Commit 26565398 authored by Tilman Steinweg's avatar Tilman Steinweg
Browse files

modified Readmes

parent ceb1fcca
......@@ -12,7 +12,7 @@ gcc:
- echo "g++ Version:" && g++ --version
- mkdir build && cd "$_"
- SCAI_DIR=/data14/software/scai_lama/build_dev/ Geographer_DIR=/data14/software/geographer/build cmake -DSOFI_VALUE_TYPE=float ../src/ -DCMAKE_INSTALL_PREFIX=./
- SCAI_DIR=/data14/software/scai_lama/build_dev/ Geographer_DIR=/data14/software/geographer/build cmake ../src/ -DCMAKE_INSTALL_PREFIX=./
- make install -j 4
when: on_success
......@@ -8,7 +8,7 @@ Get started:
- See `` in `par/`
- Type `make doc` in `build/`
- Type `make doc` or ?make pdf` in build directory
- See the contributing guidelines provided in ``
......@@ -8,7 +8,7 @@ Before the modelling code can be started, you have compile the code using cmake.
Further information can be found in `README.cmake` in the main directory
- ``mkdir build && cd build``
- ``SCAI_DIR="lama install directory" cmake -DSOFI_VALUE_TYPE=float ../src/ -DCMAKE_INSTALL_PREFIX=./``
- ``SCAI_DIR="lama install directory" cmake ../src/ -DCMAKE_INSTALL_PREFIX=./``
- ``make install -j 4``
......@@ -17,6 +17,11 @@ The simulation can be started by the example start file:
Switch of Lama warnings and Tracing
` export SCAI_TRACE=OFF`
Different kinds of parallelization are possible:
- Using OpenMP (shared-memory parallel):
......@@ -26,6 +31,12 @@ Different kinds of parallelization are possible:
The standard configuration of `` is using both kinds of parallelization.
Run Code on GPUs
`export SCAI_DEVICE=0,1,2,3 (...)` (Comma seperated list of GPU devices on one node
## Run the tests
To test the proper functionality of the installation, you can run the build in unit and integration tests.
Since the [Google Test framework]( is used for the unit tests, the environment variable `GTEST_DIR` has to be set to the location of the compiled Google Test library (`libgtest.*` and `libgtest_main.*`). The integration test has no special requirements apart the LAMA framework and MPI.
......@@ -34,12 +45,5 @@ Since the [Google Test framework]( is used
If one of the test cases fails, the script will give an error message and will abort.
## Data processing for parallel I/O
Since the modelling code supports parallel I/O operations some pre- and post-processing steps maybe required to split or merge data. In order to use the binaries mentioned below, you have to build [LAMA]( with the option `BUILD_EXAMPLES=ON`.
#### Data pre-processing
- to partition a single vector-file `${SCAI_ROOT}/lama/examples/io/vectorRepartition.exe filename.mtx 1 filename.%r.mtx {NProcessors}`
#### Data post-processing
- Merge to a single vector-file `${SCAI_ROOT}/lama/examples/io/vectorRepartition.exe filename.%r.mtx {NProcessors} filename.mtx 1`
# Parallel IO
specify .lmf or .su as file type in the configuration file to use parallel in and output.
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment