default to lmdb for database storage #1128
Merged
Conversation
This file contains 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
lmdb is 10-15% faster than leveldb although it takes ~1.1x the storage. This is usually irrelevant in prefetching since both are fast enough, but more important lmdb allows multiple, concurrent reads for training and evaluation several models on the same data.
|
SGTM |
shelhamer
force-pushed
the
default-db-lmdb
branch
from
8f1afe6
to
73f1de4
Compare
shelhamer
added a commit
that referenced
this pull request
Sep 22, 2014
default to lmdb for database storage
2 tasks
mitmul
pushed a commit
to mitmul/caffe
that referenced
this pull request
Sep 30, 2014
default to lmdb for database storage
RazvanRanca
pushed a commit
to RazvanRanca/caffe
that referenced
this pull request
Nov 4, 2014
default to lmdb for database storage
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
lmdb is faster at read and supports concurrent access for training and evaluating multiple networks on the same data at the same time.
As a further step I vote to switch the caffe.proto default backend to lmdb from leveldb in dev with a note in the release since this changes existing model defaults. Thoughts?
Certain examples like cifar and mnist siamese have their own custom data tools that are left as leveldb.