I had a similar thought as well. The questions are then, first, how do we think/work in more cache-friendly manner? Do things linearly, plan things ahead, don't branch out on random noise? And, how to minimise allocations? Don't try to ingest more information than what you can fit on your "stack"?
// and don't forget about "stack unwinding" before going to bed
// and don't forget about "stack unwinding" before going to bed