I would go with Col4, as the data on both the AMPs is distributed equally, 50 & 51.
In your case you can choose either one as the data count is small thus the PI selection won't make too much difference. The PI selection becomes more critical when the data count is huge lets say hundreds of million rows landing on the AMPs.
I would go with Col4, as the data on both the AMPs is distributed equally, 50 & 51.
In your case you can choose either one as the data count is small thus the PI selection won't make too much difference. The PI selection becomes more critical when the data count is huge lets say hundreds of million rows landing on the AMPs.