Any success stories or tips for running GPU-backed raster analysis workloads (deep learning, image classification) on ArcGIS Enterprise on Kubernetes? Looking to optimize tile caching and GPU scheduling across nodes.
