For my PhD thesis I wrote a simple tool to measure VM-entry and VM-exit turnaround time - vmlatency. The tool was initially implemented for Linux only. Later ported to macOS. Windows port wan’t finished because I have other activities to deal with prior to the PhD defense. In 2021 I ported the tool on Windows but the port only worked with ICC compiler. The main reason for using ICC was GNU-style inline assembler that was heavily used to execute various processor instructions.

I’ve spent the last ten year working at Intel. For the first four and a half years, I was working at Krylatsky Hills office in Moscow, Russia. The rest of the time at Intel I spent at Drottningholms office in Stockholm, Sweden. It was a long and interesting journey from intern to senior software engineer, from bachelor student at MIPT to PhD.

This is story of a single line of code that took about nine months to write. The story began in early 2020 with a couple of bug reports complaining about Windows host crashes when running the Intel Simics simulator. The reports indicated that the simulations were running on servers using the Microsoft Windows Server 2019 operating system. It appeared that similar hardware running Windows Server 2016 did not suffer problems. However, over time similar reports did start to appear for hosts running older Windows versions as well as Linux.

During Windows port of vmlatency I had to write a bit of assembly using MASM. I needed to execute few instructions not covert by available intrinsics. While doing the port I found a minor oddity in MASM.

In spring 2021, I received an offer to hold a course of lectures about simulation technologies for fourth year students at Moscow Institute of Physics and Technology (MIPT). The course was supposed to focus on fundamentals of computer system simulation and how different kinds of simulators are used in the industry. This was an interesting offer and I accepted it because I always liked talking to students about different research and engineering topics.