The Next Horseman, HPC Training and Value
Apr. 24th, 2020 07:51 amWorld COVID-19 confirmed cases are now at 2.7 million, with 190,000 deaths. The United States is now over 32% of the total number of confirmed cases, and almost 50,000 dead. The Western European countries, still among the worst hit, are managing to slow their number of cases and fatalies down, although the numbers are still far too high. Following pestilence, famine is the next horseman of the apocalypse. As COVID-19 begins to creep into the developing world, there are now serious challenges to the world's food supply, and the possibility of famine is growing, the extent that the leadership of the World Food Programmer are describing a potential disaster of Biblical proportions. It is in the context that I am making a fairly significant personal decision; as friends would know, I have a more-than-significant collection of tabletop RPGs; around $50,000 worth last time I bothered to do a rough calculation. Whilst I have casually traded items in the past, I am going to start doing so in earnest with the proceeds donated to Médecins Sans Frontières, one of my favourite charities. It is just another step in my future life decisions to make every effort to make life more tolerable for those in the worst situations.
The past few days I've been conducting training workshops using the University's video conferencing; the first class, in particular, had a lot more researchers attending that what had registered (close to forty). I had concerns over the capacity of my home Internet's bandwidth to cope with such a large number, but with video and audio feeds turned off (questions were raised in chat), it all coped reasonably well. Two of the courses were pretty stock standard, Introduction to High Performance Computing Using Linux and Advanced Linux and Shell Scripting for HPC, the third is less regular, GPU Applications and Programming. I've been swotting up a bit on the latter because, to be honest, I haven't done any GPU programming myself since the last time I conducted the course and there have been some changes in the system. It is, in some ways, quite flattering to have so many research on the waitlist for my courses (it's around 700), it is nice to be wanted an all that, but I don't scale. Even if I manage 30 or so at a time, once a month, the difficulty in getting through the set is obvious. Maybe I should ask the powers in charge if I can increase to multiple class sets per month. I'll be exhausted, but it will be worth it. I help maintain systems and teach researchers how to use them so that they can provide the results that help make the world a better place. This no mere fancy; massive macroeconomic modelling indicates that the social return on investment in High Performance Computing is $44 return (savings or profits) per dollar invested. I wouldn't work in this industry if it wasn't the case.
The past few days I've been conducting training workshops using the University's video conferencing; the first class, in particular, had a lot more researchers attending that what had registered (close to forty). I had concerns over the capacity of my home Internet's bandwidth to cope with such a large number, but with video and audio feeds turned off (questions were raised in chat), it all coped reasonably well. Two of the courses were pretty stock standard, Introduction to High Performance Computing Using Linux and Advanced Linux and Shell Scripting for HPC, the third is less regular, GPU Applications and Programming. I've been swotting up a bit on the latter because, to be honest, I haven't done any GPU programming myself since the last time I conducted the course and there have been some changes in the system. It is, in some ways, quite flattering to have so many research on the waitlist for my courses (it's around 700), it is nice to be wanted an all that, but I don't scale. Even if I manage 30 or so at a time, once a month, the difficulty in getting through the set is obvious. Maybe I should ask the powers in charge if I can increase to multiple class sets per month. I'll be exhausted, but it will be worth it. I help maintain systems and teach researchers how to use them so that they can provide the results that help make the world a better place. This no mere fancy; massive macroeconomic modelling indicates that the social return on investment in High Performance Computing is $44 return (savings or profits) per dollar invested. I wouldn't work in this industry if it wasn't the case.