Grid computing infrastructures were made to support execution of science applications at larger scales. One challenge today in running your science in these behemoth systems the requirement of “griddification” or “supercomputerification”. You need to know how to make the best of your hardware or grid sites in order to orchestrate beautiful workflows and process your science. So a lot of research has been done to create languages such as Swift to make life easier for these domain scientists.
I was debugging a science application for the last several months to run on petascale (100×10^3++ processors) systems. The main goal of the domain scientist was to process hundres of thousands data sequences. I got too much carried away in the debugging to make the application work and have only looked at 3000 of the set In other words, not much *real* work has been done.
Now I should always remember when debugging, remember the scientists who took pain in measuring this data or who can’t get data. (Much like an analogy of “finish your food because there are millions of children hungry in developing countries”).