The Tensor Algebra Compiler (taco) is a C++ library that computes tensor algebra expressions on sparse and dense tensors. It uses novel compiler techniques to get performance competitive with hand-optimized kernels in widely used libraries for both sparse tensor algebra and sparse linear algebra.
You can use taco as a C++ library that lets you load tensors, read tensors from files, and compute tensor expressions. You can also use taco as a code generator that generates C functions that compute tensor expressions.
Learn more about taco at tensor-compiler.org, in the paper The Tensor Algebra Compiler, or in this talk. To learn more about where taco is going in the near-term, see the technical reports on optimization and formats.
You can also subscribe to the taco-announcements email list where we post announcements, RFCs, and notifications of API changes, or the taco-discuss email list for open discussions and questions.
TL;DR build taco using CMake. Run make test
.
Build taco using CMake 3.4.0 or greater:
cd <taco-directory>
mkdir build
cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
make -j8
Building taco requires gcc
5.0 or newer, or clang
3.9 or newer. You can
use a specific compiler or version by setting the CC
and CXX
environment
variables before running cmake
.
To build taco with the Python API (pytaco), add -DPYTHON=ON
to the cmake line above. For example:
cmake -DCMAKE_BUILD_TYPE=Release -DPYTHON=ON ..
You will then need to add the pytaco module to PYTHONPATH:
export PYTHONPATH=<taco-directory>/build/lib:$PYTHONPATH
This requires Python 3.x and some development libraries. It also requires
NumPy and SciPy to be installed. For Debian/Ubuntu, the following packages
are needed: python3 libpython3-dev python3-distutils python3-numpy python3-scipy
.
To build taco with support for parallel execution (using OpenMP), add -DOPENMP=ON
to the cmake line above. For example:
cmake -DCMAKE_BUILD_TYPE=Release -DOPENMP=ON ..
If you are building with the clang
compiler, you may need to ensure that
the libomp
development headers are installed. For Debian/Ubuntu, this is
provided by libomp-dev
, One of the more specific versions like
libomp-13-dev
may also work.
To build taco for NVIDIA CUDA, add -DCUDA=ON
to the cmake line above. For example:
cmake -DCMAKE_BUILD_TYPE=Release -DCUDA=ON ..
Please also make sure that you have CUDA installed properly and that the following environment variables are set correctly:
export PATH=/usr/local/cuda/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH
export LIBRARY_PATH=/usr/local/cuda/lib64:$LIBRARY_PATH
If you do not have CUDA installed, you can still use the taco cli to generate CUDA code with the -cuda flag.
The generated CUDA code will require compute capability 6.1 or higher to run.
To generate documentation for the Python API:
cd <taco-directory>/python_bindings
make html
Before generating the documentation, you must have built the Python API (by following the instructions above) as well as installed the following dependencies:
pip install sphinx
pip install numpydoc
pip install sphinx-rtd-theme
To run all tests:
cd <taco-directory>/build
make test
Tests can be run in parallel by setting CTEST_PARALLEL_LEVEL=<n>
in the environment (which runs <n>
tests in parallel).
To run the C++ test suite individually:
cd <taco-directory>
./build/bin/taco-test
To run the Python test suite individually:
cd <taco-directory>
python3 build/python_bindings/unit_tests.py
To enable code coverage analysis, configure with -DCOVERAGE=ON
. This requires
the gcovr
tool to be installed in your PATH.
For best results, the build type should be set to Debug
. For example:
cmake -DCMAKE_BUILD_TYPE=Debug -DCOVERAGE=ON ..
Then to run code coverage analysis:
make gcovr
This will run the test suite and produce some coverage analysis. This process
requires that the tests pass, so any failures must be fixed first.
If all goes well, coverage results will be output to the coverage/
folder.
See coverage/index.html
for a high level report, and click individual files
to see the line-by-line results.
The following sparse tensor-times-vector multiplication example in C++ shows how to use the taco library.
// Create formats
Format csr({Dense,Sparse});
Format csf({Sparse,Sparse,Sparse});
Format sv({Sparse});
// Create tensors
Tensor<double> A({2,3}, csr);
Tensor<double> B({2,3,4}, csf);
Tensor<double> c({4}, sv);
// Insert data into B and c
B.insert({0,0,0}, 1.0);
B.insert({1,2,0}, 2.0);
B.insert({1,2,1}, 3.0);
c.insert({0}, 4.0);
c.insert({1}, 5.0);
// Pack inserted data as described by the formats
B.pack();
c.pack();
// Form a tensor-vector multiplication expression
IndexVar i, j, k;
A(i,j) = B(i,j,k) * c(k);
// Compile the expression
A.compile();
// Assemble A's indices and numerically compute the result
A.assemble();
A.compute();
std::cout << A << std::endl;
If you just need to compute a single tensor kernel you can use the taco online tool to generate a custom C library. You can also use the taco command-line tool to the same effect:
cd <taco-directory>
./build/bin/taco
Usage: taco [options] <index expression>
Examples:
taco "a(i) = b(i) + c(i)" # Dense vector add
taco "a(i) = b(i) + c(i)" -f=b:s -f=c:s -f=a:s # Sparse vector add
taco "a(i) = B(i,j) * c(j)" -f=B:ds # SpMV
taco "A(i,l) = B(i,j,k) * C(j,l) * D(k,l)" -f=B:sss # MTTKRP
Options:
...
For more information, see our paper on the taco tools taco: A Tool to Generate Tensor Algebra Kernels.