18:17:00.001Z

It turns out I have to actually do some talking at SC11 as well as listen to others. So one of today's jobs was to start preparing some presentations I will be giving at SC11. My normal habit is to have a custom version of a slide set for each audience/customer. I try to avoid simply re-using the same slide deck for each talk. Obviously I do re-use large chunks of previous presentations but update it, or add/remove content to get the right focus.

For the poor vendor speakers who have to give the same roadmap presentation a dozen times a day, each day of the SC11 week, each with full enthusiasm to different set of faces, I fully accept that the slide set is understandably almost unaltered between customers. I also salute their professionalism in conveying the same enthusiasm for the roadmap technologies and products by the 30th run-though at the end of the week! As I try to assimilate as much info as I can for advising customers on procurements and technology aspects of HPC service and application planning, I will bear in mind the repetitive job of the speaker on the other side of the table!

Thankfully, my own speaking efforts are much more varied. I will be speaking in various public and non-public venues. The hard ones are the short slots - trying to get a credible, useful and hopefully memorable message across in 10-15minutes is hard work. Among other things, I will be presenting about NAG's HPC services and consulting - how they can help new adopters of HPC (related to the "missing middle" agenda).

The easy ones are the ones that are not recorded - don't have to worry as much about making mistakes. but more importantly, it is also easier to give a good talk when you interact with the audience in front of you - but when being recorded, you have to bear in mind the audience that is not visible - and is giving not feedback/interaction.

Thankfully, my week is fairly evenly split between listening and talking modes - with the latter mostly meeting or conversation format rather than presentations/interviews etc. so not so much preparation of slides this year as some in the past. Although, preparation for SC11 is more than just writing presentations - I need to make sure I am up to date with the material for each meeting (whether customer, collaborator, vendor, etc.).

And the other preparations - I did book a flight didn't I? (check) And a hotel. See previous diary entry about luxury at SC. Better print out some stuff - maybe a map, schedule (although it will be out of date before the ink is dry) and so on. Currency - for the non-US attendees - better check how many dollars are left in my wallet from my last trip to the USA.

Apparently it is worth checking the weather forecast for Seattle too. Although the climate is fairly similar to the north of England where I am now, so not much adjustment will be needed. Wet, not so warm and a bit windy probably.

My SC11 diary 7

2011-10-31T16:28:00.001Z

As a cursory glance at #SC11 on twitter today will tell you, it is now only two weeks until SC11 (or less if you count the parts of the show that start over the weekend).

So perhaps this is a good time to consider the many supercomputing people who won't be joining the hordes in Seattle this year.

I wrote in one of my earlier SC11 diary entries: "... it is wise to not forget the many people using, supplying or supporting HPC products and solutions that won't be attending SC - in many cases because they are not even aware of SC11 and what it offers" and went on to talk a little about those people who did HPC or who could benefit from HPC but weren't aware of SC.

But as well as these unfortunate folk, we must bear in mind the even more unfortunate people - those who are only too aware of SC but who will be relegated to watching from afar. Some will have been in previous years, but it is not their turn this year maybe. They will linger around the home office, watching their email streams quieten, their phones calls and voice-mails contain only tantalizing messages of "are you in Seattle - shall we meet you for a drink?" They will grudgingly have to trust colleagues to handle meetings with customers, prospects, collaborators, suppliers, etc. Maybe they will try to keep up via twitter, blogs, news outlets, etc.

Perhaps even now they are painfully being asked to help with the organization of their SC11 booth, demos or meetings. Only to never see for real the effect of their work.

They will be adrift from the casual conversations and chance encounters that really create and shape collaborations and new deals. Even career opportunities.

They will be forced to only dream of the most wonderful banquets, exquisite entertainment and palatial venues that you will enjoy every night [yes, it does look that way from afar, even if from close up on the night it looks less like the marketing brochure promises :-)]. And yes, I know it's "work" or "networking" - but with a glass of wine in each hand and cheeks full of tasty snacks, you're not so convincing on the "it's work" thing :-)

Oh yes, they will be missing out - professionally and socially.

So, when you arrive in Seattle, knackered from long hours crammed into economy ~~cage~~ seat at the furtherest back end of an long haul flight across the Atlantic/Pacific/continent, mind-numbed from a day in the "air travel bubble" of boarding queues, connecting flights, immigration queues, etc., only to check into a hotel that is considerably less luxurious than your family will imagine, and looking forward to limited sleep for a week - before complaining too loudly, just think - at least you are at SC11 with all the opportunities that brings, not at home.

Someone remind me too, as I try to persuade my body to ignore it's natural clock stuck at the wrong end of the day (aren't timezones fun!).

On the other hand, away from the professional impact - a week not at SC11 is a week not being away from the family.

My SC11 diary 6

2011-10-27T22:39:00.000+01:00

Today was supposed to be a day away from the email, laptop, phone, etc. But it didn't quite turn out that way. Among other things, SC11 planning required some attention. Will try harder tomorrow (there won't be a diary entry tomorrow for example).

Which raises a question - do you find time for a day off at SC? Some people arrive over the weekend and take a day away from supercomputing to do some local tourism. Others stay on an extra day or two after the end of SC for the same reason. Personally, unless flight schedules force an extra day or two, I don't normally do this.

Sometimes it makes for good anecdotes though. I recall when we had SC in Phoenix (year anyone?) some colleagues arrived a day or two early. On their off day, they spent a couple of hours driving up to the Grand Canyon to see the sights. Unfortunately, there was heavy fog and they didn't see a thing. A couple of hours drive back and the day wasted - but for the anecdote to be used at the dinners for the rest of the week.

You also get the hotel anecdotes. Pittsburgh was bad for this (year?). There was a sports game in town (Robbers? Stealers? Some such team name :-)) in the days just before SC11 and there was a lot of overbooking of hotel rooms. Cue chaos of hotel receptions filled with people arguing and fighting for rooms. Sales folks smiling awkwardly as they got the last room, only to watch one of their big customers be "walked" to a hotel 20 miles away from the convention center.

Such anecdotes are the fun of travelling that is our HPC world. We just hope they are happy anecdotes or at least ones that work out OK in the end.

My SC11 diary 5

2011-10-26T17:09:00.000+01:00

I enjoy seeing the reactions of people attending SC for the first time. Perhaps being used to only attending other HPC events around the world, they are unprepared for the scale of the event.

Especially the exhibition. I am sure people get lost in there - properly lost, not just the few minutes disorientation that we all get several times in the SC11 week as we traverse the show floor looking for a specific booth or exit. Each year, I remember by about the 3rd day (but not the 1st!) that the booth numbers are partly logical - they are related to the rows/columns of the booth location within the hall. How handy for navigation!

The technical program can be daunting too - half way through the week you can still be finding entirely new unvisited corners of the convention complex where a workshop or BoF has decided your eligibility to attend will be judged by your ability to find it :-).

Or worse, for those who don't have full conference passes (e.g. only exhibitor, or technical program but not tutorials, etc.) suddenly find their way barred by a polite guardian who is happy to confirm that yes, the exit to the building is just 20 feet in front of you, but I'm sorry you can't pass this rope to get to it, because a room 200 yards around the corner had a tutorial in it yesterday and you don't have a tutorial badge.

OK, maybe a personal experience the previous year affected my last paragraph!

And then people can still be surprised by the amount of off-site meetings taking place - most of the meeting rooms in the major hotels within several blocks of the convention center will be booked out for the week. Many of these are used by the vendor community to host a series of meetings for customers and potential customers to learn more about their products, technologies and plans.

Finally, the many social (sorry networking) events. With so few days and so many choices of high quality networking reception, private dinners, etc., it becomes hard to select the right events to attend, or the right people to take dinner with. Sometimes duty pushes the choices one way. Sometimes we get to choose the company of people who look likely to make an enjoyable evening. Occasionally the two even combine :-) I am lucky this year to have mostly the latter (enjoyable) planned.

There is so much to participate in at SC11 that there is sometimes a sense of responsibility to choose wisely and make the best of it all; but also it is worth remembering that, with so many options, there is plenty of scope to both do good work and have a good time.

My SC11 diary 4

2011-10-25T17:52:00.000+01:00

Another good day with the SC11 schedule today - a good increase in the proportion of nailed down meetings.

Something that comes up during the process of arranging meeting at SC is the different logistics. There seems to be quite a variety of opinion as to the best way to attend SC week.

Some people will try to compress it - arrive in Seattle Monday afternoon, and leave Wednesday evening. Basically cutting it down to two days of intensive meetings. This works in a way, but misses out on the many events and workshops arranged around the core days, and misses out on the many social and networking opportunities - which are arguably one of the most important parts of the week.

Others will take the opposite extreme - arrive in Seattle Thursday before the SC11 week, and stay right through the week until the weekend after the show. That is a long time to be at one event, but it allows the maximum attendance of the various workshops arranged before and after the main event.

I know of some people (especially from outside the USA) who arrange other meetings in the USA the week before or after SC. This makes for a long time away from home - maybe a fortnight but at least it cuts down the travel, as the two meetings share the same long haul flight.

In fact I even know someone who does the opposite - a European who arranges meetings in Asia the week adjacent to SC, making one Round-the-World trip out of it all. There has to be some frequent flier miles aspect to that :-)

For myself, I try to find a balance between as short a time away from home as possible and attending an optimum amount of meetings, networking events and conference activities.

What's your habit? Maybe tell us in the comments section of this post?

My SC11 diary 3

2011-10-24T17:55:00.002+01:00

Well, shock, today so far has not been dominated by SC11! "Normal" work (and admin) has been the focus so far today. It is easy at this time of year to scan the headlines in the main HPC news outlets such as HPC Wire, InsideHPC, twitter (!), ... and assume SC is the only thing the HPC world is thinking of right now. The same is true of article preparation emails circulating for specialist publications like The Exascale Report. And it is even true to some extent for publications with a broader remit - e.g. Scientific Computing.

It is true, SC does dominate the thinking of many in the world of high performance computing at this time of year. And rightly so in many contexts. As I have already noted, it is a very focused week of meetings with existing and potential partners.

The HPC press will become swamped by SC11 related content - in fact many will switch to SC "special editions" for the duration of the event itself. Anyone thinking of press releases or other news disseminations must take this into account - release the story now and be forgotten by the time people come together to discuss at SC11 itself - or release during SC11 week and risk being lost among the SC11 week's deluge of supercomputing news releases?

It shows up in other conversations too - quite often over the last month or two meetings conclude with "we can catch up on that at SC in Seattle ...".

But it is wise to not forget the many people using, supplying or supporting HPC products and solutions that won't be attending SC - in many cases because they are not even aware of SC11 and what it offers. There are a lot of researchers, engineers, and organizations using computing technology to deliver performance and capability advantages to their research/business who do not self-identify what they do as "HPC". (I borrow this insight from Addison Snell of Intersect360.)

In addition, there are many users of technical computing who could greatly benefit from adopting HPC technologies or solutions but do not use HPC at all - often because they have not yet been engaged by the HPC community (e.g. SC11). These are a core part of the "missing middle" (or similar terms) talked about by many in the world of high performance computing recently - notably Intel, the Alliance for High Performance Digital Manufacturing, and others. (Even NAG - some of our HPC services are designed to train, advise and support organisations that are new to HPC.)

So, inevitably, SC11 does not represent the whole world of high performance computing, but it does a very good job of gathering together as much of the supercomputing ecosystem as possible. And thus it is worthy of being the default context for so many HPC conversations and supercomputing news stories in the next few weeks.