User 429 | 7/11/2014, 7:13:39 AM
Assume that one shard includes v vertices and e edges.(numofedges = n, numofedges = e)
Naively, the memory usage for loading a shard should be: Memory occupied = sizeof(vertexDatatype) v + sizeof(edgeDataType) e
Does this ideal memory usage hold to be true in the GraphChi system? I mean , except the memory usage computed above, is there something else also taking memory space during the iteration?
Hope that I explain my confusion clearly and someone can provide some clue or answer =)