Validation of random graph construction #7
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
After the graph has been constructed, dumps all the edges to a list and compares with the list of randomly generated edges.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Parallel random graph construction in GraphBIG is not implemented correctly.
To illustrate the problem, I added validation code to the end of the parallel random graph construction benchmark. It dumps the graph back to an edge list after construction, and then compares with the randomly generated list the graph was constructed from. When I run with more than 1 thread, the validation usually fails, although it occasionally passes.
C++ standard library containers like
vector,list, andunordered_mapare not thread-safe. Callingpush_backconcurrently on the same data structure without synchronization is leading to data corruption in this benchmark. Locking each vertex before adding an edge should solve the problem, but this will have a significant impact on performance.