中科学计算机科学节之系 Uriversity D三 PARTI三N esearchIssues- Parallel computation models(5) Memory LogP Model This model is based on data movement across a memory hierarchy from resource LM to target LM (Local Memory) using point-to-point memory communication inspired by lo to predict and analyze the latency of memory copy, pack and unpack Communication lost consist of the sum of memory communication and network communication times memory communication is from user local menory to network buffer, network communication is from network buffer to network buffer Estimating the cost of point-to-point communication is similar to the original LogP only parameters have different meaning NHPCC at Hefei 2021/2/6 16
NHPCC at Hefei 16 2021/2/6 ▪ Memory LogP Model: ▪ This model is based on data movement across a memory hierarchy from resource LM to target LM (Local Memory) using point-to-point memory communication inspired by LogP to predict and analyze the latency of memory copy, pack and unpack. ▪ Communication lost consist of the sum of memory communication and network communication times. Memory communication is from user local menory to network buffer, network communication is from network buffer to network buffer. ▪ Estimating the cost of point-to-point communication is similar to the original LogP only parameters have different meaning. Research Issues – Parallel computation models(5)
中科学计算机科学节之系 Uriversity D三 PARTI三N esearchIssues- Parallel computation models(5) Model Parameters : effective latency, I=f(d, s), s (data size), d(access pattern)which is the cost of data transfer for application, middleware and hardware o: ideal overhead which is the cost of data transfer for middleware and hardware g: the reciprocal of g corresponds to per-process bandwidth usually o= 9 p: of processors, p =1 (since consider only point-to-point communication) Cost Function(cost per byte (om+)+( Ln/wn)+(om+ D)which is similar to o+I+ o of LogP Om+I-average cost between packing /unpacking, Ln- word size of network communication wn- word size of instruction set NHPCC at Hefei 2021/2/6
NHPCC at Hefei 17 2021/2/6 ▪ Model Parameters: ▪ l: effective latency, l = f (d, s), s (data size), d (access pattern) which is the cost of data transfer for application, middleware and hardware. ▪ o: ideal overhead, which is the cost of data transfer for middleware and hardware. ▪ g: the reciprocal of g corresponds to per-process bandwidth, usually o = g. ▪ p: # of processors, p =1 (since consider only point-to-point communication). ▪ Cost Function (cost per byte): ▪ (om+ l) + (Ln/wn) + (om+ l) which is similar to o + l + o of LogP. ▪ Om + l ── average cost between packing / unpacking, Ln ── word size of network communication, wn ── word size of instruction set. Research Issues – Parallel computation models(5)