Urgent.News

What's breaking now, across thousands of outlets.

Tech

Multikernel Linux Tree Released: Runs Multiple Kernels On Bare Metal Without a Hypervisor

"Multikernel Technologies has put out mklinux v7.0-mk2, the first public release of the multikernel Linux tree it's been building since last year," reports the blog It's FOSS: The basic premise of the project is that one physical server can run several independent Linux kernels side by side, each with its own dedicated CPUs and memory. So, the same box could run a database on one kernel and a GPU…

Multikernel Technologies has unveiled mklinux v7.0-mk2, the initial public release of their innovative multikernel Linux tree. This advanced technology enables a single physical server to operate multiple independent Linux kernels simultaneously, each with its own set of dedicated CPUs and memory. Consequently, a single server could run a database on one kernel while simultaneously executing a GPU training job on another kernel, without any interference between the two.

The host kernel, responsible for managing the machine's resources, spins up new kernel instances using Linux's kexec mechanism, granting each instance direct access to hardware resources without the need for a hypervisor. The host kernel manages this system through device tree overlays located in /sys/fs/multikernel/. Moreover, it possesses the capability to redistribute hardware resources among running instances while maintaining system stability.

This setup differs significantly from virtual machines (VMs) and containers. While VMs maintain a hypervisor and host kernel between the application and hardware, containers operate without a hypervisor but share a single kernel for all tenants. Multikernel Technologies has only validated support for x86_64 hardware for this release, although they assert that the architecture-specific code is sufficiently modular to facilitate future expansions to other architectures.

Currently, the company is offering several products built on this kernel tree, including one tailored for running AI agents in their own kernel with direct GPU access, and another aimed at patching running kernels without disrupting system operation.

Written by urgent.news from Slashdot's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at linux.slashdot.org →

More in Tech

SaaS: Reactivation with Product Signals

Muchos equipos SaaS mandan emails de reactivación cuando un usuario lleva varios días sin entrar y esperan que eso alcance. A veces funciona un poco. Muchas veces no.

  • Reactivation emails often fail due to lack of context
  • Effective reactivation uses one clear product signal
  • Measure completion of specific step within 72-hour window

Fillable Tax Forms with Asynchronous Jobs, Retries, and Latency Control

A Node.js service can fill a tax form inline, but accepting the request is not the expensive part. Preserving page geometry while filling fields, validating the result, redacting personal data, and…

  • Two latency budgets: API budget and completion budget
  • Job ID returned in 202 Accepted response
  • Temporary files managed in per-job directory

Understanding JavaScript Proxy and Reflect APIs

JavaScript is a highly dynamic language, and with the introduction of ES6, it gained powerful features to observe and customize the behavior of objects — Proxy and Reflect.

  • Proxy API creates a wrapper around a target object, intercepting fundamental operations.
  • Reflect API provides methods to perform default behavior of these operations programmatically.
  • Together, Proxy and Reflect enable transparent and predictable modifications in JavaScript.

More from Monday 31 August →