Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Thrust voting #62

Merged
merged 15 commits into from
Jul 18, 2017
Merged

Thrust voting #62

merged 15 commits into from
Jul 18, 2017

Conversation

griwodz
Copy link
Member

@griwodz griwodz commented Jul 10, 2017

Compiling and running on CUDA 8 works now. Tested only on the GTX 980 Ti so far.

CUB code is now compiled for CUDA < 8, Thrust code for CUDA >= 8
Note that CUB code with CUDA 7.5 will still fail on GTX 980 cards and other modern cards.

Test program detection can now use DevIL to load more image formats.

The CI on travis is also updated to test the build with different cuda versions (connected to #49) and upgrading to trusty (connected to #41 )

Carsten Griwodz and others added 9 commits July 10, 2017 09:43
(ext == ".png") ||
(ext == ".ppm") ||
(ext == ".tga") ||
(ext == ".tif") )
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Are these file formats not supported by opencv, which is already in the dependencies?

- compiler: gcc
env: CXX_COMPILER=g++-4.8 C_COMPILER=gcc-4.8 TOOLSET=gcc VISUAL_DEBUG=OFF CCTAG_SERIALIZE=ON WITH_CUDA=OFF
env: CXX_COMPILER=g++-4.8 C_COMPILER=gcc-4.8 TOOLSET=gcc WITH_CUDA=ON CUDA_VERSION_MAJOR="8" CUDA_VERSION_MINOR="0" CUDA_PKG_LONGVERSION="${CUDA_VERSION_MAJOR}.${CUDA_VERSION_MINOR}.61-1" CUDA_PKG_VERSION="${CUDA_VERSION_MAJOR}-${CUDA_VERSION_MINOR}"
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it not better to add a new variant?
In order to keep one build variant with "VISUAL_DEBUG=ON CCTAG_SERIALIZE=ON"?
To ensure that the code with debug exports is builded once and not broken by new changes.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I restored it but not as a variant in the matrix: at the end of the build(s) without cuda, it restarts with those flags. It avoids to explose the matrix of possibilities and also it's faster.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

incidentally i fixed #41 and #49, so now it also builds clang with cuda, using boost and eigen from the repositories

simogasp added 5 commits July 12, 2017 20:43
removed boost #41
Now when the builds are not with CUDA, it rebuilds the library at the
end with the two flags enabled. This spares some time as builds wihout
cuda are much faster and thus avoiding overheads of loading the whole
system at the beginning.
@simogasp
Copy link
Member

Before merging maybe update the README/BUILD with the updated info: required/supported CUDA versions, supported HW etc
@griwodz @fabiencastan

@fabiencastan fabiencastan merged commit 6c2e4e2 into develop Jul 18, 2017
@fabiencastan fabiencastan deleted the thrust-voting branch July 18, 2017 15:06
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants