Computer hardware requests for CPU and GPU resources.
Many of the new advances in AI and ML require substantial hardware
support. Therefore, given the lack of direct measures of disruption
for research in the area, we examine requests for data science related infrastructure as a proxy. Specifically, we collect data about the change in requests by application year through Compute Canada’s annual resource allocation competition from 2012 to 2020. We separately identify the trends in requests for CPU and GPU resources for research support as shown in this figure.
The figure above shows a remarkable increase in the number of computer
hardware requests for CPU and GPU resources to support research over the 2012-2019 period. However, the series also indicates a negative COVID-19 impact on academic research related to AI and data science given the significant decrease in GPU allocation requests and a slowdown in the asks for CPU support in the 2020 application year (which took place in the Fall of 2020). These decreases would point to a lower level of planned R&D activity in the grant period which runs from April 2021 through March 2022, potentially further dampening the pace of Canadian R&D production at least in the short run.
Data Source: Compute Canada. Support year for the resource allocation Competition is a 1-year period during which successful applicants receive computational resources. It begins the year following the application year. For instance, winners who applied in the 2020 application year (September-November) receive support from April 2021 through March 2022
