The quick-install instructions should be enough to get started for Linux or Mac OS X. See the README for full installation instructions, including Windows support.
These instructions are for BART version 0.3.00. Newer versions
of BART may use slightly different syntax, e.g. CUDA_BASE instead of cuda.top
For Linux,
sudo apt-get install gcc make libfftw3-dev liblapack-dev libpng-devFor Mac OS X, using macports,
sudo port install fftw-3-single gcc47 libpngOptionally, install CUDA and ACML. Note ACML is not supported under Mac OS X.
Make a local Makefile to configure paths specific to your system. Change these defaults as desired.
echo "PARALLEL=1 # speed up build" > Makefile.local
echo "CUDA=1 # compile with CUDA support" >> Makefile.local
echo "cuda.top := /usr/local/cuda" >> Makefile.localFor Linux,
echo "ACML=1 # compile with ACML support" >> Makefile.local
echo "acml.top := /usr/local/acml5.3.1/gfortran64_mp_int64" >> Makefile.localFor Mac,
echo "ACML=0 # No ACML support on Mac" >> Makefile.local
echo "fftw.top := /opt/local/" >> Makefile.localUnder the bart/ directory, run
makeThis will build the binary bart, which is the entry point to the command-line tools.
./bart
#BART. Available commands are:
#bench bitmask bpsense caldir calmat cc
#cdf97 circshift conj conv cpyphs creal
#crop ecalib ecaltwo estdims estvar extract
#fakeksp fft fftmod fftshift filter flip
#fmac homodyne itsense join lrmatrix mip
#nlinv noise normalize nrmse nufft ones
#pattern phantom pics pocsense poisson repmat
#reshape resize rof rsense rss sake
#saxpy scale sdot show slice spow
#svd threshold toimg traj transpose twixread
#version walsh wave zerosUnder the bart/ directory, run
source vars.shThis adds the bart tool and the TOOLBOX_PATH environment variable to your session,
To add this for future sessions, append this line to your .bashrc.
Test the installation by benchmarking the toolbox:
bart bench
# add (md_zaxpy) | 0.1157 0.1189 0.1170 0.1160 0.1169 | Avg: 0.1169 Max: 0.1189 Min: 0.1157
# add (md_zaxpy), contiguous | 0.0727 0.0705 0.0700 0.0711 0.0703 | Avg: 0.0709 Max: 0.0727 Min: 0.0700
# add (for loop) | 0.0404 0.0460 0.0431 0.0423 0.0424 | Avg: 0.0428 Max: 0.0460 Min: 0.0404
# sum (md_zaxpy) | 0.1252 0.1211 0.1190 0.1181 0.1207 | Avg: 0.1208 Max: 0.1252 Min: 0.1181
# sum (md_zaxpy), contiguous | 0.0829 0.0810 0.0857 0.0811 0.0927 | Avg: 0.0847 Max: 0.0927 Min: 0.0810
# sum (for loop) | 0.0421 0.0423 0.0366 0.0392 0.0394 | Avg: 0.0399 Max: 0.0423 Min: 0.0366
# complex transpose | 0.0583 0.0552 0.0602 0.0586 0.0634 | Avg: 0.0592 Max: 0.0634 Min: 0.0552
# complex resize | 0.0019 0.0020 0.0021 0.0021 0.0020 | Avg: 0.0020 Max: 0.0021 Min: 0.0019
# complex matrix multiply | 0.0554 0.0646 0.0562 0.0562 0.0535 | Avg: 0.0572 Max: 0.0646 Min: 0.0535
# batch matrix multiply 1 | 0.0154 0.0141 0.0156 0.0142 0.0141 | Avg: 0.0147 Max: 0.0156 Min: 0.0141
# batch matrix multiply 2 | 0.2275 0.2258 0.2275 0.2262 0.2271 | Avg: 0.2268 Max: 0.2275 Min: 0.2258
# tall matrix multiply 1 | 0.0870 0.0875 0.0969 0.0901 0.0912 | Avg: 0.0905 Max: 0.0969 Min: 0.0870
# tall matrix multiply 2 | 0.0231 0.0199 0.0206 0.0260 0.0249 | Avg: 0.0229 Max: 0.0260 Min: 0.0199
# complex dot product | 0.0135 0.0136 0.0134 0.0137 0.0170 | Avg: 0.0142 Max: 0.0170 Min: 0.0134
# complex dot product | 0.0138 0.0138 0.0133 0.0135 0.0188 | Avg: 0.0146 Max: 0.0188 Min: 0.0133
# real complex dot product | 0.0012 0.0013 0.0012 0.0012 0.0013 | Avg: 0.0012 Max: 0.0013 Min: 0.0012
# l2 norm | 0.0011 0.0010 0.0011 0.0010 0.0015 | Avg: 0.0011 Max: 0.0015 Min: 0.0010
# l1 norm | 0.0154 0.0158 0.0163 0.0160 0.0138 | Avg: 0.0155 Max: 0.0163 Min: 0.0138
# copy 1 | 0.0182 0.0050 0.0059 0.0050 0.0050 | Avg: 0.0078 Max: 0.0182 Min: 0.0050
# copy 2 | 0.0051 0.0052 0.0050 0.0055 0.0053 | Avg: 0.0052 Max: 0.0055 Min: 0.0050
# wavelet soft thresh | 0.0551 0.0546 0.0559 0.0600 0.0588 | Avg: 0.0569 Max: 0.0600 Min: 0.0546
# wavelet soft thresh | 0.0797 0.0643 0.0635 0.0617 0.0590 | Avg: 0.0656 Max: 0.0797 Min: 0.0590