Replace MKL with Boost+Eigen3#28
Merged
shelhamer merged 7 commits intoBVLC:boost-eigenfrom Jan 22, 2014
kloudkl:boost-eigen
Merged
Replace MKL with Boost+Eigen3#28shelhamer merged 7 commits intoBVLC:boost-eigenfrom kloudkl:boost-eigen
shelhamer merged 7 commits intoBVLC:boost-eigenfrom
kloudkl:boost-eigen
Conversation
|
FWIW I tried this branch and it worked pretty well. |
Member
|
Thank you for the testing data point. We have plans to merge this soon. |
Member
|
Excellent work @kloudkl ! Thanks for the porting, fixes, and tests. We will benchmark this relative to the MKL implementation soon. Please note that after merging I will rebase the boost-eigen branch to master to ready it for integration. You will need to re-create your tracking branch. |
shelhamer
added a commit
that referenced
this pull request
Jan 22, 2014
Replace MKL with Boost+Eigen3 * commit '70c4320e436f92d0963b2622d20c7435b2f07f30': Fix test_data_layer segfault by adding destructor to join pthread Fix math funcs, add tests, change Eigen Map to unaligned for lrn_layer Fix test stochastic pooling stepsize/threshold to be same as max pooling Fixed FlattenLayer Backward_cpu/gpu have no return value Fixed uniform distribution upper bound to be inclusive Add python scripts to install dependent development libs * commit '9a7d022652d65f44bebc97576a3b4f1b5e559748': Fix test_data_layer segfault by adding destructor to join pthread Fix math funcs, add tests, change Eigen Map to unaligned for lrn_layer Fix test stochastic pooling stepsize/threshold to be same as max pooling Fixed FlattenLayer Backward_cpu/gpu have no return value Fixed uniform distribution upper bound to be inclusive * commit '958f038e9e0b1b1c0c62b9119b323f4d62a3832a': Fix test_data_layer segfault by adding destructor to join pthread Fix math funcs, add tests, change Eigen Map to unaligned for lrn_layer Fix test stochastic pooling stepsize/threshold to be same as max pooling Fixed FlattenLayer Backward_cpu/gpu have no return value Fixed uniform distribution upper bound to be inclusive
jhauswald
added a commit
to jhauswald/caffe
that referenced
this pull request
Jul 11, 2015
ghost
mentioned this pull request
Cysu
pushed a commit
to Cysu/caffe
that referenced
this pull request
Feb 19, 2016
Python wrappers for temporal/spatial prediction
mindcont
pushed a commit
to mindcont/caffe
that referenced
this pull request
Apr 25, 2017
Add instructions for deep-vis toolbox usage on AWS
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
This pull request aims to deal with the differences between Intel MKL and Boost.Random 1ff7241, Eigen::Map 68bf029.
Test failures related to FlattenLayer 842a435, stochastic pooling layer d851962, and DataLayer 6a11fa3 are also resolved.
Finally, minor enhancements to building b5badf7 and installing e853430 are added.
Whether should the Intel MKL related codes be removed completely? Currently they are just commented out. Although considering some users may want to exploit Intel MKL's performance, they could still be kept along with Boost+Eigen3 for a while so that the users could finish the transition in no hurry. This decision would probably produce duplicate codes and maintenance burdens in the long term. What are your opinions?
@rodrigob @shelhamer @lifeiteng
issue: #16