I am two types of kernel user; one is Android whose kernel is Linux and second is Windows whose kernel is Windows NT. I can feel the issues both of these have; both fight for stability when something cause system to get crashed and I feels BSOD and this is so annoying when you are working on something harder which needs system to be stable. This same came with me while training an AI model on my own system and Windows get crashed. So, I thought why not to try to build a kernel when never get this issue, but after some time while building I learnt nothing can be 100% stable but then I thought if this is not possible then why not to create a layer of problems which cause system to crash at system level. I had built this ARKM: Advanced Robotics Kernelling Machine, which is a custom built 12 core 64-bit Microkernel architecture with AMP which means Asymmetric Multiprocessing, ACPI which means Advanced Configuration and Power Interface into the core architecture of ARKM and due to this the stability of kernel became sky high. I had used the AMP architecture as follows:
Core 0 (BSP) → Kernel shell, interrupts, global scheduler.
Core 1 → Idle stability loop + diagnostics.
Core 2 → GPU compositor for zero‑latency graphics.
Core 3 → Isolated Ring 3 user‑space applications.
Its Benefits came as: Deterministic performance, no contention, guaranteed responsiveness (e.g., compositor never lags even if apps crash).
Now if I talk about how ACPI changed the architecture then it also follows the steps:
Hardware Discovery → ARKM parses ACPI tables (RSDP, MADT) to locate the Local APIC and enumerate CPU cores.
Power Management → Provides standardized methods for sleep states, CPU throttling, and device power control.
Interrupt Routing → ACPI helps ARKM configure APICs and IOAPICs for multicore interrupt handling.
Scalability → Makes ARKM capable of dynamically waking up Application Processors (APs) and bringing them into 64‑bit long mode.
Now if I talk about the 12 cores architecture it has then there is also a specialty behind it which is also as follows:
Core 0 (BSP) → Kernel shell, interrupts, timers, global scheduler.
Core 1 → Idle loop + diagnostics (keeps stability checks running).
Core 2 → Dedicated compositor (graphics/UI rendering at zero latency).
Core 3 → Ring 3 sandbox apps (user programs run isolated here).
Remaining Cores (4–11) → Can be assigned to parallel workloads, background services, or specialized tasks (e.g., networking stack, storage drivers, AI routines).
I am still building it. My Aim is if I had bought a machine, then it must work according to me. I am also creating it under my future Company AROM so that I don't want to rely on any other kernel.
ARKM is just 2 to 10 KM behind the finish line. ARKM will now come with Full-fledge Kernel and with a Server Kernel with POSIX tech Linux have just for attract the user and developers of Linux on ARKM and we are developing our own also this Linux POSIX architecture will act as bridge between present world and future world.
Support me IF you want to dive into the future tech or want to explore it.
Comments URL: [https://news.ycombinator.com/item?id=49320900](https://news.ycombinator.com/item?id=49320900)
Points: 1