Previously, we were spinning up separate instances for benchmarking
optimized and non-optimized code. This is inefficient and incurs
unnecessary cost for very large instances (e.g. metal).
This commit changes the benchmarking CI to benchmark optimized
and non-optimized code on the same EC2 instance.
Previously, we were spinning up separate instances for benchmarking optimized and non-optimized code. This is inefficient and incurs unnecessary cost for very large instances (e.g. metal).
This commit changes the benchmarking CI to benchmark optimized and non-optimized code on the same EC2 instance.