Commits · e55216f580627847b8a0b68c4fe75dea67e3db6f · Kaushik Kulkarni / viennacl-dev · GitLab

Jun 29, 2015
- compressed_matrix: Refined OpenCL kernels for NVIDIA Kepler GPUs. · e55216f5
  Karl Rupp authored Jun 29, 2015
  e55216f5
Jun 24, 2015
- Eigen: Extended wrapper to also support Eigen::Map<> for dense matrices. · 32ff5024
  Karl Rupp authored Jun 24, 2015
  32ff5024
Jun 23, 2015
- SpGEMM: Workaround for bug in NVIDIA OpenCL compiler. · 216a6ac4
  Karl Rupp authored Jun 23, 2015
  216a6ac4
Jun 11, 2015
- CUDA: Added improved CSR SpMV kernel. · b3a6f0e1
  Karl Rupp authored Jun 11, 2015
  b3a6f0e1
Jun 08, 2015
- SpGEMM: Compatibility fixes so that older NVIDIA GPUs can be used as well. · a4ddca5f
  Karl Rupp authored Jun 08, 2015
  a4ddca5f
May 31, 2015
- stdint.h: Finally removed (forgotten in previous commit) · 979abd97
  Karl Rupp authored May 31, 2015
  979abd97
- Visual Studio 2008: Fixed a couple of accidental C++11 features. · 795eec5d
  Karl Rupp authored May 31, 2015
  795eec5d
May 28, 2015
- Armadillo: Added bindings similar to Eigen, MTL4, and Boost.uBLAS. · 373f55d6
  Karl Rupp authored May 28, 2015
  373f55d6
- Merge pull request #113 from d-meiser/add-asan-to-travis · b42c06d0
  Karl Rupp authored May 28, 2015
  b42c06d0
May 27, 2015
- Scan: Fixed incorrect exclusive_scan() for large vector sizes. · 22ec5de1
  Karl Rupp authored May 27, 2015
  22ec5de1
- Scans: Fixed bug in CPU-version of exclusive scan for small vectors. · 74f6136c
  Karl Rupp authored May 27, 2015
  74f6136c
- compressed_matrix: Added support for operator<< · 4e6de156
  Karl Rupp authored May 27, 2015
  4e6de156
- compressed_matrix: generate_row_block_information() now public. · d9d5df1c
  Karl Rupp authored May 27, 2015
  d9d5df1c
- Scan: Added in-place versions, fixed issue in in-place OpenMP implementation. · de4d6ad2
  Karl Rupp authored May 27, 2015
  de4d6ad2
- SpGEMM: Fixed compilation and external linkage problems with CUDA 6.0. · 5318fef6
  Karl Rupp authored May 27, 2015
  5318fef6
May 23, 2015
- Scan: Fixed incomplete migration of OpenCL kernels from SVD module. · 54bf0c38
  Karl Rupp authored May 23, 2015
  
  54bf0c38
May 22, 2015
- Scan: Refurbished CUDA and OpenCL implementations. · aadb5b72
  Karl Rupp authored May 22, 2015
  aadb5b72
May 21, 2015
- SpGEMM: Fixed missing barrier in OpenCL kernels. · 525fc3ae
  Karl Rupp authored May 21, 2015
  525fc3ae
May 20, 2015
- Scan: Added fast OpenMP implementation, fixed OpenCL bug, shifted to vectors. · 54474134
  Karl Rupp authored May 20, 2015
  54474134
- SpGEMM: Minor improvements to OpenCL kernel. · 6347fe4d
  Karl Rupp authored May 20, 2015
  6347fe4d
May 10, 2015
- Merge branch 'karlrupp/sparse-matrix-matrix-product' · 66a8949c
  Karl Rupp authored May 10, 2015
  66a8949c
- SpGEMM: Switched back to dynamic scheduling with OpenMP. · b3e5daa0
  Karl Rupp authored May 10, 2015
  b3e5daa0
May 07, 2015
- SpGEMM: Fixed bug in OpenCL kernel. · f0e57ab7
  Karl Rupp authored May 07, 2015
  f0e57ab7
- compressed_matrix: Improved handling of dimensions for STL matrix. · 0cfd0263
  Karl Rupp authored May 07, 2015
  0cfd0263
- SpGEMM: Fixed warnings and added asserts() to check for proper dimensions. · 5bf86cc7
  Karl Rupp authored May 07, 2015
  
  5bf86cc7
- SpGEMM: Added OpenCL implementation of RMerge. · d1685c1e
  Karl Rupp authored May 07, 2015
  d1685c1e
Apr 27, 2015
- custom-cuda: Fixed incorrect passing of arguments to CUDA kernel. · f29e01e0
  Karl Rupp authored Apr 27, 2015
  f29e01e0
Apr 18, 2015
- SpGEMM: Added AVX2 to numerical stage. · fa7ba2fe
  Karl Rupp authored Apr 18, 2015
  
  fa7ba2fe
- SpGEMM: Broader use of AVX2 in symbolic stage. · c78438cf
  Karl Rupp authored Apr 18, 2015
  
  c78438cf
- SpGEMM: Now merging two rows from B before merging into scratchpad. · 278e37b3
  Karl Rupp authored Apr 18, 2015
  278e37b3
- SpGEMM: Using smaller scratchpad buffers to increase cache hit rates. · 8e4805b2
  Karl Rupp authored Apr 18, 2015
  8e4805b2
- AVX2: Allocating memory via aligned_alloc with 32-byte alignment. · b0607df1
  Karl Rupp authored Apr 18, 2015
  b0607df1
- SpGEMM: Added RMerge-like factorization if nonzeros in A are above 1024. · 764230c7
  Karl Rupp authored Apr 18, 2015
  764230c7
- SpGEMM: Added missing optimizations for single row merge in RMerge. · 616f1e1c
  Karl Rupp authored Apr 18, 2015
  616f1e1c
- SpGEMM: Fixed index bug in CUDA kernel for RMerge. · cb3a8546
  Karl Rupp authored Apr 18, 2015
  cb3a8546
Apr 17, 2015
- SpGEMM: Completed RMerge implementation. · a8ebd2f1
  Karl Rupp authored Apr 18, 2015
  
  a8ebd2f1
- SpGEMM: Added RMerge implementation. · 3943a3b6
  Karl Rupp authored Apr 17, 2015
  3943a3b6
- SpGEMM: Reduced memory footprint of CUDA implementation. · efcfd827
  Karl Rupp authored Apr 17, 2015
  efcfd827
- IO: Fixed warning in Matrix Market reader. · b6f77e61
  Karl Rupp authored Apr 17, 2015
  b6f77e61
- SpGEMM: Added profiling information to CUDA implementation, using __ldg() · 50668951
  Karl Rupp authored Apr 17, 2015
  50668951