QFabric: Guarded Cross-Domain Task Recovery and Its Limits on a Linux-RTOS Board
AuthorsAyushman Bhattacharya, Anwesha Chakraborty
AffiliationsPollinations AI ayushman.elixpo@gmail.com · Dept. CSE, JIS University anwesha.elixpo@gmail.com
QFabric tests whether tasks can escape timing trouble by moving between Linux and an RTOS, and finds that careful guards mostly prevent moves that would make performance worse.
Key results
Approximate milliseconds per MCU invocation through Bridge RPC.
Times slower on the MCU than on the Cortex-A53.
Cells with switches when hysteresis was removed.
Times higher under the ablated controller.
Harmful moves rolled back after probation.
What the paper found
QFabric tests whether a Linux task that misses an empirical soft real-time contract can recover by moving to an RTOS microcontroller. On the Arduino UNO Q, with a Cortex-A53 running Linux and a Cortex-M33 running Zephyr, the prototype uses generated task interfaces, semantic and capacity checks, hysteresis, quiescent epoch switching, and probation before committing a move. The key result is a limit: each MCU call through the stock Bridge RPC path costs about 19 ms in dispatch, and the MCU runs the heaviest FIR workload 34 times slower than the Linux processor. Because calls are still dispatched through Linux, Linux interference also degrades the MCU path. In a four-node-chain replay, Linux met the contracts in all 30 cells, so QFabric correctly made zero migrations. By contrast, removing hysteresis caused switches in 7 cells and raised p95 latency by as much as 177 times; probation rolled back 5 of those moves. The study therefore presents guarded recovery not as a demonstrated speedup, but as a way to prevent harmful migrations on this board. It also corrects an earlier reported 36.6% improvement, tracing it to measurement defects rather than a real benefit.
Original abstract
Heterogeneous embedded boards pair a Linux application processor with a real-time microcontroller (MCU), which suggests moving a task between the two domains when its timing degrades. We present QFabric, a prototype that treats cross-domain placement as guarded recovery from a violated empirical soft real-time contract: a bounded task is compiled for both domains, monitored in windows with hysteresis, switched only after semantic and capacity admission at a quiescent epoch boundary, and committed or rolled back after probation. On an Arduino UNO Q (Cortex-A53 Linux and a Cortex-M33 running Zephyr), we characterize the stock Bridge RPC interconnect and measure five workloads in both domains under idle, CPU-load, and higher-priority SCHED_FIFO contention, with ordinary and real-time callers. Each MCU invocation costs about 19 ms of dispatch regardless of compute, the MCU runs the heaviest kernel 34x slower, and because every call traverses Linux userspace, Linux contention inflates the MCU path far more than the Linux path (p99 up to 400 ms versus 17.0 ms). Across 30 cells of a four-node chain, Linux meets every contract within 0.8% misses and QFabric correctly never migrates; a controller without hysteresis migrates in 7 cells, raising p95 by up to 177x, and probation rolls back 5 of those moves. On Linux-dispatched heterogeneous boards, migration therefore cannot isolate a task from Linux interference, and recovery guards matter chiefly by preventing harmful moves. This paper supersedes an earlier artifact release whose reported improvement arose from measurement defects that we identify. All source, raw traces, and generators are public.
Read the original paperMore in AI Hardware
Browse all 40 papers →Halley: A Predictive Control Plane for Optical AI Data Center Networks
Manav Mishra, Krishna M. Sivalingam
Halley anticipates failures in optical data-center networks and coordinates lasers, switches, and thermal controls to keep more AI training capacity online.
PCBRouteAgent: How Close Are LLM Agents to Becoming PCB Routing Engineers?
Yue Zhang, Jilei Sun, Liqiang Jing, Yijie Wei, Ruohan Yang
PCBRouteAgent pairs an LLM’s board-level routing plans with KiCad’s routing engine to tackle complex PCB layouts.
A Hybrid GPU and Processing-Near-Memory Architecture with Shared DRAM for Efficient Prefill-Decode Disaggregation
Sanghyeok Han, Inseong Choi, Byungkuk Yoon, Gyeonghwan Park, Jae-Joon Kim
DuoPD pairs GPUs with memory-side processors to run LLM prefill and decoding together and improve inference throughput.