[Benchmarking] Allow number of vectors to not be divisible by number of clients #426
Labels
good first issue
Good for newcomers
Infrastructure
Changes to infrastructure, testing, CI/CD, pipelines, etc.
Related to #409 (comment).
Currently, OSB benchmark workload can only use multiple clients when the data set can be partitioned evenly. For example, for a data set with 100 vectors, 10 clients could be used but not 9 (ref).
Instead of requiring the number of vectors to be divisible by the number of clients, we could have one client execute fewer requests.
The text was updated successfully, but these errors were encountered: