Ankur Dave
bc234bf0e1
Split vTableReplicated into two RDDs
...
Previously, (vTableReplicated: IndexedRDD[Pid, VertexHashMap[VD]])
stored one hashmap per partition, taking Vid directly to VD.
To take advantage of rxin's new hashmaps (see
rxin/incubator-spark@32a79d6d13 ), this
commit splits that data structure into two RDDs:
(vTableReplicationMap: IndexedRDD[Pid, VertexIdToIndexMap]) stores a map
per partition from vertex ID to the index where that vertex's attribute
is stored. This index refers to an array in the same partition in
vTableReplicatedValues.
(vTableReplicatedValues: IndexedRDD[Pid, Array[VD]]) stores the vertex
data and is arranged as described above.
2013-10-16 19:22:23 -07:00
Ankur Dave
af8e461841
Set serialization properties in GraphSuite
2013-10-16 19:21:24 -07:00
Joseph E. Gonzalez
57ac9073ae
Introducing unique indexedrdd and adding numerous specialized joins
2013-10-16 04:08:22 -07:00
Joseph E. Gonzalez
59700c0c2a
switched to more efficienct implementation of reduce by key
2013-10-16 00:18:37 -07:00
Joseph E. Gonzalez
9058f261fe
Addressing issue where statistics are not computed correctly
2013-10-15 17:39:09 -07:00
Joseph E. Gonzalez
194bb03d16
Resolved closure capture issues by addressing capture through implicit variables.
2013-10-15 15:10:41 -07:00
Joseph E. Gonzalez
7241cf1632
Updating unit tests.
2013-10-15 14:18:03 -07:00
Joseph E. Gonzalez
345e1e94cc
Still trying to resolve issues with capture.
2013-10-15 14:01:38 -07:00
Joseph E. Gonzalez
b64337ec40
Trying to resolve issues with closure capture.
2013-10-15 13:02:17 -07:00
Joseph E. Gonzalez
e7d0320000
More refactoring and documentating including renaming data to attr for vertex and edge data and eliminating the vertex type.
2013-10-15 02:20:06 -07:00
Joseph E. Gonzalez
67bb39c54b
Removing extraneous code
2013-10-14 18:49:05 -07:00
Joseph E. Gonzalez
bff223454a
trying to address issues with GraphImpl being caught in closures.
2013-10-13 22:27:10 -07:00
Joseph E. Gonzalez
637b67da56
merging changes from upstream benchmarking branch
2013-10-13 19:54:09 -07:00
Joseph E. Gonzalez
494472a6cc
Integrated IndexedRDD into graph design.
2013-10-13 19:42:32 -07:00
Dan Crankshaw
1a961dd1f2
Fixed connected components CL params
2013-10-12 01:47:38 +00:00
Dan Crankshaw
1e5535cfcf
Added connected components back
2013-10-11 16:38:52 -07:00
Dan Crankshaw
543a54dffa
Tried to fix some indenting
2013-10-11 16:07:49 -07:00
Dan Crankshaw
c4a23f95c3
Updated code so benchmarks actually run.
2013-10-11 22:57:43 +00:00
Joseph E. Gonzalez
fa2f87ca63
added replication and balance reporting
2013-10-10 14:48:40 -07:00
Joseph E. Gonzalez
5f756fb63f
added support for random vertex cuts
2013-10-10 14:10:47 -07:00
Joseph E. Gonzalez
8dfac4ea8f
added support for random vertex cuts
2013-10-10 14:09:01 -07:00
Dan Crankshaw
9929e7b9a5
Merge branch 'benchmarks' of github.com:amplab/graphx
2013-10-10 13:36:51 -07:00
Dan Crankshaw
4b46d519db
Merge pull request #17 from amplab/product2
...
product 2 change
2013-10-10 13:35:36 -07:00
Reynold Xin
5218e46178
Updated Kryo registration.
2013-10-07 11:48:50 -07:00
Reynold Xin
4f916f5302
Created a MessageToPartition class to send messages without saving the partition id.
2013-10-07 11:31:00 -07:00
Dan Crankshaw
2a8f3db94d
Fixed groupEdgeTriplets - it now passes a basic unit test.
...
The problem was with the way the EdgeTripletRDD iterator worked. Calling
toList on it returned the last value repeatedly. Fixed by overriding
toList in the iterator.
2013-10-06 19:52:40 -07:00
Dan Crankshaw
0d3ea36fd8
Added a groupEdges and a groupEdgeTriplets method. For some reason the groupEdgeTriplets method isn't properly iterating through the set of edges and thus is returning the wrong result. groupEdges seems to be working.
2013-10-06 18:34:23 -07:00
Dan Crankshaw
6cb21ce889
groupEdges() now compiles. Still need some unit tests
2013-10-06 15:33:35 -07:00
Dan Crankshaw
730a3156d3
Added initial groupEdges code. Still a prototype, I haven't figured out quite how it should all work yet.
2013-10-05 19:44:28 -07:00
Dan Crankshaw
bfedbee13a
Edge partitioner now partitions by canonical edge so all edges between two vertices (in either direction) will be sent to same machine.
2013-10-05 16:04:57 -07:00
Dan Crankshaw
e096cbe90e
Added 2D canonical edge partitioner
2013-10-05 15:20:15 -07:00
Dan Crankshaw
da3e123afb
Removed some comments
2013-10-03 18:11:35 -07:00
Dan Crankshaw
1ee60d3b34
Fixed bug in sampleLogNormal
2013-10-03 17:46:37 -07:00
Dan Crankshaw
27b442dc06
Fixed annotation import
2013-10-03 10:29:00 -07:00
Dan Crankshaw
8edd499eff
Added rmat graph generator
2013-10-03 10:21:34 -07:00
Dan Crankshaw
3c3cc1508b
Added initial implementation of lognormal graph generator. Haven't tested it yet.
2013-09-28 16:00:44 -07:00
Ankur Dave
bf05dc7e78
Add a unit test for aggregateNeighbors
2013-09-19 23:45:15 -07:00
Ankur Dave
7cadeffdf4
Merge branch 'master' into aggregateNeighbors-returns-graph
2013-09-19 23:14:26 -07:00
Ankur Dave
f08e520f4c
Initialize sc in GraphSuite to avoid NullPointerException
2013-09-19 23:12:24 -07:00
Ankur Dave
f02d5c8c53
Fix typo in aggregateNeighbors docs
2013-09-19 23:06:37 -07:00
Ankur Dave
d3cbde0085
Import appropriate Spark core classes
2013-09-19 19:29:58 -07:00
Ankur Dave
c278907bf0
Move BytecodeUtils to the right package
2013-09-19 19:28:22 -07:00
Ankur Dave
4c694bd705
Move IndexedRDD and GraphSuite to org.apache.spark
2013-09-19 19:13:07 -07:00
Ankur Dave
4e967af6af
Return Graph from default aggregateNeighbors also
2013-09-18 16:18:33 -07:00
Ankur Dave
b04f1a4019
Implement aggregateNeighbors returning Graph
2013-09-18 16:18:33 -07:00
Ankur Dave
9ff783599b
Return Graph from aggregateNeighbors; update callers
...
This commit only affects the Graph API, not GraphImpl.
2013-09-18 16:18:33 -07:00
Joseph E. Gonzalez
55696e2584
GraphX now builds with all merged changes.
2013-09-17 22:42:12 -07:00
Joseph E. Gonzalez
5ccb60d467
Working on graph test suite
2013-08-11 14:49:22 -07:00
Joseph E. Gonzalez
ddf126edad
added subgraph
2013-08-06 17:48:04 -07:00
Joseph E. Gonzalez
b454314e07
Added 2d partitioning
2013-08-06 15:14:13 -07:00