I need to load 90 files into the same graph. What’s the recommended way to load them in parallel (for e.g. load 8 files at the same tile)?
The section on Concurrent Loading outlines the capability to run a loading job with concurrent processes, one per file.
Furthermore, if you have a distributed system, you can speed up loading by distributing your input files among the cluster. E.g., in the example in that section, m1 and m2 are the aliases for two machines in the cluster (aliases are established during initial system configuration.)
Concurrent Loading in the documentation.