Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add more Interactive factor tables #400

Merged
merged 14 commits into from
Aug 14, 2022
Merged

Conversation

szarnyasg
Copy link
Member

@szarnyasg szarnyasg commented Jun 25, 2022

Fixes #219.

Wishlist:

  • Make factor generation configurable (BI/Interactive/both/neither)
  • Optimize the queries. I'm afraid the foaf ones will be slow on large SFs. These are better computer by reusing the friend ones --> foaf_value(p) = sum_f(friends_value(f)) for all friends(p, f).

@szarnyasg szarnyasg force-pushed the interactive-more-factor-tables branch from 5a4c8b9 to 85917ee Compare June 25, 2022 21:04
@szarnyasg szarnyasg marked this pull request as ready for review June 25, 2022 21:04
@szarnyasg szarnyasg force-pushed the interactive-more-factor-tables branch from 85917ee to 64fe791 Compare June 25, 2022 21:06
@szarnyasg szarnyasg marked this pull request as draft August 13, 2022 12:02
…nNumFriendOfFriendForums, personNumFriendOfFriendCompanies
@szarnyasg
Copy link
Member Author

szarnyasg commented Aug 14, 2022

The optimizations for foaf queries brought down the overall runtime for SF1 from 1 hour to 2.5 minutes (!), running locally on an MBP:

scripts/build.sh
export LDBC_SNB_DATAGEN_JAR=$(sbt -batch -error 'print assembly / assemblyOutputPath')
rm -rf out/
time ./tools/run.py \
  --parallelism $(nproc) --memory ${LDBC_SNB_DATAGEN_MAX_MEM} \
  -- \
  --format parquet --scale-factor 1 --generate-factors

@szarnyasg szarnyasg marked this pull request as ready for review August 14, 2022 15:19
@szarnyasg szarnyasg merged commit 44cf9c2 into main Aug 14, 2022
@szarnyasg szarnyasg deleted the interactive-more-factor-tables branch August 14, 2022 18:47
@szarnyasg
Copy link
Member Author

The overhead of generating these factors is relatively small. When generating SF10 locally, they increase the total runtime from 4m 19.9s to 4m 37.3s, i.e. by 12.5%

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Factor generation for Interactive
1 participant