Menu

About

About (1)
Demian Alexandru

Q1-kernel is a hybrid kernel written in lasm (lex-studio version of assembly) and rust. Using lasm for the most low level parts where the performance is a critical hit. Using rust in higher level parts (like syscalls, filesystem some managment things).

The hybrid arhitecture lets us make the kernel very safe and very modular, in the same time being as fast as a monolithic kernel. In Q1-kernel apps get access directly to the kernel without any runtime or bottle neck. By running everything runs in main from lasm everything that is placed in main being loadded into ram. Also you can easily run the entire kernel into ram by just using "entry main" in every file or pointing every file to a file with "entry main" but this can be easily achieved from the config menu. The safe arhitecture make Q1-kernel able to run very stable and easy on anything without having the problems with drivers filesystem or other apps. The modular arhitecture lets you choose what runs in userspace and what runs in kernel space.

Q1-kernel has the best syscalls speed by using an optimized arhitecture. It uses an arhitecture called Clympda syscall. It consist of 2 main things:
-first any app has direct access to syscalls and from that to the hole kernel
-syscalls have direct access to the main component of the kernel being runned in one or two separate cores
The speed it achieves is greater than linux and right now it may be the best in the world having just 0.2-0.3 µs, linux in the same test having 0.2-0.4 µs.

On filesystem side it uses the modern XFS filesystem.

Q1-kernel multicore and multithread is one of the most important parts from it. The managment goes in a cluster like way. The kernel plus the managment runs on one core and 2 threads. If the cpu has more cores the "main core" is going to assign to each core ,trough messaging, multiple jobs. Making a modern aproach to multicore its a very efficient type but still not as advance as DragonFly BSD (btw credit to DragonFly BSD being a huge inspiration for multicore)