Commit message (Collapse) | Author | Age | |
---|---|---|---|
* | Make CMake config file relocatable | 2016-12-05 | |
| | |||
* | fix compilation of BTL's blaze interface | 2016-12-05 | |
| | |||
* | Add performance monitoring for LLT | 2016-12-05 | |
| | |||
* | use numext::abs | 2016-12-02 | |
| | | | | | (grafted from 0a08d4c60b652d1f24b2fa062c818c4b93890c59 ) | ||
* | 1. Add explicit template to abs2 (resolves deduction for some arithmetic types) | 2016-12-02 | |
| | | | | | | 2. Avoid signed-unsigned conversion in comparison (warning in case Scalar is unsigned) (grafted from 4086187e49760d4bde72750dfa20ae9451263417 ) | ||
* | Fix memory leak in Ref<Sparse> | 2016-12-05 | |
| | |||
* | Optimize SparseLU::solve for rhs vectors | 2016-12-05 | |
| | |||
* | remove temporary in SparseLU::solve | 2016-12-05 | |
| | |||
* | bug #1356: fix calls to evaluator::coeffRef(0,0) to get the address of the ↵ | 2016-12-05 | |
| | | | | | | destination by adding a dstDataPtr() member to the kernel. This fixes undefined behavior if dst is empty (nullptr). | ||
* | typo | 2016-12-05 | |
| | |||
* | extend monitoring benchmarks with transpose matrix-vector and triangular ↵ | 2016-12-05 | |
| | | | | matrix-vectors. | ||
* | Improve performance of row-major-dense-matrix * vector products for recent CPUs. | 2016-12-05 | |
| | | | | | This revised version does not bother about aligned loads/stores, and rather processes 8 rows at ones for better instruction pipelining. | ||
* | Clean debugging code | 2016-12-05 | |
| | |||
* | Merged in srvasude/eigen (pull request PR-265) | 2016-12-05 | |
|\ | | | | | | | Add Expm1 support to Eigen. | ||
* | | Add missing generic load methods. | 2016-12-03 | |
| | | |||
* | | Complete rewrite of column-major-matrix * vector product to deliver higher ↵ | 2016-12-03 | |
| | | | | | | | | | | | | | | | | | | | | performance of modern CPU. The previous code has been optimized for Intel core2 for which unaligned loads/stores were prohibitively expensive. This new version exhibits much higher instruction independence (better pipelining) and explicitly leverage FMA. According to my benchmark, on Haswell this new kernel is always faster than the previous one, and sometimes even twice as fast. Even higher performance could be achieved with a better blocking size heuristic and, perhaps, with explicit prefetching. We should also check triangular product/solve to optimally exploit this new kernel (working on vertical panel of 4 columns is probably not optimal anymore). | ||
* | | Merged eigen/eigen into default | 2016-12-02 | |
|\ \ | |||
| | * | Fix small nit where I changed name of plog1p to pexpm1. | 2016-12-02 | |
| | | | |||
| | * | Sync from Head. | 2016-12-02 | |
| | |\ | | |/ | |/| | |||
| | * | Added support for expm1 in Eigen. | 2016-12-02 | |
| | | | |||
| * | | Ease compiler job to generate clean and efficient code in mat*vec. | 2016-12-02 | |
| | | | |||
| * | | Operators += and -= do not resize! | 2016-12-02 | |
| | | | |||
* | | | Makingt default numeric_list works with sycl. | 2016-12-02 | |
| | | | |||
| * | | merge | 2016-12-02 | |
| |\ \ | |||
| * | | | Add perf monitoring for gemv | 2016-12-02 | |
| | | | | |||
| * | | | Re-enable A^T*A action in BTL | 2016-12-02 | |
| | | | | |||
| | * | | bug #1355: Fixed wrong line-endings on two files | 2016-12-02 | |
| |/ / | |||
| * | | typo UIntPtr | 2016-12-01 | |
| | | | | | | | | | | | | | | | (grafted from b6f04a2dd4d68fe1858524709813a5df5b9a085b ) | ||
| * | | fix two warnings(unused typedef, unused variable) and a typo | 2016-12-01 | |
| |/ | | | | | | | | | (grafted from a9aa3bcf50d55b63c8adb493a06c903ec34251c6 ) | ||
| * | Clean up SparseCore module regarding ReverseInnerIterator | 2016-12-01 | |
| | | |||
| * | fix member order | 2016-12-01 | |
| | | |||
| * | Merged in rmlarsen/eigen (pull request PR-256) | 2016-12-01 | |
| |\ | | | | | | | | | | Add a default constructor for the "fake" __half class when not using the __half class provided by CUDA. | ||
| * \ | merge | 2016-12-01 | |
| |\ \ | |/ / |/| | | |||
* | | | Merged eigen/eigen into default | 2016-12-01 | |
|\ \ \ | |||
* | | | | Adding sycl backend for TensorPadding.h; disbaling __unit128 for sycl in ↵ | 2016-12-01 | |
| | | | | | | | | | | | | | | | | TensorIntDiv.h; disabling cashsize for sycl in tensorDeviceDefault.h; adding sycl backend for StrideSliceOP ; removing sycl compiler warning for creating an array of size 0 in CXX11Meta.h; cleaning up the sycl backend code. | ||
| | * | | cmake: remove architecture dependency from Eigen3ConfigVersion.cmake | 2016-11-30 | |
| | | | | | | | | | | | | | | | | | | | | | | | | Also, install Eigen3*.cmake under $prefix/share/eigen3/cmake by default. (grafted from 86ab00cdcfc81ef7ebc92ddc3314d7df67c65354 ) | ||
| | * | | doc: mention the NO_MODULE option and target availability | 2016-11-30 | |
| | | | | | | | | | | | | | | | | | | | | (grafted from 65f09be8d2aaeda054cce574ea14a74b00507011 ) | ||
| | | * | Add a default constructor for the "fake" __half class when not using the | 2016-11-29 | |
| | | | | | | | | | | | | | | | | __half class provided by CUDA. | ||
| | * | | cmake: added Eigen3::Eigen imported target | 2016-11-22 | |
| |/ / | | | | | | | | | | | | | (grafted from a287140f7292b9c15719bc6a3a4494ac7874e3cd ) | ||
| * / | Fix misleading-indentation warnings. | 2016-12-01 | |
|/ / | |||
* | | Cleaned up forward declarations | 2016-11-30 | |
| | | |||
* | | Moved the MemCopyFunctor back to TensorSyclDevice since it's the only caller ↵ | 2016-11-30 | |
| | | | | | | | | and it makes TensorFlow compile again | ||
* | | Added the deallocate_all() method back | 2016-11-30 | |
| | | |||
* | | Simplified includes | 2016-11-30 | |
| | | |||
* | | Improved formatting | 2016-11-30 | |
| | | |||
* | | Merged eigen/eigen into default | 2016-11-30 | |
|\ \ | |||
* | | | Udated the Sizes class to work on AMD gpus without requiring a separate ↵ | 2016-11-30 | |
| | | | | | | | | | | | | implementation | ||
* | | | Added an implementation of numeric_list that works with sycl | 2016-11-30 | |
| | | | |||
| * | | Fix slection of product implementation for dynamic size matrices with fixed ↵ | 2016-11-30 | |
| | | | | | | | | | | | | max size. | ||
| * | | Pulled latest update from trunk | 2016-11-30 | |
| |\ \ |