In their Paper „Fast CSV Loading Using GPUs and RDMA for In-Memory Data Processing“ Alexander Kumaigorodski, Clemens Lutz and Volker Markl devise a new CSV parsing approach that streamlines the control flow while correctly handling context-sensitive CSV features. By offloading I/O and parsing to the GPU, their approach enables databases to load CSVs at high throughput from main memory with NVLink 2.0, as well as directly from the network with RDMA.
A preprint version of the paper is available here.