Back to Table of contents

Primeur weekly 2020-01-06

Focus

The LUMI supercomputer is not just a very fast supercomputer, it is first of all a competence development platform - Interview with Kimmo Koski, CSC, Finland ...

Quantum computing

ORNL researchers advance performance benchmark for quantum computers ...

In leap for quantum computing, silicon quantum bits establish a long-distance relationship ...

The Quantum Information Edge launches to accelerate quantum computing R&D ...

Focus on Europe

The coolest LEGO in the universe ...

Middleware

BP looks to ORNL and ADIOS to help rein in data ...

Hardware

New year brings new directory structure for OLCF's high-performance storage system ...

GIGABYTE brings AI, Cloud solutions and smart applications to CES 2020 to enable future today ...

During its final hours of operation, the Titan supercomputer simulated the birth of supernovae ...

Big iron afterlife: How ORNL's Titan supercomputer was recycled ...

Applications

Stanford researchers build a particle accelerator that fits on a chip ...

Brain-like functions emerging in a metallic nanowire network ...

Award-winning engineer helps keep US nuclear deterrent safe from radiation ...

New algorithm could mean more efficient, accurate equipment for Army ...

Paul Ginsparg named winner of the 2020 AIP Karl Compton Medal ...

'Super' simulations offer fresh insight into serotonin receptors ...

Researchers accelerate plasma turbulence simulations on Oak Ridge supercomputers to improve fusion design models ...

Award-winning engineer helps keep US nuclear deterrent safe from radiation


Sandia National Laboratories' Alan Mar was recognized this year by the Society of Asian Scientists and Engineers. Credit: Sandia National Laboratories. Photo by Lonnie Anderson.
23 Dec 2019 Albuquerque - When nuclear radiation hits electronics, it cuts through semiconductors, leaving scars of charged particles that can flip computing bits and corrupt memory circuits, potentially disabling devices or causing erratic errors.

Sandia National Laboratories' Alan Mar was recognized this year by the Society of Asian Scientists and Engineers.

Experts like engineer Alan Mar ensure components made for the U.S. nuclear stockpile pass stringent standards to resist radiation and remain safe and reliable in extremely harsh environments. The 25-year Sandia National Laboratories employee received a Professional Achievement Award from the Society of Asian Scientists and Engineers this year.

The society recognized Alan Mar as someone who "has made significant discoveries, made important advances in his or her chosen career path and is acknowledged as a leader of large initiatives".

Sandia Labs Director Steve Younger said Alan Mar "has approached his remarkable career with a sense of continuous growth. The depth and breadth of knowledge he has gained in diverse research and technical areas makes him both an invaluable contributor to our national security mission and an important mentor to the next generation of scientists and engineers".

Jason Shelton, Alan Mar's manager, added: "I've come to appreciate his dedication to the work and to Sandia's national security mission."

Alan Mar acknowledges a lot of the work Sandia does is invisible to the public.

"When we insert a new technology that makes us more confident that our stockpile systems are more reliable under storage over decades, I take satisfaction in that", he stated. "And that's completely different than developing some new product that generates a billion-dollar business. I think they're both very worthy things to strive for, but this pride is something sort of special to Sandia in fulfilling our core mission."

Alan Mar is now leading a team of researchers to make a push over the next five years to reach a major accomplishment in his field.

Currently, the models Alan Mar uses focus only on specific parts of a system because a comprehensive calculation could take months or longer to churn through, even on high-performance computing platforms. An integrated circuit - or chip - can have millions of transistors, each one requiring its own set of calculations. And that's just one small part of a whole component.

Using a Sandia-developed code called Xyce, sponsored by the National Nuclear Security Administration's Advanced Simulation and Computing programme, Alan Mar's team wants to build a computer model that can predict the radiation effects on a whole weapons component.

Advanced modelling speeds up weapons research, development and qualification. It also lets researchers model changes in experimental conditions that increase the total radiation dose, change how fast a device gets that dose, and mix and match destructive elements like neutrons, energy and heat in environments that cannot be recreated in experimental facilities.

"We can't always replicate an environment that our systems may see", Jason Shelton stated. "It may take years to build a facility, but with this modelling capability that's being established, sometimes you may be able to get an answer sooner by running a simulation."

The path to Alan Mar's whole-model holy grail, he said, is largely marked by two goal posts.

The first goal is eliminating inefficiencies in the current code.

"We work very closely with the code developers and we show them the long poles in the tent when we see them, in terms of calculations that take a lot of time to run", Alan Mar stated. This feedback enables developers to improve the efficiency of their models.

The second goal is to lump together parts of a system into a kind of mathematical abstraction "that doesn't incorporate all the devices underneath, so that block runs a lot faster than if you did it brute-force with all the circuit elements that were originally there", he stated.

The technique has already been shown to shrink computation times from weeks to minutes while maintaining an accurate estimation of a device's response to radiation, he said.

Jason Shelton said Alan Mar works tirelessly to push the forefront of technology, especially in modelling.

"'How can we make the models run faster? How can we make them more accurate?' He's not happy with the status quo", Jason Shelton stated. "He asks: 'What's next?'"

Source: DOE/Sandia National Laboratories

Back to Table of contents

Primeur weekly 2020-01-06

Focus

The LUMI supercomputer is not just a very fast supercomputer, it is first of all a competence development platform - Interview with Kimmo Koski, CSC, Finland ...

Quantum computing

ORNL researchers advance performance benchmark for quantum computers ...

In leap for quantum computing, silicon quantum bits establish a long-distance relationship ...

The Quantum Information Edge launches to accelerate quantum computing R&D ...

Focus on Europe

The coolest LEGO in the universe ...

Middleware

BP looks to ORNL and ADIOS to help rein in data ...

Hardware

New year brings new directory structure for OLCF's high-performance storage system ...

GIGABYTE brings AI, Cloud solutions and smart applications to CES 2020 to enable future today ...

During its final hours of operation, the Titan supercomputer simulated the birth of supernovae ...

Big iron afterlife: How ORNL's Titan supercomputer was recycled ...

Applications

Stanford researchers build a particle accelerator that fits on a chip ...

Brain-like functions emerging in a metallic nanowire network ...

Award-winning engineer helps keep US nuclear deterrent safe from radiation ...

New algorithm could mean more efficient, accurate equipment for Army ...

Paul Ginsparg named winner of the 2020 AIP Karl Compton Medal ...

'Super' simulations offer fresh insight into serotonin receptors ...

Researchers accelerate plasma turbulence simulations on Oak Ridge supercomputers to improve fusion design models ...