allow torch to link with Intel MKL _ilp64 model (large integers) #1040
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
update my fork from master
# Conflicts: # lib/TH/generic/THBlas.c
update from master
add MKL_ILP64 flag in cmake and pass it to C to switch pointer types in BLAS declarations
add LAPACK_INT to allow ILP^$ model fix compiler warning about return in THTensorMath.c
|
Also, apologies for the un-squashed commits. I yet have to learn the intricacies of Git (I'm a SVN user). |
* Fast transposed copy * Add scatterAdd * lua 5.3 changes and gcc constants * Adding support for ADD_TORCH_LIBRARY macro
merge from torch master
Those modes are already supported by torch.load.
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Outdated suggestions cannot be applied.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.
Sorry for the large pull request.
I improved upon the reverted pull request #981. Thanks for the suggestions.
The goal is to allow compilation and link with Intel MKL _ILP64 (64b integers) model. This is necessary when Torch is linked with another application that itself is linked to _ILP64, otherwise the linker on linux links to the incorrect version resulting in a MKL runtime error.
The default build will not change will not be impacted. To trigger linking to _ILP64, the environment variable MKL_ILP64 has to be defined prior to building.
The INTEL_MKL_DIR environment variable now can be set so that Intel MKL can be found by the build.
I tested this patch on both linux and Windows, both with and without MKL_ILP64 defined.
Again, there should be no impact when MKL_ILP64 is not defined.