Commits · 3246d0d6e9879223bd0aaf7c90df02313c891d5d · Kaushik Kulkarni / viennacl-dev · GitLab

Apr 08, 2015
- SpGEMM: Added CUDA first implementation. Needs further tweaks. · 3246d0d6
  Karl Rupp authored Apr 08, 2015
  3246d0d6
Apr 07, 2015
- SpGEMM: Rearranged code in host backend for better navigation. · d7846a11
  Karl Rupp authored Apr 07, 2015
  d7846a11
- SpGEMM: Now using an adaptive algorithm (vector vs. list) · a94082b9
  Karl Rupp authored Apr 07, 2015
  a94082b9
Apr 06, 2015
- compressed_matrix: Merging of two rows in a list completed. · eacd3d98
  Karl Rupp authored Apr 06, 2015
  eacd3d98
Apr 02, 2015
- Optimization of SpGEMM to load and merge multiple rows from B. · 34d878d6
  Karl Rupp authored Apr 02, 2015
  34d878d6
- SpGEMM: Attempt to introduce a list for faster row merging. · 90c8ee9b
  Karl Rupp authored Apr 02, 2015
  90c8ee9b
- compressed_matrix: Removed use of an auxiliary array in SpGEMM. · 09476768
  Karl Rupp authored Apr 02, 2015
  09476768
Apr 01, 2015
- compressed_matrix: Optimizing host-based sparse matrix-matrix product. · 87edf140
  Karl Rupp authored Apr 01, 2015
  87edf140
Mar 08, 2015
- compressed_matrix: Implementation of SpGEMM for OpenCL now working for general matrices. · a0227178
  Karl Rupp authored Mar 08, 2015
  a0227178
- compressed_matrix: First implementation of SpGEMM for OpenCL. · b601a35f
  Karl Rupp authored Mar 08, 2015
  b601a35f
Mar 07, 2015
- compressed_matrix: Enabled C = prod(A, B) and fixed conversion warnings. · 5e25cdb5
  Karl Rupp authored Mar 07, 2015
  5e25cdb5
Mar 03, 2015
- compressed_matrix: Added first host-based prototype for SpGEMM. · 75695f04
  Karl Rupp authored Mar 03, 2015
  75695f04
Jan 09, 2015
- Builtin-DB : Bugfix in CPU fallback · 64958965
  Philippe Tillet authored Jan 09, 2015
  
  64958965
Dec 31, 2014
- Built-in DB: More conservative CPU Fallback · ecf6e1c4
  Philippe Tillet authored Dec 31, 2014
  
  ecf6e1c4
Dec 11, 2014
- Updated version.hpp to 1.6.2 · 38b6b067
  Karl Rupp authored Dec 11, 2014
  
  38b6b067
- Changelog: Last updates for 1.6.2 release. · 6a26de59
  Karl Rupp authored Dec 11, 2014
  
  6a26de59
- OpenMP: Ensured compatibility with OpenMP 2.0 (only signed integers as loop vars) · 605a1a90
  Karl Rupp authored Dec 11, 2014
  605a1a90
- Doxygen: Added reference to CSR-adaptive in manual. · 548d8b14
  Karl Rupp authored Dec 11, 2014
  
  548d8b14
- hyb_matrix: Using default workgroup sizes. · 9ce4b367
  Karl Rupp authored Dec 11, 2014
  9ce4b367
- coordinate_matrix: Fixed bug in row_info() kernels for OpenCL and CUDA. · 41189e24
  Karl Rupp authored Dec 11, 2014
  41189e24
- coordinate_matrix: Changed local thread size to 128 for better compatibility. · e6faa56e
  Karl Rupp authored Dec 11, 2014
  e6faa56e
- AMG: Fixed possibly incorrect matrix sizes of GPU-based operators. · 731a7810
  Karl Rupp authored Dec 11, 2014
  731a7810
Dec 10, 2014
- Updated version to 1.6.2 · 12c5ed2c
  Karl Rupp authored Dec 10, 2014
  
  12c5ed2c
- Device database: Removed unused member in template class. · 54450e8b
  Karl Rupp authored Dec 10, 2014
  54450e8b
- compressed_compressed_matrix: Fixed wrong buffer size in clear() · 384185b0
  Karl Rupp authored Dec 10, 2014
  384185b0
- GCC: Fixed all conversion warnings. · 3881ea8b
  Karl Rupp authored Dec 10, 2014
  3881ea8b
Dec 09, 2014
- GCC, Clang: Fixed compiler warnings · 390c47c1
  Karl Rupp authored Dec 09, 2014
  
  390c47c1
- Solver Bench: Added sliced_ell_matrix and added pipelined runs. · ecb1088a
  Karl Rupp authored Dec 09, 2014
  
  ecb1088a
- FFT: Fixed flag passed to CUDA kernel. · c09831bb
  Karl Rupp authored Dec 09, 2014
  c09831bb
- OpenMP: Fixed unspecified behavior for operations using reductions. · 71e46368
  Karl Rupp authored Dec 09, 2014
  71e46368
- OpenMP: Removed use of private and shared clauses. · 0a8e2999
  Karl Rupp authored Dec 09, 2014
  0a8e2999
- Iterative: Moved local array declaration out of CSR-adaptive kernel. · 40686dbf
  Karl Rupp authored Dec 09, 2014
  40686dbf
- Iterative: Fixed overloads for pipelined iterative solvers. · b825e5d2
  Karl Rupp authored Dec 09, 2014
  b825e5d2
Dec 06, 2014
- Direct solve bench: Removed accidental uBLAS dependency. · 8682205a
  Karl Rupp authored Dec 06, 2014
  
  8682205a
- CUDA: Fixing uses of 'uint' and performance warnings. · 4ebc5c60
  Karl Rupp authored Dec 06, 2014
  
  4ebc5c60
- CUDA: Fixed complaints about destructor in Visual Studio. · 0e2a3758
  Karl Rupp authored Dec 06, 2014
  0e2a3758
Dec 05, 2014
- vector_iterator: Fixed internal handling of smart-pointer. · 55b05254
  Karl Rupp authored Dec 05, 2014
  55b05254
Dec 04, 2014
- GMRES: Reverted to previous kernel in first pipelined stage for non-NVIDIA GPUs. · a4b7354a
  Karl Rupp authored Dec 04, 2014
  a4b7354a
- GMRES: Improved kernel first first stage of pipelined orthogonalization. · 4381e000
  Karl Rupp authored Dec 04, 2014
  4381e000
- compressed_matrix: Fixed missing context switch for CSR-adaptive metainfo. · 9d8bae24
  Karl Rupp authored Dec 04, 2014
  9d8bae24