Therefore, we observe that memory efficiency depend on both machine balance and computational intensity. T m/t f is known as machine balance and is important to ensure machine efficiency. Minimum possible computation time = t f * f (assuming the requested memory is already in fast memory)Īctual computation time = computational cost + data fetch cost Q factor is also known as the “computational intensity” and it’s vital to algorithm efficiency because per slow memory access we are intend to do more arithmetic operations. Q = f/m (average number of flops per slow memory access) ![]() M = number of memory elements moved between slow and fast memory You can think as these to be the RAM and one level of cache. We will assume two levels of memory for simplicity: fast and slow. For efficient access, it is vital that the algorithms are aware of the underlying storage pattern. ![]() Order is important because of how these values are stored in memory. Likewise, in row major, values are stored sequentially following row order. In the column major order, values are stored sequentially following the column order.
0 Comments
Leave a Reply. |