Lecture 12

Question 1

Keep active threads together - better!

Answer 1

Threads and memory both behave like this for coalescing

Question 2

__manage__ memory

Answer 2

accessible from both CPU and GPU

Question 3

atomic operations

Answer 3

atomicAdd, atomicSub, AromicExch, atomicMin,
atomicMax, atomicInc, atomicDec, atomicCAS,
atomicAND, atomicOR, atomicXor

Question 4

why don’t always use atomic operations?

Answer 4

leads to sequential performances