BusDK

BusDK Update

riscv64 becomes a native Bus Engine OS target: a fast boot with memory hotplug

Bus Engine OS now has an experimental native riscv64 image target. The recorded proof run booted in about 70 seconds, with the riscv64 image chain, cross-build fixes, and browser-hosted harness work landing around the same target.

The same work also brought kernel-side memory hotplug requirements into the riscv64 configuration and checker. That opens the live grow/shrink path for guest memory without reboot, which is important for right-sizing AI VMs as workloads change.

July 5, 2026Bus Engine OSriscv64memory hotplug

In Brief

TL;DR

  • Bus Engine OS gained an early native riscv64 image target, with a proof boot recorded at about 70 seconds.
  • The riscv64 kernel configuration now carries memory-hotplug requirements and checker enforcement for live guest-memory right-sizing work.
  • RISC-V is under evaluation as the main QEMU/WASM acceleration path because it appears to fit in-browser execution better than the harder x86 accelerator path and has produced the strongest results so far.

The riscv64 target is a native Bus Engine OS image path, not just a cross-compile curiosity. Reaching it meant a riscv64 builder toolchain, package-level cross-build fixes, root-disk boot plumbing, and an image-chain proof that reached a roughly 70-second boot. That is enough to make riscv64 useful for focused system work while the target remains early.

Memory hotplug is the practical VM feature in this slice. The riscv64 kernel profile now requires sparse-memory and memory-hotplug settings, and the checker enforces that requirement for the architecture. The intended operating shape is simple: grow or shrink a guest's memory while it keeps running, then use that path to right-size AI VMs around changing model, batch, and interactive workloads.

The product reason to care about RISC-V here is the browser path. Bus Engine OS is evaluating QEMU/WASM acceleration, and RISC-V has given the strongest browser-hosted results so far. An x86 accelerator is also in development, but it carries more complexity; the current estimate is that RISC-V maps better to WASM and has a clearer route toward near-native in-browser performance. That is an evaluation signal, not a promise of final parity.

As the riscv64 target and hotplug workflow settle for broader hands-on use, docs.busdk.com is where the operator recipe will live.