ProWAFT: A ROMA-LPD Instance for Workload-Aware and Dynamic Fault Tolerance in FPGA-Based CNN Accelerators
arXiv:2607.01602v1 Announce Type: new Abstract: SRAM-based FPGAs provide an attractive platform for energy- and latency-constrained CNN inference at the network edge, yet transient faults can lead to silent errors that compromise reliability. Always-on redundancy (e.g., full TMR) improves correctness but incurs substantial performance and energy overhead, while reactive recovery may introduce unacceptable latency on the critical path. We propose \textbf{ProWAFT}, a proactive workload-aware fault-tolerance framework for FPGA-based CNN accelerators that uses partial reconfiguration to selectivel
As AI deployment at the edge intensifies, the need for robust and efficient hardware solutions in unreliable environments becomes critical.
This research addresses a key reliability challenge for energy- and latency-constrained AI inference, crucial for widespread adoption in sectors like autonomous systems and IoT.
The proposed ProWAFT framework offers a more efficient alternative to traditional fault-tolerance methods, potentially reducing the performance and energy overheads of reliable AI hardware.
- · FPGA manufacturers
- · Edge AI providers
- · Critical infrastructure relying on AI
- · Semiconductor industry
- · Companies reliant solely on software-based reliability
- · Less energy-efficient hardware solutions
Increased reliability and efficiency of FPGA-based CNN accelerators at the network edge.
Faster and more widespread adoption of AI in previously challenging environments due to improved fault tolerance.
Reduced total cost of ownership for AI-powered edge devices, accelerating economic impact in various industries.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL