Right In-Place (RiP) Convolution: A Simple, General, and Near-Optimal Strategy for Memory-Efficient CNN Inference

Excerpt

arXiv:2610.00586v1 Announce Type: new Abstract: Activation memory, not compute, limits CNN inference on constrained hardware such as microcontrollers. Direct in-place convolution removes the dual-buffer cost, but the memory-optimal formulation of Gural and Murmann assumes valid padding, unit stride, unit dilation, and odd square kernels, and needs a non-sequential traversal costing $2\times$ inference time in transposes. We identify two regimes in which their published closed form does not hold: