From ROOT_DIR:

cmake -S . -B build
cmake --build build
ctest --test-dir build --output-on-failure

Build Options

• Build DEBUG version with sanitizer check: cmake -S . -B build -DCMAKE_BUILD_TYPE=Debug -DENABLE_SANITIZERS=ON; run the tests (see below) to utilize sanitizer checking.
• Build enabling SIMD operations cmake -S . -B build -DDSO_SIMD=ON.
• Build enabling pre-computation of a series of SH factors that include square roots (see dso::CunninghamWeights). This will most probably make the code faster but augment required memory resources (~1 to 1.5 Mb). To use this optimization, compile with PRECOMPUTED_SQRT_SHFACS=ON.

Testing

example running tests on parallel: ctest --test-dir build -j8 --output-on-failure

Benchmarks

axpy for MatrixStorageType::LwTriangularColWise

• build the benchmark with no DSO_SIMD:

cmake -S . -B build-simd-off -DCMAKE_BUILD_TYPE=Release -DDSO_SIMD=OFF -DENABLE_BENCHMARKS=ON
cmake --build build-simd-off -j

• build the benchmark with DSO_SIMD:

cmake -S . -B build-simd-on -DCMAKE_BUILD_TYPE=Release -DDSO_SIMD=ON -DENABLE_BENCHMARKS=ON
cmake --build build-simd-on -j

• compare:

./build-simd-off/bench_axpy_lwtricolwise
./build-simd-on/bench_axpy_lwtricolwise

Precomputing SH (Cunningham) Weight

• build the benchmark with no PRECOMPUTED_SQRT_SHFACS=[ON|OFF] and then compare using the program bench/bench_sh2gradient.