Commit message (Collapse) | Author | Age | ||
---|---|---|---|---|
... | ||||
* | | | Made the cost model cwiseMax and cwiseMin methods consts to help the PowerPC ↵ | 2016-08-18 | ||
| | | | | | | | | | | | | cuda compiler compile this code. | |||
* | | | Force the inlining of a simple accessor. | 2016-08-18 | ||
| | | | ||||
* | | | Merged in ibab/eigen/double-tensor-reduction (pull request PR-216) | 2016-08-18 | ||
|\ \ \ | | | | | | | | | | | | | Enable efficient Tensor reduction for doubles on the GPU (continued) | |||
| * | | | Fix remaining CUDA >= 300 checks | 2016-08-18 | ||
| | | | | ||||
| * | | | Add the necessary CUDA >= 300 checks back | 2016-08-18 | ||
| | | | | ||||
* | | | | Properly detect the type of the result of a contraction. | 2016-08-16 | ||
| | | | | ||||
* | | | | Fix compilation on CUDA 8 by removing call to h2log1p | 2016-08-15 | ||
| | | | | ||||
* | | | | Use array_prod instead of calling TotalSize since TotalSize is only ↵ | 2016-08-15 | ||
| | | | | | | | | | | | | | | | | available on DSize. | |||
* | | | | Merged in ibab/eigen/extend-log1p (pull request PR-218) | 2016-08-15 | ||
|\ \ \ \ | | | | | | | | | | | | | | | | Fix compilation on CUDA 8 due to missing h2log1p function | |||
| * | | | | Close branch extend-log1p | 2016-08-15 | ||
| | | | | | ||||
* | | | | | Fixed a bug in the documentation. | 2016-08-12 | ||
| | | | | | ||||
* | | | | | bug #1273: Add parentheses when redefining eigen_assert | 2016-08-12 | ||
| | | | | | ||||
* | | | | | bug #1272: Disable assertion when total number of columns is zero. | 2016-08-12 | ||
| | | | | | | | | | | | | | | | | | | | | Also moved assertion to finished() method and adapted unit-test | |||
* | | | | | std::erfcf doesn't exist: use numext::erfc instead | 2016-08-11 | ||
| | | | | | ||||
* | | | | | Don't attempt to optimize partial reductions when the optimized ↵ | 2016-08-08 | ||
| | | | | | | | | | | | | | | | | | | | | implementation doesn't buy anything. | |||
* | | | | | Merged in ibab/eigen/extend-log1p (pull request PR-217) | 2016-08-08 | ||
|\| | | | | | | | | | | | | | | | | | | | Add log1p support for CUDA and half floats | |||
| * | | | | Add log1p support for CUDA and half floats | 2016-08-08 | ||
|/ / / / | ||||
* | | | | Merged in suiyuan2009/eigen/fix_tanh_inconsistent_for_tensorflow (pull ↵ | 2016-08-08 | ||
|\ \ \ \ | | | | | | | | | | | | | | | | | | | | | | | | | | request PR-215) Fix_tanh_inconsistent_for_tensorflow | |||
* | | | | | bug #1272: Let CommaInitializer work for more border cases (enhances fix of ↵ | 2016-08-08 | ||
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | bug #1242). The unit test tests all combinations of 2x2 block-sizes from 0 to 3. | |||
| | * | | | Remove CUDA >= 300 checks and enable outer reductin for doubles | 2016-08-06 | ||
| | | | | | ||||
| * | | | | fix tanh inconsistent | 2016-08-06 | ||
| | | | | | ||||
| * | | | | create fix_tanh_inconsistent branch | 2016-08-06 | ||
|/ / / / | ||||
| * | | | Merge upstream changes | 2016-08-05 | ||
| |\ \ \ | |/ / / |/| | | | ||||
| * | | | Make use of atomicExch for atomicExchCustom | 2016-08-05 | ||
| | | | | ||||
* | | | | Add aliases Eigen_*_DIR to Eigen3_*_DIR | 2016-08-05 | ||
| | | | | | | | | | | | | | | | | This is to make configuring work again after project was renamed from Eigen to Eigen3 | |||
* | | | | Fixed the constructors of the new half_base class. | 2016-08-04 | ||
| | | | | ||||
* | | | | Fixed cut and paste bug in debud message | 2016-08-04 | ||
| | | | | ||||
* | | | | Fixed the isnan, isfinite and isinf operations on GPU | 2016-08-04 | ||
| | | | | ||||
* | | | | Extended a regression test to validate that we basic fp16 support works with ↵ | 2016-08-03 | ||
| | | | | | | | | | | | | | | | | cuda 7.0 | |||
* | | | | Deleted redundant regression test. | 2016-08-03 | ||
| | | | | ||||
* | | | | Check that it's possible to forward declare the hlaf type. | 2016-08-03 | ||
| | | | | ||||
* | | | | Move Eigen::half_impl::half to Eigen::half while preserving the free ↵ | 2016-08-04 | ||
| | | | | | | | | | | | | | | | | functions to the Eigen::half_impl namespace together with ADL | |||
* | | | | Merged in ibab/eigen (pull request PR-206) | 2016-08-03 | ||
|\ \ \ \ | | | | | | | | | | | | | | | | Expose real and imag methods on Tensors | |||
* | | | | | Cleaned up the new float16 test a bit | 2016-08-03 | ||
| | | | | | ||||
* | | | | | Added a test for fp16 | 2016-08-03 | ||
| | | | | | ||||
* | | | | | CUDA_ARCH isn't always defined, so avoid relying on it too much when ↵ | 2016-08-03 | ||
| | | | | | | | | | | | | | | | | | | | | figuring out which implementation to use for reductions. Instead rely on the device to tell us on which hardware version we're running. | |||
* | | | | | List PARDISO solver. | 2016-08-02 | ||
| | | | | | ||||
* | | | | | Change project name to Eigen3, to be compatible with FindEigen3.cmake and ↵ | 2016-08-02 | ||
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Eigen3Config.cmake. This is related to pull-requests 214. | |||
* | | | | | Use numext::conj instead of std::conj | 2016-08-01 | ||
| | | | | | ||||
* | | | | | Avoid unecessary object copies | 2016-08-01 | ||
| | | | | | ||||
* | | | | | Fix vectorization logic for coeff-based product for some corner cases. | 2016-07-31 | ||
| | | | | | ||||
* | | | | | half implementation has been moved to half_impl namespace | 2016-07-29 | ||
| | | | | | ||||
* | | | | | bug #1266: half implementation has been moved to half_impl namespace | 2016-07-29 | ||
| | | | | | ||||
* | | | | | bug #1266: half implementation has been moved to half_impl namespace | 2016-07-29 | ||
| |_|/ / |/| | | | ||||
* | | | | Enable slice-vectorization+inner-unrolling when unaligned vectorization is ↵ | 2016-07-28 | ||
| | | | | | | | | | | | | | | | | allowed. For instance, this permits to vectorize 5x5 matrices (including product) | |||
* | | | | Update and fix Cholesky mini benchmark | 2016-07-28 | ||
| | | | | ||||
* | | | | Vectorize more small product expressions by letting the general assignement ↵ | 2016-07-28 | ||
| | | | | | | | | | | | | | | | | logic decides on the sizes that are OK for vectorization. | |||
* | | | | bug #1264: fix compilation | 2016-07-27 | ||
| | | | | ||||
* | | | | Add instructions for LAPACKE+Accelerate | 2016-07-27 | ||
| | | | | ||||
* | | | | Big 1261: add missing max(ADS,ADS) overload (same for min) | 2016-07-27 | ||
| | | | |