def call_louvain(sID, data, num_verts, num_edges, vertex_partition_offsets,
aggregate_segment_offsets, max_level, resolution):
wid = Comms.get_worker_id(sID)
handle = Comms.get_handle(sID)
local_size = len(aggregate_segment_offsets) // Comms.get_n_workers(sID)
segment_offsets = \
aggregate_segment_offsets[local_size * wid: local_size * (wid + 1)]
return c_mg_louvain.louvain(data[0], num_verts, num_edges,
vertex_partition_offsets, wid, handle,
segment_offsets, max_level, resolution)
def call_sssp(sID, data, num_verts, num_edges, vertex_partition_offsets,
aggregate_segment_offsets, start):
wid = Comms.get_worker_id(sID)
handle = Comms.get_handle(sID)
local_size = len(aggregate_segment_offsets) // Comms.get_n_workers(sID)
segment_offsets = \
aggregate_segment_offsets[local_size * wid: local_size * (wid + 1)]
return mg_sssp.mg_sssp(data[0], num_verts, num_edges,
vertex_partition_offsets, wid, handle,
segment_offsets, start)
def call_bfs(sID, data, src_col_name, dst_col_name, num_verts, num_edges,
vertex_partition_offsets, aggregate_segment_offsets, start,
depth_limit, return_distances):
wid = Comms.get_worker_id(sID)
handle = Comms.get_handle(sID)
local_size = len(aggregate_segment_offsets) // Comms.get_n_workers(sID)
segment_offsets = \
aggregate_segment_offsets[local_size * wid: local_size * (wid + 1)]
return mg_bfs.mg_bfs(data[0], src_col_name, dst_col_name, num_verts,
num_edges, vertex_partition_offsets, wid, handle,
segment_offsets, start, depth_limit, return_distances)
def call_pagerank(sID, data, num_verts, num_edges, vertex_partition_offsets,
aggregate_segment_offsets, alpha, max_iter, tol,
personalization, nstart):
wid = Comms.get_worker_id(sID)
handle = Comms.get_handle(sID)
local_size = len(aggregate_segment_offsets) // Comms.get_n_workers(sID)
segment_offsets = \
aggregate_segment_offsets[local_size * wid: local_size * (wid + 1)]
return mg_pagerank.mg_pagerank(data[0], num_verts, num_edges,
vertex_partition_offsets, wid, handle,
segment_offsets, alpha, max_iter, tol,
personalization, nstart)
def call_katz_centrality(sID, data, src_col_name, dst_col_name, num_verts,
num_edges, vertex_partition_offsets,
aggregate_segment_offsets, alpha, beta, max_iter, tol,
nstart, normalized):
wid = Comms.get_worker_id(sID)
handle = Comms.get_handle(sID)
local_size = len(aggregate_segment_offsets) // Comms.get_n_workers(sID)
segment_offsets = \
aggregate_segment_offsets[local_size * wid: local_size * (wid + 1)]
return mg_katz_centrality.mg_katz_centrality(
data[0], src_col_name, dst_col_name, num_verts, num_edges,
vertex_partition_offsets, wid, handle, segment_offsets, alpha, beta,
max_iter, tol, nstart, normalized)
def call_wcc(sID,
data,
src_col_name,
dst_col_name,
num_verts,
num_edges,
vertex_partition_offsets,
aggregate_segment_offsets):
wid = Comms.get_worker_id(sID)
handle = Comms.get_handle(sID)
local_size = len(aggregate_segment_offsets) // Comms.get_n_workers(sID)
segment_offsets = \
aggregate_segment_offsets[local_size * wid: local_size * (wid + 1)]
return mg_connectivity.mg_wcc(data[0],
src_col_name,
dst_col_name,
num_verts,
num_edges,
vertex_partition_offsets,
wid,
handle,
segment_offsets)
def shuffle(dg, transposed=False):
"""
Shuffles the renumbered input distributed graph edgelist into ngpu
partitions. The number of processes/gpus P = prows*pcols. The 2D
partitioning divides the matrix into P*pcols rectangular partitions
as per vertex partitioning performed in renumbering, and then shuffles
these partitions into P gpus.
Parameters
----------
transposed : bool, optional (default=False)
"""
ddf = dg.edgelist.edgelist_df
ngpus = Comms.get_n_workers()
prows, pcols, partition_type = Comms.get_2D_partition()
renumber_vertex_count = dg.renumber_map.implementation.\
ddf.map_partitions(len).compute()
renumber_vertex_cumsum = renumber_vertex_count.cumsum()
if transposed:
row_dtype = ddf['dst'].dtype
col_dtype = ddf['src'].dtype
else:
row_dtype = ddf['src'].dtype
col_dtype = ddf['dst'].dtype
vertex_partition_offsets = cudf.Series([0], dtype=row_dtype)
vertex_partition_offsets = vertex_partition_offsets.append(
cudf.Series(renumber_vertex_cumsum, dtype=row_dtype))
num_verts = vertex_partition_offsets.iloc[-1]
if partition_type == 1:
vertex_row_partitions = []
for i in range(prows + 1):
vertex_row_partitions.append(vertex_partition_offsets.iloc[i *
pcols])
vertex_row_partitions = cudf.Series(vertex_row_partitions,
dtype=row_dtype)
else:
vertex_row_partitions = vertex_partition_offsets
vertex_col_partitions = []
for i in range(pcols + 1):
vertex_col_partitions.append(vertex_partition_offsets.iloc[i * prows])
vertex_col_partitions = cudf.Series(vertex_col_partitions, dtype=col_dtype)
meta = ddf._meta._constructor_sliced([0])
partitions = ddf.map_partitions(
_set_partitions_pre,
vertex_row_partitions=vertex_row_partitions,
vertex_col_partitions=vertex_col_partitions,
prows=prows,
pcols=pcols,
transposed=transposed,
partition_type=partition_type,
meta=meta)
ddf2 = ddf.assign(_partitions=partitions)
ddf3 = rearrange_by_column(
ddf2,
"_partitions",
max_branch=None,
npartitions=ngpus,
shuffle="tasks",
ignore_index=True,
).drop(columns=["_partitions"])
partition_row_size = pcols
partition_col_size = prows
return (ddf3, num_verts, partition_row_size, partition_col_size,
vertex_partition_offsets)
