Details
![Zahra Ebrahimi Headshot](https://confcats-catavault.s3.amazonaws.com/CATAVault/ieeecass/master/files/styles/cc_user_photo/s3/user-pictures/15211.png?h=6b9bf2e9&itok=SGY-p464)
- Affiliation
-
AffiliationTechnische Universität Dresden
- Country
CGRAs with high parallelization/customization capabilities are viable platforms for ever-evolving health-monitoring applications. We propose BioCare, an area/energy-efficient CGRA for health-monitoring edge devices, which exploits approximations across HW/SW stack while still provides acceptable final QoR. BioCare offers different energy-accuracy trade-off levels through the plasticity of its PEs, each can support functional-versatility and precision-adaptability with a SIMD manner. BioCare surpasses SoAs, by achieving up to 32%/67% area/energy savings and 3.6x higher throughput, when analysed on multiple widely-used ECG/EEG kernels/application. Our implementations will be open-sourced to springboard future research for bio-signal processing/reconfigurable/approximate computing communities.