__host__ __device__ CUDA macro added to all header functions under Imath/ #41
Conversation
…and msc needed functions to compile with CUDA Signed-off-by: Owen Thompson <oxt3479@rit.edu>
Signed-off-by: Owen Thompson <oxt3479@rit.edu>
…compilation of quaternion with CUDA) Signed-off-by: Owen Thompson <oxt3479@rit.edu>
…ion. Signed-off-by: Owen Thompson <oxt3479@rit.edu>
Signed-off-by: Owen Thompson <oxt3479@rit.edu>
Signed-off-by: Owen Thompson <oxt3479@rit.edu>
Signed-off-by: Owen Thompson <oxt3479@rit.edu>
Signed-off-by: Owen Thompson <oxt3479@rit.edu>
|
It looks like maybe you have some methods declared constexpr but defined without, or maybe the other way around. They need to match. |
Signed-off-by: Owen Thompson <oxt3479@rit.edu>
|
It looks like Rebase and merge is causing issues. Is squashing all these into a single commit an acceptable solution, or should that be avoided? |
|
It should be fine to squash.
…On Wed, Aug 12, 2020 at 11:40 AM owdt ***@***.***> wrote:
It looks like Rebase and merge is causing issues. Is squashing all these
into a single commit an acceptable solution, or should that be avoided?
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#41 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AFC3DGLEJCIBI6PWJQXXASTSALOZXANCNFSM4P46VWSA>
.
--
Cary Phillips | R&D Supervisor | ILM | San Francisco
|
|
If it were up to me, squash and merge would be the only possible choice. If you're adding fixes in response to review comments or CI failures on the original PR, then rebase & merge will end up with the repo containing commits that are known to be broken. That seems like a terrible idea, and really inhibits "git bisect" among other things. Author the code with as many individual commits as you want, but each accepted PR should result in exactly one commit being added to the head of master. |
Changes include the addition of exceptionless versions of a few functions included under ImathFrustum.h, specifically:
This is included in this PR so that these functions can be used on the CUDA kernel.
Additionally, Macros were used in ImathRoots.h to deal with std::complex host only.
These functions should be unchanged when compiling normally but when compiling for CUDA instead of std::complex thrust::complex and its associated functions are used.