def test_quiver_ogbnproducts()

in benchmarks/api/bench_sampler.py [0:0]

• 30 lines of code
• 4 McCabe index (conditional complexity)

def test_quiver_ogbnproducts(mode='GPU'):
  import quiver
  if mode == 'GPU':
    quiver_mode = 'GPU'
  else:
    quiver_mode = 'UVA'
  root = osp.join(osp.dirname(osp.dirname(osp.realpath(__file__))),
                  '..', 'data', 'products')
  dataset = PygNodePropPredDataset('ogbn-products', root)
  train_idx = dataset.get_idx_split()["train"]
  train_loader = torch.utils.data.DataLoader(train_idx,
                                             batch_size=1024,
                                             pin_memory=True,
                                             shuffle=True)
  csr_topo = quiver.CSRTopo(dataset[0].edge_index)
  quiver_sampler = quiver.pyg.GraphSageSampler(csr_topo, [15, 10, 5],
                                               device=0,
                                               mode=quiver_mode)
  total_time = 0
  sampled_edges = 0
  for seeds in train_loader:
    seeds = seeds.to(0)
    torch.cuda.synchronize()
    start = time.time()
    _, _, adjs = quiver_sampler.sample(seeds)
    torch.cuda.synchronize()
    total_time += time.time() - start
    for adj in adjs:
      sampled_edges += adj.edge_index.shape[1]
  print('Sampled Edges per secs: {} M'.format(sampled_edges / total_time / 1000000))