qualify rank() over (partition) - question - response (2) by gksenthilkumar

Sorry for confusion..
I have data like below..
here I want to pick one row for each inr & rd for latest timestamp(dt col)..
if we have multiple rows for latest timestamp for that inr & rd then I want to get the latest row id rows for that latest(max) dt for each inr & rd...
like below highlighted rows are my output for this scenario
table contains 5+ billion rows, 100+ columns, for specific process it will fetch only 1MM rows
any suggestion in performance stand point..

inr

rd

dt

c1

c2

…

c100

1

1

2000-01-01 10:10:10 1111

1

1

…

1

1

2000-01-01 10:10:10 2222

2

2

…

1

1

2000-01-01 10:10:10 3333

3

3

…

1

1

2000-01-01 10:10:10 4444

4

4

…

1

2

2000-01-01 10:10:10 1111

1

1

…

1

2

2000-01-01 10:10:10 4444

4

4

…

1

2

2000-01-01 10:10:10 4444

5

5

…

1

2

2000-01-01 10:10:10 4444

5

5

…

1

2

2000-01-01 10:10:10 2222

2

2

…

1

2

2000-01-01 10:10:10 3333

3

3

…

1

3

2000-01-01 10:10:10 1111

1

1

…

1

3

2000-01-01 10:10:10 3333

3

3

…

1

3

2000-01-01 10:10:10 3333

4

4

…

1

3

2000-01-01 10:10:10 3333

5

5

…

1

3

2000-01-01 10:10:10 2222

2

2

…