[hotfix] Replace `request_format` and related APIs with `create_format_` (#1924)

* upd

* upd

* upd

* upd

examples/pytorch/gcmc/train_sampling.py

@@ -214,18 +214,6 @@ def thread_wrapped_func(func):
             raise exception.__class__(trace)
     return decorated_function
 
-def prepare_mp(g):
-    """
-    Explicitly materialize the CSR, CSC and COO representation of the given graph
-    so that they could be shared via copy-on-write to sampler workers and GPU
-    trainers.
-    This is a workaround before full shared memory support on heterogeneous graphs.
-    """
-    for etype in g.canonical_etypes:
-        g.in_degree(0, etype=etype)
-        g.out_degree(0, etype=etype)
-        g.find_edges([0], etype=etype)
-
 def config():
     parser = argparse.ArgumentParser(description='GCMC')
     parser.add_argument('--seed', default=123, type=int)
@@ -468,8 +456,8 @@ if __name__ == '__main__':
         run(0, n_gpus, args, devices, dataset)
     # multi gpu
     else:
-        prepare_mp(dataset.train_enc_graph)
-        prepare_mp(dataset.train_dec_graph)
+        dataset.train_enc_graph.create_format_()
+        dataset.train_dec_graph.create_format_()
         procs = []
         for proc_id in range(n_gpus):
             p = mp.Process(target=run, args=(proc_id, n_gpus, args, devices, dataset))

examples/pytorch/graphsage/train_cv.py

@@ -147,18 +147,6 @@ class NeighborSampler(object):
             hist_blocks.insert(0, hist_block)
         return blocks, hist_blocks
 
-def prepare_mp(g):
-    """
-    Explicitly materialize the CSR, CSC and COO representation of the given graph
-    so that they could be shared via copy-on-write to sampler workers and GPU
-    trainers.
-
-    This is a workaround before full shared memory support on heterogeneous graphs.
-    """
-    g.in_degree(0)
-    g.out_degree(0)
-    g.find_edges([0])
-
 def compute_acc(pred, labels):
     """
     Compute the accuracy of prediction given the labels.
@@ -332,7 +320,7 @@ if __name__ == '__main__':
     train_mask = g.ndata['train_mask']
     val_mask = g.ndata['val_mask']
     g.ndata['features'] = features
-    prepare_mp(g)
+    g.create_format_()
     # Pack data
     data = train_mask, val_mask, in_feats, labels, n_classes, g
 

examples/pytorch/graphsage/train_cv_multi_gpu.py

@@ -175,18 +175,6 @@ def thread_wrapped_func(func):
             raise exception.__class__(trace)
     return decorated_function
 
-def prepare_mp(g):
-    """
-    Explicitly materialize the CSR, CSC and COO representation of the given graph
-    so that they could be shared via copy-on-write to sampler workers and GPU
-    trainers.
-
-    This is a workaround before full shared memory support on heterogeneous graphs.
-    """
-    g.in_degree(0)
-    g.out_degree(0)
-    g.find_edges([0])
-
 def compute_acc(pred, labels):
     """
     Compute the accuracy of prediction given the labels.
@@ -397,7 +385,7 @@ if __name__ == '__main__':
     g.ndata['features'] = features.share_memory_()
     create_history_storage(g, args, n_classes)
 
-    prepare_mp(g)
+    g.create_format_()
     # Pack data
     data = train_mask, val_mask, in_feats, labels, n_classes, g
 

examples/pytorch/graphsage/train_sampling.py

@@ -89,18 +89,6 @@ class SAGE(nn.Module):
             x = y
         return y
 
-def prepare_mp(g):
-    """
-    Explicitly materialize the CSR, CSC and COO representation of the given graph
-    so that they could be shared via copy-on-write to sampler workers and GPU
-    trainers.
-
-    This is a workaround before full shared memory support on heterogeneous graphs.
-    """
-    g.in_degree(0)
-    g.out_degree(0)
-    g.find_edges([0])
-
 def compute_acc(pred, labels):
     """
     Compute the accuracy of prediction given the labels.
@@ -241,9 +229,9 @@ if __name__ == '__main__':
     else:
         train_g = val_g = test_g = g
 
-    prepare_mp(train_g)
-    prepare_mp(val_g)
-    prepare_mp(test_g)
+    train_g.create_format_()
+    val_g.create_format_()
+    test_g.create_format_()
     # Pack data
     data = in_feats, n_classes, train_g, val_g, test_g
 

examples/pytorch/graphsage/train_sampling_multi_gpu.py

@@ -90,18 +90,6 @@ class SAGE(nn.Module):
             x = y
         return y
 
-def prepare_mp(g):
-    """
-    Explicitly materialize the CSR, CSC and COO representation of the given graph
-    so that they could be shared via copy-on-write to sampler workers and GPU
-    trainers.
-
-    This is a workaround before full shared memory support on heterogeneous graphs.
-    """
-    g.in_degrees(0)
-    g.out_degrees(0)
-    g.find_edges([0])
-
 def compute_acc(pred, labels):
     """
     Compute the accuracy of prediction given the labels.
@@ -269,9 +257,9 @@ if __name__ == '__main__':
     else:
         train_g = val_g = test_g = g
 
-    prepare_mp(train_g)
-    prepare_mp(val_g)
-    prepare_mp(test_g)
+    train_g.create_format_()
+    val_g.create_format_()
+    test_g.create_format_()
     # Pack data
     data = in_feats, n_classes, train_g, val_g, test_g