We are doing unbuffered I/O while reading the input file, which can lead leads to very slow computation on some environments (in our case PVCs in a kube clusters). I didn't dig into specific details but I saw that osmpbfreader doesn't wrap the reader with a buffer and makes atomic reads which would lead to an I/O operation each (often just a u32).
We are doing unbuffered I/O while reading the input file, which can lead leads to very slow computation on some environments (in our case PVCs in a kube clusters). I didn't dig into specific details but I saw that osmpbfreader doesn't wrap the reader with a buffer and makes atomic reads which would lead to an I/O operation each (often just a u32).