This paper caused me to start doubting the Linux scheduler on medium sized boxes ( 36 physical cores in 2016 ) and helped me find an annoying bug in the rhel 7

This paper caused me to start doubting the Linux scheduler on medium sized boxes ( 36 physical cores in 2016 ) and helped me find an annoying bug in the rhel 7 kernel which prevented some processes from being scheduled at all in some cases( it was an issue with Numa load balancing if I remember well). Note my bug was most likely unrelated to the paper, but it marked in my mind a specific class of bugs as possible. So it helped me at least :-)