Recently read a price to performance comparison for GPU instances on AWS, Google Cloud, Azure and Linode (https://www.linode.com/content/gpu-price-performance-benchmarking/).
Have you seen anything similar for cloud VMs?
How would you set up such a benchmark?