Rearchitect and Reinstate GPU

[onurulgen]

• Reinstated GPU into the project by fixing or implementing missing functions.
• Rearchitected the project to combine GPU and CPU implementations by using virtual functions.
• Increased the performance up to 11 times by using GPU.
• Implemented unit and regression tests for GPU and CPU by using Catch2 framework.
• Developed GitHub Actions for tests, coverage, static code analysis, and producing CPU and GPU executables for Windows, Linux and macOS.

[codecov]

Welcome to Codecov 🎉

Once you merge this PR into your default branch, you're all set! Codecov will compare coverage reports and display results in all future pull requests.

Thanks for integrating Codecov - We've got you covered ☂️

[github-actions]

✅Code Analysis Results - no issues found! ✅

[msseibel]

I had to add set(CMAKE_CUDA_ARCHITECTURES "native") to the CMakeLists.txt file. Otherwise, I got the following error, "CMAKE_CUDA_ARCHITECTURES must be non-empty if set".

Besides that, the build worked, and I was able to run the program.

[onurulgen]

I had to add set(CMAKE_CUDA_ARCHITECTURES "native") to the CMakeLists.txt file. Otherwise, I got the following error, "CMAKE_CUDA_ARCHITECTURES must be non-empty if set".

Besides that, the build worked, and I was able to run the program.

CMAKE_CUDA_ARCHITECTURES are set in here. If you need to set that, your cmake configure command must've failed while detecting CUDA, right?

[msseibel]

Yes, that is correct. It failed as soon as it reached line 156 enable_language(CUDA).

[Copilot]

Copilot reviewed 670 out of 685 changed files in this pull request and generated 1 comment.

Files not reviewed (15)

• .github/workflows/linux.yml: Language not supported
• .github/workflows/macos.yml: Language not supported
• .github/workflows/windows.yml: Language not supported
• CMakeLists.txt: Language not supported
• Doxyfile.in: Language not supported
• cmake/FindOPENCL.cmake: Language not supported
• cmake/NIFTYREGConfig.cmake.in: Language not supported
• niftyreg_build_version.txt: Language not supported
• reg-apps/CMakeLists.txt: Language not supported
• reg-apps/reg_average.cpp: Language not supported
• reg-apps/reg_benchmark.cpp: Language not supported
• reg-apps/reg_gpuinfo.cpp: Language not supported
• reg-apps/reg_jacobian.cpp: Language not supported
• reg-apps/reg_measure.cpp: Language not supported
• reg-apps/reg_resample.cpp: Language not supported

Comments suppressed due to low confidence (1)

.github/workflows/tests.yml:32

• The use of ${{ matrix.sudo }} might not be necessary for all platforms. Consider removing it or using a conditional statement.

${{ matrix.sudo }} cmake --build build/ --target install --config ${{ matrix.build_type }}

To fix the problem, we need to ensure that the multiplication is performed using the larger type (size_t) to avoid overflow. This can be achieved by casting one of the operands to size_t before performing the multiplication. This way, the entire multiplication will be done in the size_t type, preventing overflow.

We will modify the line where the multiplication occurs to cast image->nbyper to size_t before the multiplication.

To fix the problem, we need to ensure that the multiplication is performed using a larger integer type to avoid overflow. This can be done by casting one of the operands to unsigned long before performing the multiplication. This way, the multiplication will be done using the larger type, and the result will be correctly represented.

The specific change involves casting one of the variables (width, height, or channels) to unsigned long before the multiplication. This change should be made on line 83 of the file reg-io/png/reg_png.cpp.

To fix the problem, we need to ensure that the multiplication is performed using the larger type (double) to avoid overflow. This can be achieved by casting the operands to double before performing the multiplication. Specifically, we need to update the lines where the multiplication occurs to cast array2Ptr[i] and array1Ptr[i] to double.

To fix the problem, we need to ensure that the multiplication is performed using the larger type double to avoid overflow. This can be done by casting the operands to double before performing the multiplication. This change should be made in the file reg-lib/Optimiser.cpp on line 278.

To fix the problem, we need to ensure that the multiplication is performed using double precision to avoid overflow. This can be achieved by casting one of the operands to double before performing the multiplication. Specifically, we will cast array2PtrBw[i] to double in the multiplication expression.

To fix the problem, we need to ensure that the multiplication is performed using a larger integer type to avoid overflow. This can be achieved by casting one of the operands to size_t before performing the multiplication. This way, the multiplication will be done using the larger type, and the result will be correctly handled by the malloc function.

Specifically, we will cast this->activeBlockNumber to size_t before multiplying it by this->dim on lines 35 and 36. This change will ensure that the multiplication is performed using the size_t type, preventing any potential overflow.

To fix the problem, we need to cast the operands of the multiplication to a larger type before performing the multiplication. This will ensure that the multiplication is done using the larger type, preventing overflow. Specifically, we should cast params->activeBlockNumber and params->dim to size_t before multiplying them. This change should be made on line 269 in the file reg-lib/cpu/_reg_blockMatching.cpp.

To fix the problem, we need to ensure that the multiplication is performed using a larger integer type to avoid overflow. We can achieve this by casting one of the operands to size_t before performing the multiplication. This will ensure that the multiplication is done using the larger type, preventing overflow.

• Cast one of the operands to size_t before performing the multiplication.
• Specifically, cast z to size_t on line 106.
• No additional methods, imports, or definitions are needed to implement this change.

To fix the problem, we need to ensure that the multiplication is performed using the larger type (size_t) to prevent overflow. This can be achieved by casting one of the operands to size_t before performing the multiplication. This way, the multiplication will be done using the larger type, and the result will be correctly stored in the size_t variable.

The specific change involves casting referenceBinNumber[currentTimePoint] to size_t before multiplying it with floatingBinNumber[currentTimePoint].

To fix the problem, we need to ensure that the multiplication is performed using a larger integer type to prevent overflow. This can be achieved by casting the operands to size_t before performing the multiplication. This way, the multiplication will be done using the larger type, and the result will be correctly assigned to the size_t variable.