Showing posts with label supercomputers. Show all posts
Showing posts with label supercomputers. Show all posts

Friday, 22 May 2020

What makes a Supercomputer Centre a Supercomputer Centre?

When is a Supercomputer Center not a Supercomputer Center?

The world of HPC has always been a place of rapid change in technology with slower change in business models and skill profiles, but what actually makes a supercomputer center a supercomputer center?

Tin (or Silcon maybe)

Is it having a big HPC system? How big counts? Does it matter what type of "big" system you have?

Does it matter if there is not one big supercomputer but instead a handful of medium sized ones of different types?

Does it count if the supercomputers are across the street, or in a self-owned/operated datacentre the other side of town? What if the supercomputers are located hundreds of miles away from the HPC (eg to get cheap power & cooling)?

Who and How

Or is it having a team of HPC experts able to help users? How many experts? What level of expertise counts? How many have to be RSE (Research Software Engineer) types?

Is it having the vision and processes to recognise they are primarily a service provider to their users ("customers") rather than thinking of themselves mainly as a buyer of HPC kit?

What if you mainly have AI workloads rather than "traditional" HPC? What if you only run many small simulation jobs and no simulations that span thousands of cores? What if users only ever submit jobs via web portals and never log in to the supercomputers directly?

Is it essential to have a .edu, .gov, etc. address? Or can .com be a supercomputer center too?

This but not that?

If you have no supercomputers of your own, but have 50 top class HPC experts who work with users on other supercomputers and also research future technologies - is that a supercomputer center?

If you have a very large HPC system but only the bare miuminm of HPC staff and no technology R&D efforts - is that a supercopmputer center?

Which of the last two adds more value to your users?

Declare or Earn?

Is it merely a matter of declaration - "we are a supercomputer center"? Or it is a matter of other supercomputer centers accepting you as a peer? But then who counts as other supercomputer centers to accept you? What if some do and some don't?

Is there a difference between a supercomputer center and a supercomputing center?

What do you think? And does your answer depend on whether you are a user, or work at a "traditional" supercomputer center, or a new type of supercomputing center, or a HPC vendor, or from outside the HPC field?

Monday, 31 July 2017

HPC Getting More Choices - Technology Diversity

HPC has been easy for a while ...

When buying new workstations or personal computers, it is easy to adopt the simple mantra that a newer processor or higher clock frequency means your application will run faster. It is not totally true, but it works well enough. However, with High Performance Computing, HPC, it is more complicated.

HPC works by using parallel computing – the use of many computing elements together. The nature of these computing elements, how they are combined, the hardware and software ecosystems around them, and the challenges for the programmer and user vary significantly – between products and across time. Since HPC works by bringing together many technology elements, the interaction between those elements becomes as important as the elements themselves.

Whilst there has always been a variety of HPC technology solutions, there has been a strong degree of technical similarity of the majority of HPC systems in the last decade or so. This has meant that (i) code portability between platforms has been relatively easy to achieve and (ii) attention to on-node memory bandwidth (including cache optimization) and inter-node scaling aspects would get you a long way towards a single code base that performs well on many platforms.

Increase in HPC technology diversity

However, there is a marked trend of an increase in diversity of technology options over the last few years, with all signs that this is set to continue for the next few years. This includes breaking the near-ubiquity of Intel Xeon processors, the use of many-core processors for the compute elements, increasing complexity (and choice) of the data storage (memory) and movement (interconnect) hierarchies of HPC systems, new choices in software layers, new processor architectures, etc.

This means that unless your code is adjusted to effectively exploit the architecture of your HPC system, your code may not run faster at all on the newer system.

It also means HPC clusters proving themselves where custom supercomputers might have previously been the only option, and custom supercomputers delivering value where commodity clusters might have previously been the default.

Tuesday, 12 March 2013

Name that supercomputer (Quiz)

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!

  1. Which supercomputer are we looking underneath?

  2. Acceptance of this leading system became a HPC news topic recently

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

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

  5. Racing cattle powered by Greek letters

  6. Spock was one of these

  7. Which supercomputer does this photo show the inner rows of?

  8. Memory with a deerstalker & pipe

  9. Put an end to Ming (or did he)?

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