Merge pull request rapidsai#716 from jwyles/bug_fix

[REVIEW] Fixing bug with disappearing get_column_data_ptr function
oorliu · Feb 13, 2020 · 81f776b · 81f776b
2 parents e0ce323 + 0376229
commit 81f776b
Show file tree

Hide file tree

Showing 7 changed files with 32 additions and 14 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -13,6 +13,7 @@
 
 - PR #692 Add check after opening golden result files in C++ Katz Centrality tests.
 - PR #702 Add libcypher include path to target_include_directories
+- PR #716 Fixed bug due to disappearing get_column_data_ptr function in cudf
 
 
 # cuGraph 0.12.0 (04 Feb 2020)

diff --git a/python/cugraph/community/ecg_wrapper.pyx b/python/cugraph/community/ecg_wrapper.pyx
@@ -61,7 +61,8 @@ def ecg(input_graph, min_weight=.05, ensemble_size=16):
     g.adjList.get_vertex_identifiers(&c_index_col)
 
     df['partition'] = cudf.Series(np.zeros(num_verts, dtype=np.int32))
-    cdef uintptr_t c_ecg_ptr = get_column_data_ptr(df['partition']._column)
+    #cdef uintptr_t c_ecg_ptr = get_column_data_ptr(df['partition']._column)
+    cdef uintptr_t c_ecg_ptr = df['partition'].__cuda_array_interface__['data'][0]
 
     if g.adjList.edge_data.dtype == np.float32:
         c_ecg.ecg[int32_t, float] (<Graph*>g, min_weight, ensemble_size, <int32_t*>c_ecg_ptr)

diff --git a/python/cugraph/community/louvain_wrapper.pyx b/python/cugraph/community/louvain_wrapper.pyx
@@ -64,8 +64,9 @@ def louvain(input_graph, max_iter=100):
 
 
     df['partition'] = cudf.Series(np.zeros(num_verts,dtype=np.int32))
-    cdef uintptr_t c_louvain_parts_ptr = get_column_data_ptr(df['partition']._column)
-
+    #cdef uintptr_t c_louvain_parts_ptr = get_column_data_ptr(df['partition']._column)
+    cdef uintptr_t c_louvain_parts_ptr = df['partition'].__cuda_array_interface__['data'][0]
+
     cdef bool single_precision = False
     # this implementation is tied to cugraph.cu line 503
     # cudaDataType_t val_type = graph->adjList->edge_data?

diff --git a/python/cugraph/link_analysis/pagerank_wrapper.pyx b/python/cugraph/link_analysis/pagerank_wrapper.pyx
@@ -79,7 +79,9 @@ def pagerank(input_graph, alpha=0.85, personalization=None, max_iter=100, tol=1.
     #TODO FIX ME when graph class is upgraded to remove gdf_column
     cdef gdf_column c_identifier = get_gdf_column_view(df['vertex'])
 
-    cdef uintptr_t c_pagerank_val = get_column_data_ptr(df['pagerank']._column)
+    #cdef uintptr_t c_pagerank_val = get_column_data_ptr(df['pagerank']._column)
+    cdef uintptr_t c_pagerank_val = df['pagerank'].__cuda_array_interface__['data'][0];
+
     cdef uintptr_t c_pers_vtx = <uintptr_t>NULL
     cdef uintptr_t c_pers_val = <uintptr_t>NULL
     cdef sz = 0
@@ -93,11 +95,15 @@ def pagerank(input_graph, alpha=0.85, personalization=None, max_iter=100, tol=1.
             renumber_df['map'] = input_graph.edgelist.renumber_map
             renumber_df['id'] = input_graph.edgelist.renumber_map.index.astype(np.int32)
             personalization_values = personalization.merge(renumber_df, left_on='vertex', right_on='map', how='left').drop('map')
-            c_pers_vtx = get_column_data_ptr(personalization_values['id']._column)
-            c_pers_val = get_column_data_ptr(personalization_values['values']._column)
+            #c_pers_vtx = get_column_data_ptr(personalization_values['id']._column)
+            c_pers_vtx = personalization_values['id'].__cuda_array_interface__['data'][0]
+            #c_pers_val = get_column_data_ptr(personalization_values['values']._column)
+            c_pers_val = personalization_values['values'].__cuda_array_interface__['data'][0]
         else:
-            c_pers_vtx = get_column_data_ptr(personalization['vertex']._column)
-            c_pers_val = get_column_data_ptr(personalization['values']._column)
+            #c_pers_vtx = get_column_data_ptr(personalization['vertex']._column)
+            c_pers_vtx = personalization['vertex'].__cuda_array_interface__['data'][0]
+            #c_pers_val = get_column_data_ptr(personalization['values']._column)
+            c_pers_val = personalization['values'].__cuda_array_interface__['data'][0]
 
     if (df['pagerank'].dtype == np.float32): 
         c_pagerank.pagerank[int, float](g, <float*> c_pagerank_val, sz, <int*> c_pers_vtx, <float*> c_pers_val,

diff --git a/python/cugraph/traversal/bfs_wrapper.pyx b/python/cugraph/traversal/bfs_wrapper.pyx
@@ -64,8 +64,10 @@ def bfs(input_graph, start, directed=True):
     df['distance'] = cudf.Series(np.zeros(num_verts, dtype=np.int32))
     df['predecessor'] = cudf.Series(np.zeros(num_verts, dtype=np.int32))
     cdef gdf_column c_vertex_col = get_gdf_column_view(df['vertex'])
-    cdef uintptr_t c_distance_ptr = get_column_data_ptr(df['distance']._column)
-    cdef uintptr_t c_predecessors_ptr = get_column_data_ptr(df['predecessor']._column)
+    #cdef uintptr_t c_distance_ptr = get_column_data_ptr(df['distance']._column)
+    cdef uintptr_t c_distance_ptr = df['distance'].__cuda_array_interface__['data'][0]
+    #cdef uintptr_t c_predecessors_ptr = get_column_data_ptr(df['predecessor']._column)
+    cdef uintptr_t c_predecessors_ptr = df['predecessor'].__cuda_array_interface__['data'][0]
 
     g.adjList.get_vertex_identifiers(&c_vertex_col)
 

diff --git a/python/cugraph/traversal/sssp_wrapper.pyx b/python/cugraph/traversal/sssp_wrapper.pyx
@@ -77,8 +77,10 @@ def sssp(input_graph, source):
     df['distance'] = cudf.Series(np.zeros(num_verts, dtype=data_type))
     df['predecessor'] = cudf.Series(np.zeros(num_verts, dtype=np.int32))
 
-    cdef uintptr_t c_distance_ptr = get_column_data_ptr(df['distance']._column)
-    cdef uintptr_t c_predecessors_ptr = get_column_data_ptr(df['predecessor']._column)
+    #cdef uintptr_t c_distance_ptr = get_column_data_ptr(df['distance']._column)
+    cdef uintptr_t c_distance_ptr = df['distance'].__cuda_array_interface__['data'][0]
+    #cdef uintptr_t c_predecessors_ptr = get_column_data_ptr(df['predecessor']._column)
+    cdef uintptr_t c_predecessors_ptr = df['predecessor'].__cuda_array_interface__['data'][0]
 
     g.adjList.get_vertex_identifiers(&c_identifier_col)
 

diff --git a/python/cugraph/utilities/column_utils.pyx b/python/cugraph/utilities/column_utils.pyx
@@ -17,14 +17,18 @@
 # cython: language_level = 3
 
 #from cugraph.structure.graph cimport *
-from cudf._lib.cudf cimport get_column_data_ptr, get_column_valid_ptr
 from libc.stdint cimport uintptr_t
 from libc.stdlib cimport calloc, malloc, free
 
 import cudf
 import cudf._lib as libcudf
 import numpy as np
 
+cdef uintptr_t get_column_data_ptr(col):
+    """
+    Get the data pointer from a cudf.Series object
+    """
+    return col.__cuda_array_interface__['data'][0]
 
 cdef gdf_column get_gdf_column_view(col):
     """
@@ -47,7 +51,8 @@ cdef gdf_column get_gdf_column_view(col):
     if col._column._mask is None:
         valid_ptr = 0
     else:
-        valid_ptr = get_column_valid_ptr(col._column)
+        #valid_ptr = get_column_valid_ptr(col._column)
+        valid_ptr = 0
     cdef uintptr_t category = 0
     cdef gdf_dtype_extra_info c_extra_dtype_info = gdf_dtype_extra_info(
         time_unit=TIME_UNIT_NONE,