Get all your news in one place.
100’s of premium titles.
One app.
Start reading
Tom’s Hardware
Tom’s Hardware
Technology
Matthew Connatser

TikTok parent company used AI to optimize Linux kernel, boosting performance and efficiency

Fedora Linux.

A technical presentation from Chinese tech company ByteDance — which is best known for creating TikTok — detailed how it used AI and machine learning to make the Linux kernel run better on any hardware (via TechSpot). ByteDance believes that in the future computer engineers will likely have to lean on AI for kernel optimization. And with the gains touted in the presentation, those claims might be right.

ByteDance gave its presentation at the Linux Plumbers Conference on Nov. 14 — and though you might think the developer of TikTok is out of place here, you'd be wrong. The presentation, delivered by ByteDance engineer Cong Wang, was heavily detailed both technically and academically (it was made for computer engineers, after all).

The general gist of the presentation: ByteDance used AI to make the Linux kernel (the core of the operating system) much more efficient and performant across all kinds of hardware. We've recently seen AI help make GPU drivers more efficient, but doing the same thing in the kernel of an OS is a significant step up as a technical feat.

That this AI-powered solution worked universally is a big deal, as hardware-specific optimizations are often required to achieve good performance — and that can be challenging for developers because there are so many possible combinations of components.

The presentation detailed how AI optimizations were able to reduce memory usage by 30% — and that was using existing Linux tools, just more efficiently. Network latency was also improved by up to 12% with AI that has prior knowledge (which wouldn't be hard to obtain on a computer used regularly).

ByteDance concluded that AI-assisted kernel optimization could also help balance CPU usage, use cache more effectively, and even detect malware. At the same time, it also acknowledged that machine learning and AI wasn't a silver bullet: real human engineers will apparently not be replaced by computers any time soon for coding kernels.

Sign up to read this article
Read news from 100’s of titles, curated specifically for you.
Already a member? Sign in here
Related Stories
Top stories on inkl right now
One subscription that gives you access to news from hundreds of sites
Already a member? Sign in here
Our Picks
Fourteen days free
Download the app
One app. One membership.
100+ trusted global sources.