09:23:51 From Eric Freudenthal : the tools and infrastructure that makes this meeting possible is a remarkable pdc success 09:24:11 From Eric Freudenthal : (imagine this pandemic without it) 09:56:59 From Jim Teresco, Siena College : Thanks all for doing this. I am trying to register for ipdps but the cvent system says “capacity full”. 09:59:40 From Sushil Prasad : For IPDPS registration glitch, please email esaule@uncc.edu 12:17:44 From Eric Freudenthal : I wonder about plans make pdcunplugged sustainable (in that it might live beyond your career) 12:19:26 From SGhafoor : question to 1st presenter: Do you have class room example available to share with the community? 12:31:42 From Suzanne : I will be around during lunch if anyone has any more questions 12:31:54 From Jim Teresco, Siena College : thanks to all presenters, very interesting work 12:35:00 From Erik Saule : Some people have reported that they could not register for IPDPS becasue the conference was full. The issue is now resolved. 12:44:00 From Trilce Estrada : Can we test the slides for Session 2 at 12:20 Central Time? just quick to make sure everything is okay, and because we won't have a break after the curriculum and CS materials ? 13:19:46 From Mark Lewis : Thanks! 13:36:30 From Charles Weems : https://tcpp.cs.gsu.edu/curriculum/?q=NSF_Cybertraining 13:49:20 From Charles Weems : https://tcpp.cs.gsu.edu/curriculum/?q=NSF_Cybertraining 14:16:54 From Sushil Prasad : CFP for JPDC Special Issue: https://nam03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.journals.elsevier.com%2Fjournal-of-parallel-and-distributed-computing%2Fcall-for-papers%2Fteaching-parallel-distributed-and-high-performance-computing&data=02%7C01%7Csushil.prasad%40utsa.edu%7Cebce05d5db3046afa1e908d7fac3b114%7C3a228dfbc64744cb88357b20617fc906%7C0%7C0%7C637253590713557522&sdata=WexD3jOjX4lt%2BceaRvdWiyosgxyIyc8TW1xlN8putuE%3D&reserved=0 14:17:41 From Sushil Prasad : JPDC Special Issue: https://www.journals.elsevier.com/journal-of-parallel-and-distributed-computing/call-for-papers/teaching-parallel-distributed-and-high-performance-computing 14:20:19 From Ana Gonzalez : Thank you for the links 14:51:31 From Erik Saule : In case, you have not registered to IPDPS, you can register for free at ipdps.org 14:51:45 From Erik Saule : There was a quirk this morning that was fixed 15:07:04 From Henry Gabb : How about hosting the repository on GitHub? 15:08:10 From Henry Gabb : The taxonomy of the GitHub repo could mirror the organization of the curriculum. 16:14:22 From Benson : Patrick - would you consider adapting materials to RISC V? 16:19:07 From Massimo Canonico : ttps://sites.google.com/uniupo.it/massimocanonico/teaching-cloud-computing?authuser=0 16:20:13 From Rocío Carratalá-Sáez : https://sites.google.com/uniupo.it/massimocanonico/teaching-cloud-computing?authuser=0 (missing h) 16:27:28 From Suzanne : FWIW, we have multiple instructors teaching scala over many courses in our curriculum at West Point (Data Structures, ProgLang and Algo) 16:28:38 From Massimo Canonico : Thanks, nice session 16:31:39 From Sushil Prasad : Thanks Suzzane! 16:41:32 From Jim Teresco : I have to run, might be able to return for at least part of the panel. Thank you to the organizers and speakers for an informative and enjoyable day. It’s not the same as a real in-person workshop by any stretch, but the virtual format allowed me to attend as a non-author, which I would not have been able to do even under normal circumstances given time and money constraints. I look forward to seeing all of you in person at future meetings. 16:42:35 From Massimo Canonico : Hi Jim! Have a good day 17:05:42 From Mark Lewis : Central Texas has a really strong showing at this meeting. 17:13:00 From Almadena Chtchelkanova : I would like to elaborate on the future goal for High Performance, Parallel and Distribute computing: To Enable Real Time Robust Reliable Decision Making Process Based on Scientific Modeling, Simulation, and Data Science while minimizing time to solution (productivity) and energy to solution. There is a need for proper curriculum/training to insure that Everybody will make a transition from serial to heterogeneous ( including reconfigurable architectures, GPUs and multicores) computing systems AND from serial to parallel programming to deal with computation-&data-intensive applications, Exploiting parallelism and concurrency at multiple levels and/or scales and execution/programming models; And taking into consideration Interactions between architecture, language, compiler, systems software, I/O & storage and application layers While minimizing time and energy to correct repeatable SOLUTION 17:26:07 From Uzi Vishkin : I just looked up a 2008 presentation by Intel's lead parallel education reach out person. He cited Bill Dally, Ed Lee, Dave Patterson and Burton Smith all warning about teaching & deploying broad multi-threaded programming aiming at multi/many cores and expect effective usability. 15 years since 2005 (the transition away from unicores) and a decade into EDU PAR, where do we stand: Was the ~2005 paranoia on target or a false alarm? Do we have sufficient evidence on what works? (as opposed to what you or me think is right.) Also: on one hand, we heard today concerns that co-processors such as Xeon Phi are disappearing. On the other hand, the keynote called to teach programming heterogeneous systems, which is significantly more challenging. 17:32:49 From Uzi Vishkin : Personally, I identify most with Henry advocating pressure on vendors to make programming as easy and abstract as possible. I wanted to ask him: how to make them listen and ...act? 17:36:54 From Susan and Arny : what are th educational implications of your vision? 17:42:14 From Suzanne : i also have a question for the panel (about education) 17:43:33 From Uzi Vishkin : I was focusing on the manycore space. Not the the large Google/BF computers 17:44:59 From Margaret Martonosi : I have advised a set of undergraduate research projects in the area of memory consistency models and unpredictable aspects of concurrency. One fascinating educational experience is seeing the pedagogic value of having students explore these unpredictable scenarios. It motivates them to learn about systems deeply in order to understand how these surprising concurrency variations occur. That’s a fun pedagogical lesson learned for me! 17:48:51 From Eric Freudenthal : I think there's no conflict between needing more people to be able to work with /tune low level parallelism and needing all to understand the big ideas of parallelism. I think that we'll have more success if we don't conflate those goals. 17:48:51 From Andrew Lumsdaine : The most exciting phrase to hear in science, the one that heralds new discoveries, is not “Eureka” but “That’s funny...” —Isaac Asimov (1920–1992) 17:49:23 From Uzi Vishkin : declarative programming remains relevant. Parallel declarative programming is as essecntial 17:54:25 From Susan and Arny : numbers are different from numerals 17:56:58 From Margaret Martonosi : Common APIs are great. We need much better specifications of the functionality. For example, if I have an FFT accelerator today that I invoke through an API, and then map to a different FFT accelerator in a different implementation, how do we specify what sort of accuracy is “good enough” for these different FFT accelerators to be interchangeable? 18:05:07 From Charles Weems : A responsive user interface is a performance problem encountered in early courses. A multi-user application (e.g., game) encounters race conditions. 18:06:20 From Margaret Martonosi : When I ask my comp arch classes who has experienced programs that are “too slow”: It’s always the gamers who raise their hands. That’s their first/only experience of programs not running fast enough. 18:07:04 From Joel Adams : Those of us who teach the introductory data structures course have an ideal opportunity to motivate parallelism: just load up an elementary data structure with enough values (e.g., an organism’s genome) that it takes more than 10 secs to process. 10 secs is an eternity for today’s students. 18:26:16 From Eric Freudenthal : the folk who study career differentiation have evidence that people's proclivities vary widely. Some people find helping others highly satisfying, others do not. 18:29:54 From Margaret Martonosi : Thank you!!! 18:30:30 From Satish : Congratulations Suzzane! 18:30:51 From Suzanne : Thanks :)