Martin Drozdik
64b29e06b9
bug #956 : Fixed bug in move constructors of DenseStorage which caused "moved-from" objects to be in an invalid state.
2015-02-16 18:18:46 +09:00
Gael Guennebaud
98604576d1
Merged in chtz/eigen-indexconversion (pull request PR-92)
...
bug #877 , bug #572 : Get rid of Index conversion warnings, summary of changes:
- Introduce a global typedef Eigen::Index making Eigen::DenseIndex and AnyExpr<>::Index deprecated (default is std::ptrdiff_t).
- Eigen::Index is used throughout the API to represent indices, offsets, and sizes.
- Classes storing an array of indices uses the type StorageIndex to store them. This is a template parameter of the class. Default is int.
- Methods that *explicitly* set or return an element of such an array take or return a StorageIndex type. In all other cases, the Index type is used.
2015-02-16 15:29:00 +01:00
Gael Guennebaud
45cbb0bbb1
The usage of DenseIndex is deprecated, so let's replace DenseIndex by Index
2015-02-16 15:05:41 +01:00
Gael Guennebaud
cc641aabb7
Remove deprecated usage of expr::Index.
2015-02-16 14:46:51 +01:00
Gael Guennebaud
aa6c516ec1
Fix many long to int conversion warnings:
...
- fix usage of Index (API) versus StorageIndex (when multiple indexes are stored)
- use StorageIndex(val) when the input has already been check
- use internal::convert_index<StorageIndex>(val) when val is potentially unsafe (directly comes from user input)
2015-02-16 13:19:05 +01:00
Benoit Steiner
e2cfddf75f
Pulled latest updates from trunk
2015-02-13 16:21:59 -08:00
Benoit Steiner
0927801a84
Optimized version of the sin(), exp(), log() and sqrt() function for AVX
2015-02-13 16:07:08 -08:00
Benoit Jacob
e972b55ec4
bug #953 - Fix prefetches in 3px4 product kernel
...
This gives a 10% speedup on nexus 4 and on nexus 5.
2015-02-13 14:52:36 -05:00
Gael Guennebaud
fc202bab39
Index refactoring: StorageIndex must be used for storage only (and locally when it make sense). In all other cases use the global Index type.
2015-02-13 18:57:41 +01:00
Gael Guennebaud
fe51319980
Merge Index-refactoring branch with default, fix PastixSupport, remove some useless typedefs
2015-02-13 10:03:53 +01:00
Gael Guennebaud
0918c51e60
merge Tensor module within Eigen/unsupported and update gemv BLAS wrapper
2015-02-12 21:48:41 +01:00
Gael Guennebaud
409547a0c8
update EIGEN_FAST_MATH documentation
2015-02-12 21:04:31 +01:00
Benoit Steiner
f669f5656a
Marked a few functions as EIGEN_DEVICE_FUNC to enable the use of tensors in cuda kernels.
2015-02-10 14:29:47 -08:00
Gael Guennebaud
029d236ceb
merge
2015-02-10 23:12:47 +01:00
Gael Guennebaud
fe25f3b8e3
FMA has been wrongly disabled
2015-02-10 23:11:35 +01:00
Benoit Steiner
cc5d7ff523
Added vectorized implementation of the exponential function for ARM/NEON
2015-02-10 14:02:38 -08:00
Gael Guennebaud
d4ec48575e
Make Block<SparseMatrix> inherit SparseCompressedBase in the case of an inner-panels and fix valuePtr() innerIndexPtr()
2015-02-09 11:14:36 +01:00
Gael Guennebaud
7838fda82c
Add a SparseCompressedBase class providing (un)compressed accessors (like data()/*Stride() for dense matrices),
...
and a CompressedAccessBit flag (similar to DirectAccessBit for dense matrices).
2015-02-07 22:00:46 +01:00
Benoit Steiner
01f7918788
Pulled latest fixes
2015-02-06 05:30:20 -08:00
Gael Guennebaud
b50ffaddf2
merge
2015-02-06 14:27:12 +01:00
Gael Guennebaud
74e460b995
Fix symmetric product
2015-02-06 14:26:24 +01:00
Benoit Steiner
c739102ef9
Pulled the latest changes from the trunk
2015-02-06 05:25:03 -08:00
Benoit Steiner
dcb2a8b184
Added the EIGEN_HAS_CONSTEXPR define
...
Gate the tensor index list code based on the value of EIGEN_HAS_CONSTEXPR
2015-02-06 02:51:59 -08:00
Benoit Jacob
5ef95fabee
bug #936 , patch 3/3: Properly detect FMA support on ARM (requires VFPv4)
...
and use it instead of MLA when available, because it's both more accurate,
and faster.
2015-01-30 17:45:03 -05:00
Benoit Jacob
0f21613698
bug #936 , patch 2/3: Remove EIGEN_VECTORIZE_FMA, was redundant with EIGEN_HAS_SINGLE_INSTRUCTION_MADD
2015-01-30 17:44:26 -05:00
Benoit Jacob
340b8afb14
bug #936 , patch 1.5/3: rename _FUSED_ macros to _SINGLE_INSTRUCTION_,
...
because this is what they are about. "Fused" means "no intermediate rounding
between the mul and the add, only one rounding at the end". Instead,
what we are concerned about here is whether a temporary register is needed,
i.e. whether the MUL and ADD are separate instructions.
Concretely, on ARM NEON, a single-instruction mul-add is always available: VMLA.
But a true fused mul-add is only available on VFPv4: VFMA.
2015-01-31 14:15:57 -05:00
Benoit Jacob
9f99f61e69
bug #936 , patch 1/3: some cleanup and renaming for consistency.
2015-01-30 17:43:56 -05:00
Benoit Jacob
759bd92a85
bug #935 : Add asm comments in GEBP kernels to work around a bug
...
in both GCC and Clang on ARM/NEON, whereby they spill registers,
severely harming performance. The reason why the asm comments
make a difference is that they prevent the compiler from
reordering code across these boundaries, which has the effect
of extending the lifetime of local variables and increasing
register pressure on this register-tight code.
2015-01-30 17:27:56 -05:00
Gael Guennebaud
c6eb84aabc
Enable vectorization of transposeInPlace for PacketSize x PacketSize matrices
2015-01-26 17:09:01 +01:00
Gael Guennebaud
e1f1091fde
Add support for dense ?= diagonal
2015-01-24 10:32:49 +01:00
Gael Guennebaud
279786e987
Fix missing evaluator in outer-product
2015-01-13 10:25:50 +01:00
Gael Guennebaud
ae4644cc68
bug #907 , ARM64: workaround ICE in xcode/clang
2015-01-13 10:03:00 +01:00
Gael Guennebaud
36f7c1337f
bug #907 , ARM64: workaround vreinterpretq_u64_* not defined in xcode/clang
2015-01-13 09:57:37 +01:00
Gael Guennebaud
63974bcb88
Big 907: workaround some missing intrinsics in current NDK's gcc version (ARM64)
2015-01-07 09:44:25 +01:00
Gael Guennebaud
79f4a59ed9
bug #907 : fix compilation with ARM64
2015-01-07 09:41:56 +01:00
Benoit Steiner
9f98650d0a
Ensured that contractions that can be reduced to a matrix vector product work correctly even when the input coefficients aren't aligned.
2015-01-06 09:29:13 -08:00
Gael Guennebaud
f5f6e2c6f4
bug #921 : fix utilization of bitwise operation on enums in first_aligned
2014-12-19 14:41:59 +01:00
Gael Guennebaud
25c7d9164f
bug #920 : fix MSVC 2015 compilation issues
2014-12-18 22:58:15 +01:00
Gael Guennebaud
7dad5f797e
bug #821 : workaround MSVC 2013 issue with using Base::Base::operator=
2014-12-16 13:33:43 +01:00
Christoph Hertzberg
e8cdbedefb
bug #877 , bug #572 : Introduce a global Index typedef. Rename Sparse*::Index to StorageIndex, make Dense*::StorageIndex an alias to DenseIndex. Overall this commit gets rid of all Index conversion warnings.
2014-12-04 22:48:53 +01:00
Gael Guennebaud
775f7e5fbb
bug #697 : make sure empty classes are at the end in case of multiple inheritence
2014-12-02 14:40:19 +01:00
Gael Guennebaud
a819fa148d
Fix MSVC compilation issue
2014-12-02 14:35:31 +01:00
Gael Guennebaud
b1f9f603a0
Simplify return type of diagonal(Index) (and ease compiler job)
2014-11-28 14:39:47 +01:00
Gael Guennebaud
4371911861
Remove useless and non standard numext::atanh2 function.
2014-12-08 16:44:34 +01:00
Gael Guennebaud
bea36925db
bug #876 : implement a portable log1p function
2014-12-08 16:26:53 +01:00
Gael Guennebaud
80ed5bd90c
Workaround various "returning reference to temporary" warnings.
2014-12-05 12:49:30 +01:00
Benoit Steiner
509e4ddc02
Added reduction packet primitives for CUDA
2014-11-19 10:34:11 -08:00
Gael Guennebaud
722916e19d
bug #903 : clean swap API regarding extra enable_if parameters, and add failtests for swap
2014-11-06 09:25:26 +01:00
Gael Guennebaud
c6fefe5d8e
Big 853: replace enable_if in Ref<> ctor by static assertions and add failtests for Ref<>
2014-11-05 16:15:17 +01:00
Gael Guennebaud
ee06f78679
Introduce unified macros to identify compiler, OS, and architecture. They are all defined in util/Macros.h and prefixed with EIGEN_COMP_, EIGEN_OS_, and EIGEN_ARCH_ respectively.
2014-11-04 21:58:52 +01:00
