DESCRIPTION
The artifact describes LARS, an automated framework to perform reordering optimization on straight-line codes; the algorithmic details of the framework are described in the SC'18 paper "Associative Instruction Reordering to Alleviate Register Pressure". A copy of the paper is in the main directory (sc-18.pdf).
The downloaded package comes with a. item The source code for LARS b. the benchmarks in the examples/ directory c. scripts for code installation and benchmarking
DEPENDENCIES
We tested LARS on ubuntu 16.04 and Red Hat Enterprise Linux Server release 6.7 using GCC 5.3.0, LLVM 5.0.0. The following are hardware requirements for compiling LARS:
- flex >= 2.6.0
- bison >= 3.0.4
- cmake >= 3.8
- boost >=1.58
- GCC >=4.9 with c++11 support for compiling
For benchmarking: 6. GCC >= 7.2.0 7. LLVM >= 5.0
STEPS TO INSTALL AND REPRODUCE RESULTS
-
Download and install LLVM. If you cannot get the latest version of LLVM from apt or any other repo, you can download it from http://releases.llvm.org/download.html. The installation steps are in https://llvm.org/docs/GettingStarted.html. I downloaded LLVM into source directory, and created two separate build and install directories. Then, from the build directory, I used the following command to configure LLVM: cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=
pwd/../install ../source/llvm/ -DLLVM_TARGETS_TO_BUILD="X86" -DGCC_INSTALL_PREFIX= -DCMAKE_C_COMPILER=/bin/gcc -DCMAKE_CXX_COMPILER=/bin/g++ You may need to adjust the paths according to your machine configuration. -
Set the PATH and LD_LIBRARY_PATH variables to point to LLVM installation export PATH=/include:$PATH export LD_LIBRARY_PATH=/lib:$LD_LIBRARY_PATH
-
Now go to the LARS download, and simply run 'make all' in the main directory. The makefile will create the 'test' executable.
-
Set the paths to benchmarking gcc and clang in the run-benchmarks.sh script. Also, set the target (multi-core CPU or Xeon Phi).
Some scripts will not run if the paths are not set
-
Go to the examples directory, and run the benchmarking script as './run-benchmarks.sh'. This will create a file 'output.txt' with all the results. Alternatively, you can set the paths to the benchmarking gcc and clang (first 4 lines in run-benchmarks.sh), then go to an independent benchmark directory, execute 'run.sh', and see the printed results on standard output.
Note that the benchmarks may take >40 hours to run.
COPYRIGHT
All files in this archive which do not include a prior copyright are by default included in this tool and copyrighted 2018 Ohio State University.
MORE INFORMATION
For more information on how to add a new benchmark, see the docs/ folder or contact me at [email protected]