Leveraging Dynatrace for supporting novel use-cases via a custom-made NVML-based Python Extension

Tomasz Gajger

Tomasz Gajger

December 22, 2020

For a long time now, GPUs have been more of a versatile number crunching devices, than just tools for graphics rendering.


The GPGPU paradigm becomes prevalent throughout the industry and a steadily growing number of commercial applications depends on them for workloads processing. This in turn means, that - a formerly homogeneous - CPU-based hardware stack is joined by new entities, which also must have their performance continuously monitored.


While Dynatrace does not provide built-in GPU support, it offers a flexible way of extending its capabilities via Extensions. In the recently published article, I explored possibility of leveraging Dynatrace for supporting novel use-cases via a custom-made NVML-based Python Extension. All that wouldn't be possible if not for the welcoming attitude the company exhibits towards such initiatives. When I approached Gdańsk lab leadership with the idea for my PhD research, I not only received encouragement, but also an approval for free access to a real Dynatrace environment, no strings attached.


Get to know more:

About Author:
Tomasz is a Principal Software Engineer at Dynatrace.
He joined the company in 2015 as an intern in one of Gdańsk OneAgent teams and quickly rose through the ranks. Currently he works as the Product Owner for OneAgent full-stack installation, a team he built from scratch. Besides that, Tomasz is a PhD student at ETI faculty, GUT, where he conducts research on GPU performance monitoring and analysis.

Related jobs

Director of Software Engineering


27k+ PLN GROSS /mo.

Frontend Engineer


9k - 23k PLN GROSS /mo.

Frontend Developer in a developer relations team


10k - 23k PLN GROSS /mo.