ZeroSwap received the RTAS 2026 Best Paper Award

Our paper, “ZeroSwap: Minimizing Swap Overhead for Real-Time Multi-DNN Inference via SSD-based GPU Memory Extension”, received a Best Paper Award at the 32nd IEEE Real-Time and Embedded Technology and Applications Symposium.

ZeroSwap combines selective swapping, shared pinned memory, and computation-transfer overlapping to make SSD-backed memory extension practical for timing-sensitive DNN inference on integrated GPU platforms.

Source: RTCL@DGIST announcement