It was great to meet and shake hands with so many of our readers at the Nvidia GPU Technology Conference (GTC) in San Jose. For this month’s Community Feedback section, we are presenting the results of our very informal surveys conducted both online and onsite at GTC.
The Exascale Report asked:
What is the most promising development or research agenda you are aware of that represents potential positive impact for HPC advancement and progress toward achieving exascale?
Of nearly 100 responses to this question, the top three responses for each question (summarized in our own words) are:
- Power management techniques dealing with dynamic voltage adjustments
- High temperature cooling technologies
- ARM architecture
What can / should we be doing today to prepare critical applications in a way that will enable them to scale for eventual exascale-levels of computation?
- We need more cross-discipline teams looking at modular tools and libraries
- Focus on improving scalability on current petascale hardware
- Rewarding or creating incentives for breakthroughs in scaling of applications
What do you believe will be the single most difficult barrier on the road to achieving practical exascale-class systems?
- Politics behind the funding
- Power management
- Interconnect technology
And, as a result of our most recent poll, we’re pleased to report there has been a slight improvement in the Exascale progress meter.
The Exascale Progress Meter is intended to be a reflection of the collected opinion from recent surveys of readers throughout the global HPC community. Readers were polled in the U.S., U.K., Germany, France, Spain, Russia, China, Japan, India, and Canada. The scale has 20 points with 10 being right in the middle between “Little” and “Good”.
Most respondents feel they are starting to see signs of progress, (this has been covered in this issue, but overall, there is still a general, widespread feeling of concern over lack of appropriate funding (more in the U.S. than other countries), a lack of creative, out-of-the-box thinking in terms of how to get there, and a lack of coordinated, cross-discipline research teams.
For related stories, visit The Exascale Report Archives.
