After Andres's excellent talk at PGConf we tried benchmarking effective_io_concurrency on some of our servers and found that those which have a number of NVME storage volumes could not fill the I/O queue even at the maximum setting (1000).
Before we start benchmarking patched versions of PostgreSQL to bump this setting up to ever higher values, I am wondering if anyone has done this yet and if so if you would be willing to share results.