search

mlcommons / training

Reference implementations of MLPerf™ training benchmarks
https://mlcommons.org/en/groups/training
Apache License 2.0
1.62k stars 560 forks source link

Hardware Configuration #735

Closed BhAem closed 3 months ago

BhAem commented 6 months ago

Why do hardware configurations only consider the number of CPU cores and the number of Accelerators, without taking into account the size of the server's memory?

ShriyaPalsamudram commented 3 months ago

If this is regarding the benchmark results table, then you can use the System Name + Accelerator Model Name to get the max available memory for that hardware by looking at the hardware specs.

Each submission may not use the entire memory, thus it does not make sense to add a separate column for it in the table.