Skip to content

Commit

Permalink
last minute commits
Browse files Browse the repository at this point in the history
  • Loading branch information
kab163 committed Aug 15, 2023
1 parent b863d26 commit 2118526
Show file tree
Hide file tree
Showing 3 changed files with 17 additions and 17 deletions.
2 changes: 1 addition & 1 deletion Intermediate_Tutorial/01-SEQ/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,7 @@ documentation:
https://raja.readthedocs.io/en/develop/sphinx/user_guide/feature/policies.html#raja-loop-kernel-execution-policies

The `seq_exec` policy is a good first step because it allows us to get a sense of the
performance using serial nested seqs with RAJA.
performance using sequential, nested loops with RAJA.
From here, we have a good baseline to compare against when transitioning to
CUDA, HIP, etc.

Expand Down
28 changes: 14 additions & 14 deletions Intermediate_Tutorial/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -11,32 +11,32 @@ as OpenMP and even a native CUDA implementation just for comparison. You can ref
these implementation to study the differences in implementation and runtime comparison.
However, anything beyond that is outside the scope of this tutorial.

In the following lessons, you will compare RAJA implementations of a fractal generating code.
In the following lessons, you will compare RAJA implementations of the fractal generating code.
We will start with a sequential implementation of the fractal and gradually build our
way up to a more complex RAJA launch implementation.
You will notice that these lessons will employ the RAJA kernel and launch abstractions.
Additionally, as the lessons progress, we will be exploring the performance portability
of RAJA by looking at how we can change the targeted backend hardware from CUDA to HIP.
of RAJA by looking at how we can change the targeted backend from CUDA to HIP.
(Refer to lessons 02-CUDA and 03-HIP).

As before the exercises have COMPILE macro guards, to compile the code uncomment the
COMPILE define on top of the file.
As before, the exercises have COMPILE macro guards. To compile the code, uncomment the
COMPILE define at the top of the file.

To start, let's build the tutorial within a newly created, empty `build` directory located
in the `raja-suite-tutorial` repo::
If you are doing this tutorial outside of the RADIUSS tutorial series, be sure to build
the tutorial within a newly created, empty `build` directory located
in the `raja-suite-tutorial` repo. If you're on a LC machine, you can run these commands:
```
module load cuda/11.2.0
module load cmake/3.20.2
module load gcc/8.3.1
cmake -DENABLE_CUDA=On -DENABLE_OPENMP=Off -DCMAKE_CUDA_ARCHITECTURES=70 -DCMAKE_CUDA_COMPILER=/usr/tce/packages/cuda/cuda-11.2.0/bin/nvcc -DCUDA_TOOLKIT_ROOT_DIR=/usr/tce/packages/cuda/cuda-11.2.0 -DBLT_CXX_STD=c++14 -DCMAKE_BUILD_TYPE=Release -DRAJA_ENABLE_EXERCISES=On -DRAJA_ENABLE_OPENMP=Off -DCMAKE_CUDA_FLAGS=--extended-lambda -DCUDA_ARCH=sm_70 ../
module load cuda/11.2.0
module load cmake/3.20.2
module load gcc/8.3.1
cmake -DENABLE_CUDA=On -DENABLE_OPENMP=Off -DCMAKE_CUDA_ARCHITECTURES=70 -DCMAKE_CUDA_COMPILER=/usr/tce/packages/cuda/cuda-11.2.0/bin/nvcc -DCUDA_TOOLKIT_ROOT_DIR=/usr/tce/packages/cuda/cuda-11.2.0 -DBLT_CXX_STD=c++14 -DCMAKE_BUILD_TYPE=Release -DRAJA_ENABLE_EXERCISES=On -DRAJA_ENABLE_OPENMP=Off -DCMAKE_CUDA_FLAGS=--extended-lambda -DCUDA_ARCH=sm_70 ../
```

I am building this code on LC's lassen machine. If these build instructions don't work for you, you can refer to the build documentation from RAJA's ReadTheDocs or use one of the provided build scripts.


Now, we can build the RAJA loop-exec implementation with `./bin/fractal 1024`. The first argument
is the width of the fractal (1024). It may be interesting to see how the fractal changes with
different width values.
Now, we can build the RAJA loop-exec implementation with `./bin/fractal 1024`. The
first argument is the width of the fractal (1024). It may be interesting to see how
the fractal changes with different width values.

To verify your results in each lesson, you can look at the resulting .bmp file output. If you
have completed everything correctly, you will see a complete image of the fractal.
Expand Down
4 changes: 2 additions & 2 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -8,8 +8,8 @@ advanced (i.e. Intermediate) tutorial meant for those who would like more
hands-on instruction for RAJA. We would suggest starting with the Intro
tutorial and then moving on to the Intermediate tutorial.

If you would like to build locally, be sure to do the following
to build and run the tutorials:
If you are running on an LC machine and would like to build locally,
be sure to do the following to build and run the tutorials:
```
module load cmake/3.20.2
module load gcc/8.3.1
Expand Down

0 comments on commit 2118526

Please sign in to comment.