https://hgpu.org/?p=15949
Challenges for a GPU-Accelerated Dynamic Programming Approach for Join-Order Optimization