Feature: Passing arguments to NVCC compiler (#26)

* Add option to give nvcc extra arguments

* Add test for nvcc options that changes c++ dialect from c++17 to c++14

* Add make and the english language pack to devcontainer to be able to build the documentation

* Update documentation config to automatically import the current version of the package

* Document new --compiler-args argument

* Improve tests coverage by testing for bad arguments and the error output during a failed compilation

* Add IPython to docs requirements to allow the __version__ import for readthedocs env

* Change devcontainer base image to have the latest CUDA toolkit

* Mock the nsight compute tool with a bash script

* Add test to compile with opencv

* Add new page to documentation that contains a new notebook that explains compiling with external libraries

* Add autodocstring vscode extension to devcontainer

* Add function that modifies the default profiler/compiler arguments to allow reusing them in multiple magic command calls

* Update pylint exceptions

* Update contributing instructions

* Change version from 1.0.3 to 1.1.0 due to adding features in a backward-compatible manner

* Install latest CUDA toolkit on the test runner to pass the OpenCV compilation test

* Install opencv in test runner and update code coverage install

* Add CUDA bin to PATH in test and coverage runners

* Add cuda bin to path variable in .bashrc

* Update way to set environment variable PATH in github action

* Change devcontainer base image back to ubuntu:22.04 to match the environment from the test runner

docs/source/conf.py

@@ -6,11 +6,18 @@
 # -- Project information -----------------------------------------------------
 # https://www.sphinx-doc.org/en/master/usage/configuration.html#project-information
 
+import os
+import sys
+
+sys.path.append(os.path.join("..", ".."))
+from nvcc4jupyter.__init__ import __version__  # noqa: E402
+
 project = "nvcc4jupyter"
 copyright = "2024, Andrei Nechaev & Cosmin Stefan Ciocan"
 author = "Andrei Nechaev & Cosmin Stefan Ciocan"
-release = "1.0.1"
-version = "1.0.1"
+release = __version__
+version = __version__
+
 
 # -- General configuration ---------------------------------------------------
 # https://www.sphinx-doc.org/en/master/usage/configuration.html#general-configuration

docs/source/index.rst

@@ -10,4 +10,5 @@ which provides CUDA capable GPUs with the CUDA toolkit already installed.
    :caption: Contents:
 
    usage
+   notebooks
    magics

docs/source/magics.rst

@@ -21,24 +21,40 @@ Usage
    - ``%%cuda``: Compile and run this cell.
    - ``%%cuda -p``: Also runs the Nsight Compute profiler.
    - ``%%cuda -p -a "<SPACE SEPARATED PROFILER ARGS>"``: Also runs the Nsight Compute profiler.
+   - ``%%cude -c "<SPACE SEPARATED COMPILER ARGS"``: Passes additional arguments to "nvcc".
    - ``%%cuda -t``: Outputs the "timeit" built-in magic results.
 
 Options
 -------
 
+.. _timeit:
+
 -t, --timeit
    Boolean. If set, returns the output of the "timeit" built-in
    ipython magic instead of stdout.
 
+.. _profile:
+
 -p, --profile
    Boolean. If set, runs the NVIDIA Nsight Compute profiler whose
    output is appended to standard output.
 
+.. _profiler_args:
+
 -a, --profiler-args
    String. Optional profiler arguments that can be space separated
    by wrapping them in double quotes. See all options here:
    `Nsight Compute CLI <https://docs.nvidia.com/nsight-compute/NsightComputeCli/index.html#command-line-options>`_
 
+.. _compiler_args:
+
+-c, --compiler-args
+   String. Optional compiler arguments that can be space separated
+   by wrapping them in double quotes. They will be passed to "nvcc".
+   See all options here:
+   `NVCC Options <https://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc/index.html#nvcc-command-options>`_
+
+
 .. note::
    If both "\-\-profile" and "\-\-timeit" are used then no profiling is
    done.
@@ -47,10 +63,11 @@ Examples
 --------
 ::
 
-   # compile, run, and profile the code in the cell with the Nsight
-   # compute profiler while collecting only metrics from the
-   # "MemoryWorkloadAnalysis" section.
-   %%cuda --profile --profiler-args "--section MemoryWorkloadAnalysis"
+   # compile, run, and profile the code in the cell with the Nsight compute
+   # profiler while collecting only metrics from the "MemoryWorkloadAnalysis"
+   # section; also provides the "--optimize 3" option to "nvcc" during
+   # compilation to optimize host code
+   %%cuda -p -a "--section MemoryWorkloadAnalysis" -c "--optimize 3"
 
 ------
 

docs/source/notebooks.rst

@@ -0,0 +1,34 @@
+*********
+Notebooks
+*********
+
+This page provides a list of useful Jupyter notebooks written with the
+**nvcc4jupyter** library.
+
+.. note::
+   These notebooks are written for Google's Colab, but you may run them in
+   other environments by installing all expected dependencies. If running in
+   Colab, make sure to set the runtime type to a GPU instance (at the time of
+   writing this, T4 is the GPU offered for free by Colab).
+
+------
+
+.. _compiling_with_external_libraries:
+
+Compiling with external libraries
+=================================
+
+[`NOTEBOOK <https://colab.research.google.com/drive/1iuY46DCwv4hy3SqDhJgFeO8kgpHnzjTh?usp=sharing>`_]
+
+If you need to compile CUDA C++ code that uses external libraries in the host
+code (e.g. OpenCV for reading and writing images to disk) then this section is
+for you.
+
+To achieve this, use the :ref:`compiler-args <compiler_args>` option of the
+:ref:`cuda <cuda_magic>` magic command to pass the correct compiler options
+of the OpenCV library to **nvcc** for it to link the OpenCV code with the
+code in your Jupyter cell. Those compiler options can be provided by the
+`pkg-config <https://www.freedesktop.org/wiki/Software/pkg-config/>`_ tool.
+
+In the notebook we show how to use OpenCV to load an image, blur it with a CUDA
+kernel, and then save it back to disk using OpenCV again.

docs/source/usage.rst

@@ -255,3 +255,47 @@ Running the cell above will compile and execute the vector addition code in the
     SM Active Cycles                cycle       383.65
     Compute (SM) Throughput             %         1.19
     ----------------------- ------------- ------------
+
+Compiler arguments
+------------------
+
+In the same way profiler arguments can be passed to the profiling tool,
+compiling arguments can be passed to **nvcc**:
+
+.. code-block:: c++
+
+    %cuda_group_run --group "vector_add" --compiler-args "--optimize 3"
+
+Running the cell above will compile and execute the vector addition code in the
+"vector_add" group. During compilation, **nvcc** receives the "\-\-optimize"
+option which specifies the optimization level for host code.
+
+Set default arguments
+---------------------
+
+In the case where you execute multiple magic commands with the same compiler or
+profiler arguments you can avoid writing them every time by setting the default
+arguments:
+
+.. code-block:: python
+
+    from nvcc4jupyter import set_defaults
+    set_defaults(compiler_args="--optimize 3", profiler_args="--section SpeedOfLight")
+
+The same effect can be achieved by running "set_defaults" once for each config
+due to the fact that the default value is not changed if an a value is not
+given to the "set_defaults" function.
+
+.. code-block:: python
+
+    from nvcc4jupyter import set_defaults
+    set_defaults(compiler_args="--optimize 3")
+    set_defaults(profiler_args="--section SpeedOfLight")
+
+
+Now we can run the following cell without specifying the compiler and profiler
+arguments once again.
+
+.. code-block:: c++
+
+    %cuda_group_run --group "vector_add" --profile