Added CUDA __host__ __device__ with macro to Vector, Matrix, Limits, … #38
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
…and msc needed functions to compile with CUDA Signed-off-by: Owen Thompson <[email protected]>
meshula
reviewed
Aug 7, 2020
| constexpr static T baseTypeMax() { return limits<T>::max(); } | ||
| constexpr static T baseTypeSmallest() { return limits<T>::smallest(); } | ||
| constexpr static T baseTypeEpsilon() { return limits<T>::epsilon(); } | ||
| IMATH_HOSTDEVICE constexpr static T baseTypeMin() { return limits<T>::min(); } |
There was a problem hiding this comment.
I can't remember why these constexpr haven't been replaced by the IMATH_CONSTEXPR14 macro?
There was a problem hiding this comment.
If it has not been replaced then it means that it is capable of being compiled as constexr under the c++11 standard. If they were to be replaced with IMATH_CONSTEXR14 then it would not be constexr for c++11, but can be for c++14.
These functions are only a return statement which allows for c++11 to compile them as constexpr.
There was a problem hiding this comment.
Oh! Thanks for the explanation!
Signed-off-by: Owen Thompson <[email protected]>
lgritz
approved these changes
Aug 10, 2020
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Outdated suggestions cannot be applied.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.
These changes are verified to compile with CUDA compiler version 11.0 (and should work on earlier versions). All changes made in this PR are in service of facilitating compilation.
Most of the changes are added macros but there are some additional changes that were necessary.
1:
This allows for the CUDA compiler to use device versions of these functions when it is necessary. If this were not the case then the function that calls sqrt/sin/cos (for instance length()) could not be executed on the device
2:
There are now two versions of each of these functions now, one which takes a bool singExc to specify whether it can throw an exception and one which takes in no parameters and never throws. The original implementation defaulted singExc to false. Therefore, the behaviour of any existing code should not change, it may in fact run slightly faster as it does not need to check for singExc if there is no bool passed in.
These changes are focused on Vector and Matrix operations. Changes to other files were necessary as Vector and Matrix operations call some of them (for instance those in ImathLimits.h)