I’m trying different algorithms to optimize DRAM access patterns to make parallel processing by a GPU of nodes in a graph faster. My biggest takeaway so far is

I’m trying different algorithms to optimize DRAM access patterns to make parallel processing by a GPU of nodes in a graph faster. My biggest takeaway so far is that you first need to check whether or not your dataset fits into the L2 cache…