Tales of working with Nvidia in Linux — A rant
<p>For several years, I’ve had the pleasure (and occasional headache) of working closely with Nvidia. Along this journey, I’ve stumbled upon some quirks that oscillate between amusing and mildly exasperating.</p>
<p>Let me paint a picture for you: I was a part of a startup that developed a cutting-edge machine vision system. This wasn’t just any tech — it was dispatched to factories for performing meticulous visual inspections. Quite revolutionary, if you ask me! But, as with any innovative endeavor, we ran into some peculiar constraints. The most notable? A strict “NO INTERNET CONNECTION” policy.</p>
<p>Now, here’s the twist. These visual inspections demanded powerful GPU support. The solution? Dispatching an individual computer, complete with GPU, to every single inspection station in every factory. As the guardian of deployment, it was my task to ensure that these systems were always up and running. After all, any hiccup or downtime would halt an entire factory line — a scenario we couldn’t afford.</p>
<p><a href="https://medium.com/@idohlevi/tales-of-working-with-nvidia-in-linux-a-rant-5cd1199342b7"><strong>Website</strong></a></p>