diff options
author | Rasmus Munk Larsen <rmlarsen@google.com> | 2020-04-24 17:29:25 -0700 |
---|---|---|
committer | Rasmus Munk Larsen <rmlarsen@google.com> | 2020-04-28 16:12:47 +0000 |
commit | ab773c7e914633ec4a3ee1f7cdea8b168d3bce1a (patch) | |
tree | 5a82a3071e3c5de39e934ce144023c20f62b2de4 /unsupported/Eigen/CXX11 | |
parent | b47c7779937c1984b7cd2f1d2f8df33d67c396f7 (diff) |
Extend support for Packet16b:
* Add ptranspose<*,4> to support matmul and add unit test for Matrix<bool> * Matrix<bool>
* work around a bug in slicing of Tensor<bool>.
* Add tensor tests
This speeds up matmul for boolean matrices by about 10x
name old time/op new time/op delta
BM_MatMul<bool>/8 267ns ± 0% 479ns ± 0% +79.25% (p=0.008 n=5+5)
BM_MatMul<bool>/32 6.42µs ± 0% 0.87µs ± 0% -86.50% (p=0.008 n=5+5)
BM_MatMul<bool>/64 43.3µs ± 0% 5.9µs ± 0% -86.42% (p=0.008 n=5+5)
BM_MatMul<bool>/128 315µs ± 0% 44µs ± 0% -85.98% (p=0.008 n=5+5)
BM_MatMul<bool>/256 2.41ms ± 0% 0.34ms ± 0% -85.68% (p=0.008 n=5+5)
BM_MatMul<bool>/512 18.8ms ± 0% 2.7ms ± 0% -85.53% (p=0.008 n=5+5)
BM_MatMul<bool>/1k 149ms ± 0% 22ms ± 0% -85.40% (p=0.008 n=5+5)
Diffstat (limited to 'unsupported/Eigen/CXX11')
-rw-r--r-- | unsupported/Eigen/CXX11/src/Tensor/TensorMorphing.h | 9 |
1 files changed, 5 insertions, 4 deletions
diff --git a/unsupported/Eigen/CXX11/src/Tensor/TensorMorphing.h b/unsupported/Eigen/CXX11/src/Tensor/TensorMorphing.h index d04dcae17..f107d1b19 100644 --- a/unsupported/Eigen/CXX11/src/Tensor/TensorMorphing.h +++ b/unsupported/Eigen/CXX11/src/Tensor/TensorMorphing.h @@ -456,7 +456,9 @@ struct TensorEvaluator<const TensorSlicingOp<StartIndices, Sizes, ArgType>, Devi // slice offsets and sizes. IsAligned = false, PacketAccess = TensorEvaluator<ArgType, Device>::PacketAccess, - BlockAccess = TensorEvaluator<ArgType, Device>::BlockAccess, + BlockAccess = TensorEvaluator<ArgType, Device>::BlockAccess && + // FIXME: Temporary workaround for bug in slicing of bool tensors. + !internal::is_same<typename internal::remove_const<Scalar>::type, bool>::value, PreferBlockAccess = true, Layout = TensorEvaluator<ArgType, Device>::Layout, CoordAccess = false, @@ -525,7 +527,6 @@ struct TensorEvaluator<const TensorSlicingOp<StartIndices, Sizes, ArgType>, Devi EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const Dimensions& dimensions() const { return m_dimensions; } - EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE bool evalSubExprsIfNeeded(EvaluatorPointerType data) { m_impl.evalSubExprsIfNeeded(NULL); if (!NumTraits<typename internal::remove_const<Scalar>::type>::RequireInitialization @@ -559,14 +560,14 @@ struct TensorEvaluator<const TensorSlicingOp<StartIndices, Sizes, ArgType>, Devi } return true; } - + #ifdef EIGEN_USE_THREADS template <typename EvalSubExprsCallback> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE void evalSubExprsIfNeededAsync( EvaluatorPointerType data, EvalSubExprsCallback done) { m_impl.evalSubExprsIfNeededAsync(nullptr, [done](bool) { done(true); }); } -#endif // EIGEN_USE_THREADS +#endif // EIGEN_USE_THREADS EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE void cleanup() { m_impl.cleanup(); |