An AMD engineer has contributed to the upstream FFmpeg library an ONNX Runtime back-end for its DNN filter. The FFmpeg Deep Neural Network (DNN) filters allow for running AI models natively inside the video processing pipeline for upscaling, object detection, background segmentation, and more. This ONNX Runntime back-end support is notable in that it expands the GPU and NPU capabilities with FFmpeg...
Way back in 2017 for the Linux 4.12 kernel the Realtek rtl8723bs WiFi driver was added to the kernel's staging area. Nearly a decade later, it's still being cleaned-up to suit the more rigorous non-staging area of the kernel in the formal networking subsystem. For Linux 7.2, the staging pull request is once again dominated by clean-ups to this Realtek WiFi driver...
Merged back in Linux 5.15 in 2021 was KSMBD as an in-kernel SMB3 file server. There hasn't been much KSMBD news to report on recently but for Linux 7.2 there is now SMB2 compression support...
Every network operations (NetOps) professional knows the anxiety of a manual configuration update. Even when you meticulously plan the change, a single unrecorded variable or an ad hoc configuration tweak can cascade into unexpected downtime. Across large enterprises, network teams are constantly fighting a losing battle against configuration drift—the gradual, undocumented departure of live network state from the intended design—caused by these sorts of errors.This problem is commonly rooted in human data entry. When network engineers must manually copy and paste details between systems,
Platform engineering, security architecture, and operations teams are being asked to support 2 realities at once: modern application platforms such as Red Hat OpenShift, and long-lived Red Hat Enterprise Linux (RHEL) fleets that still run critical automation. These parallel systems introduce risk, especially around how users, workloads, or automation identities can be authenticated across environments. But there’s also larger questions at play: What boundary exists after privileged work starts? How are exceptions approved? How can operators prove that the intended boundary was active when t
AI models are outpacing human-scale security operations. AI can surface vulnerabilities across major systems faster than teams can act, and most organizations lack the patching capacity to keep up. The solution isn't more intelligence—it's a smarter approach to remediation and risk management.The bottleneck isn’t intelligence, it’s response. IT teams must react to threats while maintaining a hardened posture, and automation makes that possible at machine speed. Before AI models, security teams, site reliability engineers, and IT operations teams were already challenged with the volume of
The many memory management "MM" related improvements were recently merged to Git for the Linux 7.2 kernel. As typical most kernel cycles, some of the low-level improvements can yield nice efficiency wins and better performance in different areas...
Pages