* Start adding support for running darknet from OpenGL textures
* Get CUDA-GL interop working (at least without segfaults...)
* Copy the GL texture to the CUDA input buffer
* Remove printf statements
* Code cleanup
* Fix a function call
* Remove the #ifdef GPU guard in darknet.h
* Use the width and height from the network instead of hard-coding
* Install libgles2-mesa-dev for some ubuntu targets
* Get the build working on Windows
* Add an OpenGL-CUDA integration option for CMake
* Bump the version-date in vcpkg.json
* Remove ifdef in darknet.h for the network_predict_gl_texture() function
* Use target_compile_definitions to set CUDA_OPENGL_INTEGRATION flag
* Re-add the #ifdef CUDA_OPENGL_INTEGRATION guard
---------
Co-authored-by: Brian Schwind <brianmschwind@gmail.com>
* re-enable opencv on osx
* enable all and all-major architecture flags for cuda
* update build script
* use proper path separator
* [windows-cuda] bump to 11.6.1
* [windows-cuda] some fixes from CI runs
* make deploy-cuda.ps1 work also when called from outside its folder
* bump to cuda 11.6 also on linux
* deploy custom cmake modules if requested by the user
* fix cmake module source path
* use proper clause for dll copy
* [build.ps1] do not export vcpkg_root anymore
* [README.md] add a build step related to Set-ExecutionPolicy on Windows
* use scope currentuser to avoid admin powershell
* simplify even more set-executionpolicy command order
* remove unnecessary decoration to shell commands to improve user experience with copy button
* write instructions to build using only cmake
* Update README.md
* add target install
* improve setup scripts for future tests
* improve setup.sh script for more tool installations and add command line flags handling
* add CI tests for setup scripts
* use nuget artifact cache to improve CI build times
* disable interactivity for build script when called from setup script, add kitware gpg key in CI
* enable cudnn in setup.sh
* add a workaround for missing nvidia drivers in CI
* setup CUDA env variables
* remove some very slow builds from travis and move them to github actions
* chmod sh scripts
* use deploy-cuda scripts also in PR pipelines
* install asm tools
* restore links missing for CI when using deploy-cuda.sh
* enable cudnn on ubuntu ci pipelines
* pass CUDA env variables
* add missing cuda install for opencv2 and opencv3 ci pipelines
* remove unnecessary steps
Co-authored-by: Alexey <AlexeyAB@users.noreply.github.com>
* [build.ps1] use x64 also on developer powershell for VS
* improve handling of errors on windows platform
* [build.ps1] bump version
* improve logic handling in uselib_track
* test for shared libs
* improve flag handling
* build so lib in ci
* clone vcpkg if not found
* fix
* improve clang handling
* bump vcpkg.json version
* gemm.c compatible with ARM
* remove unnecessary vcpkg.json field
* do not unnecessarily rebuild vcpkg in the CI build stage
* use alexeyab nuget cache
* enable self-removal of build folder
* add interactivity in build.ps1 if not opt-out
* spellcheck README
* fix another leftover spelling error
* also auto-update darknet if possible
* do not self-update darknet in CI
* update to run-vcpkg@v6
* bump to cmake 3.18
* use newer cuda syntax, fail if cuda enabled but not found
* update win build script
* improve travis with non-cuda-enabled builds
* improve windows building process
* CMake: Fail generation if CUDA was requested but not found
..instead of resorting to CPU build silently.
This makes builds predictable and
removes the confusion of `ENABLE_CUDA` option being merely a suggestion.
Signed-off-by: Ievgen Popovych <ievgenp@seetrue.ai>
* CMake: Be more precise with ZED SDK messages
CMake generation with `ENABLE_CUDA=OFF` and `ENABLE_ZED_CAMERA=OFF`
would spit out "ZED SDK not enabled, since it requires CUDA" messages
even though ZED SDK was explicitly disabled.
Fix this by being a bit more precise with messages.
Signed-off-by: Ievgen Popovych <ievgenp@seetrue.ai>
* CMake: Honor CUDNN_ENABLE option
Setting CUDNN_ENABLE to OFF would not disable CUDNN support if
CUDNN libraries are available.
Fix this behavior by only looking for CUDNN if it was requested (CUDNN_ENABLE)
and predicate actual settings on CUDNN_FOUND.
Signed-off-by: Ievgen Popovych <ievgenp@seetrue.ai>
* github: actions: Explicitly set CUDA state in CI
Stefano Sinigardi