Closed tim-pugh closed 5 years ago
We've discussed a couple of possible approaches.
Today, just as a fun first step to an MVP, we've simply shuffled the order of the fields.
Moving forward, we could:
Approach 1:
Shuffle the order of the fields
Iterate through the new order, adding fields into buckets such that each bucket is the size of a cache line (except for any left-out remainders)
We can then shuffle the order of the buckets (or leave them alone?)
We'll write down any other approaches here.
First iteration exists and is closed by #17
We will require multiple approaches for randomization. To meet one of our project goals, we'll need a performance mode that will require a different algorithm for re-arranging the fields with respect to cache lines.