Commits · 9cf4f1f2f5e95f93fdaa378349bd31ebe7c98766 · Kaushik Kulkarni / viennacl-dev · GitLab

Nov 29, 2012
- * Renamed viennacl::linalg::single_threaded to viennacl::linalg::host_based · 9cf4f1f2
  Karl Rupp authored Nov 29, 2012
  9cf4f1f2
Nov 28, 2012
- * Added least-squares example · 11b7e84f
  Karl Rupp authored Nov 28, 2012
  11b7e84f
- * Fixed a minor bug in lu_factorize that showed up with the CUDA backend only · 8f971c47
  Karl Rupp authored Nov 28, 2012
  8f971c47
- * Reimplementation of LU factorization in viennacl/linalg/lu.hpp. Better... · e8a6e5b3
  Karl Rupp authored Nov 27, 2012
  e8a6e5b3
Nov 21, 2012
- * Reduced overhead for copying to/from ublas::compressed_matrix<> · 68ec5e72
  Karl Rupp authored Nov 21, 2012
  68ec5e72
- * Added CUDA examples/tutorials/tests to build system · ede1ed5c
  Karl Rupp authored Nov 21, 2012
  ede1ed5c
Nov 18, 2012
- * Added level scheduling to ILU0. Solver cycle times look good, but setup is still quite expensive. · 730b17be
  Karl Rupp authored Nov 17, 2012
  730b17be
Nov 16, 2012
- * Added support for row-/Jacobi-preconditioner with coordinate_matrix · 013e159c
  Karl Rupp authored Nov 16, 2012
  013e159c
- * Added missing kernel initialization call to row_info() for OpenCL · 88598ec9
  Karl Rupp authored Nov 15, 2012
  88598ec9
Nov 15, 2012
- * Added element-wise operations for vectors · 119785b2
  Karl Rupp authored Nov 15, 2012
  119785b2
- Simplified implementation of inner_prod(). Might yield better performance on AMD GPUs. · 6cca4eb9
  Karl Rupp authored Nov 15, 2012
  
  6cca4eb9
- * Typesafe multi-backend transfer now working, making implementations based... · 8213cb0c
  Karl Rupp authored Nov 14, 2012
  8213cb0c
Nov 14, 2012
- Implemented support for typesafe cross-domain transfer of memory buffers. More testing required. · 1c0f1224
  Karl Rupp authored Nov 14, 2012
  
  1c0f1224
- * Pimped incomplete Cholesky factorization. Speed now comparable to ILU. · 7510910c
  Karl Rupp authored Nov 13, 2012
  7510910c
Nov 13, 2012
- Added first implementation of incomplete Cholesky preconditioner. Requires improvements. · 5118cac9
  Karl Rupp authored Nov 13, 2012
  
  5118cac9
- Added missing diagonal_assign_cpu-kernels for matrices. Initializer types now all working. · a4d0d439
  Karl Rupp authored Nov 13, 2012
  
  a4d0d439
- * Added matrix initializers (work for CPU and OpenCL, CUDA-testing required) · 32af3402
  Karl Rupp authored Nov 12, 2012
  32af3402
Nov 12, 2012
- Transfer RAM<->OpenCL<->CUDA<->RAM now implemented. · 24b5e039
  Karl Rupp authored Nov 12, 2012
  
  24b5e039
- Pimped ILU(0,T)-preconditioners, up to one order of magnitude faster with new... · 1b975b43
  Karl Rupp authored Nov 11, 2012
  1b975b43
Nov 11, 2012
- Reimplementation of ILU0 for compressed_matrix. Using the new multi-backend,... · 3a53cf71
  Karl Rupp authored Nov 11, 2012
  3a53cf71
- Sparse triangular solvers for compressed_matrix now working (CPU, OpenCL, CUDA). · 113e0c7f
  Karl Rupp authored Nov 11, 2012
  
  113e0c7f
- Sparse triangular solver for compressed_matrix now working on CPU. Moved in... · 49010db8
  Karl Rupp authored Nov 10, 2012
  49010db8
Nov 09, 2012
- Working on sparse triangular solvers. Performance still not great, but soon sufficient for ILU. · 98af43d5
  Karl Rupp authored Nov 09, 2012
  
  98af43d5
Nov 08, 2012
- Some progress with triangular solver for compressed_matrix: unit-lower-solve... · 81969744
  Karl Rupp authored Nov 08, 2012
  81969744
- * Corrected deprecated VIENNACL_HAVE_XYZ in examples · 43434a19
  Karl Rupp authored Nov 07, 2012
  43434a19
Nov 07, 2012
- Added VIENNACL_WITH_OPENMP guard to existing OpenMP stuff · 3b0b1dbe
  Karl Rupp authored Nov 07, 2012
  
  3b0b1dbe
- * Unified preprocessor defines for external toolkit: VIENNACL_WITH_XYZ. Old... · b35f793b
  Karl Rupp authored Nov 07, 2012
  b35f793b
Nov 06, 2012
- Completed CUDA backend by adding direct triangular solvers and LU factorization. All tests pass. · 5476f4ba
  Karl Rupp authored Nov 06, 2012
  
  5476f4ba
Nov 05, 2012
- * Reduced generic vector kernel (av, avbv, avbv_v) startup by 10-20 percent by packing arguments · 575300ac
  Karl Rupp authored Nov 05, 2012
  575300ac
- Fixed minor bugs in vector CUDA kernels (accidental uses of 'float' instead of... · a844270f
  Karl Rupp authored Nov 05, 2012
  a844270f
Nov 04, 2012
- Sparse matrices now working with CUDA. · b759bf80
  Karl Rupp authored Nov 04, 2012
  
  b759bf80
Nov 03, 2012
- Vector operations and matrix operations now work for CUDA. No matrix-matrix... · 8c815ad1
  Karl Rupp authored Nov 03, 2012
  8c815ad1
- Added CUDA skeleton for vector operations. Needs testing. · 4b9eb57d
  Karl Rupp authored Nov 02, 2012
  
  4b9eb57d
Nov 02, 2012
- Added memory handling and scalar operations for CUDA. · 6bc0e300
  Karl Rupp authored Nov 02, 2012
  
  6bc0e300
Nov 01, 2012
- Compilation with nvcc now works. Fixed all warnings and errors reported by nvcc. · d036b567
  Karl Rupp authored Nov 01, 2012
  
  d036b567
- Refurbished build system, now deals correctly with OpenCL-enabled and -disabled environments. · 9611f739
  Karl Rupp authored Nov 01, 2012
  
  9611f739
- * Added VIENNACL_HAVE_OPENCL flag for including OpenCL specific stuff. · 8c84096f
  Karl Rupp authored Oct 31, 2012
  8c84096f
Oct 30, 2012
- Cleaned up norm-kernels. Now using a single kernel for norm_1, norm_2 and... · 4baa4d13
  Karl Rupp authored Oct 30, 2012
  4baa4d13
- Purged old/unused kernels in auxiliary folder · 0ed4f0f7
  Karl Rupp authored Oct 30, 2012
  
  0ed4f0f7
- Fixed remaining issues with matrix(-range/slice) interactions. Removed... · 648d3c50
  Karl Rupp authored Oct 30, 2012
  648d3c50