ACR is a compiler technique combining code instrumentation and a runtime library which will automaticaly optimize a computation kernel at runtime by analysing the dataset.
By instrumenting your code with easy to use pragmas, you will provide the tool with enouth information for it to do it's job.
Given the following stencil kernel doing a mean in a "plus" shape:
double data[M][N];
for (int k = 0; k < P; ++k)
for (int i = 0; i < M-2; ++i)
for (int j = 0; j < N-2; ++j)
data[i+1][j+1] +=
(data[i][j+1] + data[i+1][j] + data[i+1][j+2] + data[i+2][j+1]) / 4.;
Pretend this kernel, a 2D Von Neumann neighbourhood applied P times, is part of a simulation to smoothen the data after an other algorithm and that both are called many times.
If the neighbourhood of the data is between a range of delta you have chosen wisely, you obviously does not need to smoothen them further. This is where ACR shines, you can ask him to monitor the "data" array and change the value of the "P" parameter. This way you can lower the computation overhead by doing less pass of this algorithm where it does not matter that much.
So the by adding the following pragmas, the kernel will automatically generate an optimized kernel depending on the data.
double data[M][N];
#pragma acr grid(25)
#pragma acr monitor(double data[i][j], avg, data_to_strategy)
#pragma acr alternative high (parameter, P = 5)
#pragma acr alternative low (parameter, P = 1)
#pragma acr strategy direct (0, high)
#pragma acr strategy direct (1, low)
for (int k = 0; k < P; ++k)
for (int i = 0; i < M-2; ++i)
for (int j = 0; j < N-2; ++j)
data[i+1][j+1] +=
(data[i][j+1] + data[i+1][j] + data[i+1][j+2] + data[i+2][j+1]) / 4.;
The data_to_strategy is a function matching the data value to a strategy unsigned integer one.
mkdir build && cd build
cmake ..
make
Make sure Doxygen is installed on your system.
- Fedora
sudo dnf install doxygen - ArchLinux
sudo pacman -S doxygen - Debian / Ubuntu
sudo apt-get install doxygen
To build the documentation:
mkdir build && cd build
cmake ..
make doxygen
After that you will find the generated documentation as html or latex
format in the doc folder.
Copyright (C) 2016 Maxime Schmitt
ACR is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with this program. If not, see http://www.gnu.org/licenses/.