Commits · 6d8feb56d9588f695830a3e49ab19586ef57a281 · Kaushik Kulkarni / loopy · GitLab

Mar 03, 2012
- Multple things, mainly related to prefetch. · 6d8feb56
  Andreas Klöckner authored Mar 03, 2012
  6d8feb56
Mar 02, 2012
- Tweaks to (help) the n-body test case. · 5def2580
  Andreas Klöckner authored Mar 02, 2012
  
  5def2580
- Add SEM test case from Reagan airport with Tim. · d68e69a8
  Andreas Klöckner authored Mar 02, 2012
  
  d68e69a8
- Add footprint generators for prefetch. · d20e0267
  Andreas Klöckner authored Mar 02, 2012
  
  d20e0267
Feb 21, 2012
- Remove TODO from FEM assembly ilp.seq example. · fd78a38d
  Andreas Klöckner authored Feb 21, 2012
  
  fd78a38d
- Make ILP go through regular loop emission, implement ilp.seq. · de71343d
  Andreas Klöckner authored Feb 21, 2012
  de71343d
Feb 20, 2012
- Before ILP refactor. · 04e4ee91
  Andreas Klöckner authored Feb 20, 2012
  
  04e4ee91
Feb 13, 2012
- Parse "N" as a shape/stride spec in kernel input. · b16f4dc8
  Andreas Klöckner authored Feb 12, 2012
  
  b16f4dc8
Feb 05, 2012
- Default sweep_dims kwarg on add_prefetch() to empty. · 8e4493cd
  Andreas Klöckner authored Feb 05, 2012
  
  8e4493cd
- Provide a better error message for inappropriate assignment lhs. · 3933b49a
  Andreas Klöckner authored Feb 05, 2012
  
  3933b49a
- Add some doc hints that I'm likely to forget. · cda499e4
  Andreas Klöckner authored Feb 05, 2012
  
  cda499e4
- Add parsing for "if()". · 59587042
  Andreas Klöckner authored Feb 05, 2012
  
  59587042
- "Properly" parse instructions with multiple equal signs. · ff0933a5
  Andreas Klöckner authored Feb 05, 2012
  
  ff0933a5
- Make switch to turn off complex-valued expression codegen. · ea359a6a
  Andreas Klöckner authored Feb 05, 2012
  
  ea359a6a
- Meeting notes w/ Tim at Reagan. · 8dd4ed28
  Andreas Klöckner authored Feb 05, 2012
  
  8dd4ed28
Jan 24, 2012
- Implement complex-valued arithmetic. · 9e30e55e
  Andreas Klöckner authored Jan 23, 2012
  
  9e30e55e
Jan 18, 2012
- Provide interactive assistance if the scheduler fails. · 72fdc4d2
  Andreas Klöckner authored Jan 18, 2012
  
  72fdc4d2
- Fix Fig 3.3 variant of FEM assembly. · 0902408b
  Andreas Klöckner authored Jan 18, 2012
  
  0902408b
- Expose reduction realization to the user. Add more notes. · 8ba555dc
  Andreas Klöckner authored Jan 18, 2012
  
  8ba555dc
Jan 17, 2012
- Fix FEM assembly code. · c8064886
  Andreas Klöckner authored Jan 17, 2012
  
  c8064886
- Fix prefill criterion. · 9e2640d2
  Andreas Klöckner authored Jan 17, 2012
  
  9e2640d2
Jan 16, 2012
- Test float4 functionality. · 26a50987
  Andreas Klöckner authored Jan 16, 2012
  
  26a50987
- Parse vectorized reductions. Spelling: 'sum_vec_float4' · 4b34ba94
  Andreas Klöckner authored Jan 16, 2012
  
  4b34ba94
- Hack auto test rig to support vector types, add flag about output filling. · 5e72d5e3
  Andreas Klöckner authored Jan 16, 2012
  
  5e72d5e3
- Use PyOpenCL to translate dtypes into C types. · 0327e979
  Andreas Klöckner authored Jan 16, 2012
  
  0327e979
Nov 20, 2011
- Merge branch 'master' of ssh://forge.tiker.net:2234/loopy · 04ff7222
  Tim Warburton authored Nov 20, 2011
  
  04ff7222
- Tim's SEM tweaks. · 0d8961db
  Tim Warburton authored Nov 20, 2011
  
  0d8961db
- First shot at instance-based subst rule precomputation. · bf38b67c
  Andreas Klöckner authored Nov 20, 2011
  
  bf38b67c
- Only include floor_div code if needed. · ac02f82d
  Andreas Klöckner authored Nov 20, 2011
  
  ac02f82d
Nov 12, 2011
- Make better message about device used for ref calculation. · 421c41d9
  Andreas Klöckner authored Nov 12, 2011
  
  421c41d9
- One more FEM tweak. · c7913fcf
  Andreas Klöckner authored Nov 12, 2011
  
  c7913fcf
- Fix up FEM quadrature example. · c05844ee
  Andreas Klöckner authored Nov 12, 2011
  
  c05844ee
- Many scattered fixes. Full-prefetch SEM with ILP generates code. · 6d6de311
  Andreas Klöckner authored Nov 12, 2011
  
  6d6de311
- Fix doc upload script. · 8fe68119
  Andreas Klöckner authored Nov 11, 2011
  
  8fe68119
- Switch ill-advised outer 'unroll' on split automatic axis back to sequential. · c700eb6d
  Andreas Klöckner authored Nov 11, 2011
  
  c700eb6d
- Fix exposure of with_annotation flag. · e2e48066
  Andreas Klöckner authored Nov 11, 2011
  
  e2e48066
- Upgrade precomputation to make storage dims and sweep dims independent, allow iname sweeps. · 847ffd67
  Andreas Klöckner authored Nov 11, 2011
  
  847ffd67
- Add a better output format for kernels. · 2f5da323
  Andreas Klöckner authored Nov 11, 2011
  
  2f5da323
Nov 11, 2011
- Add some documentation. Expose 'with_annotation' flag. · 3cd2819c
  Andreas Klöckner authored Nov 11, 2011
  
  3cd2819c
- Pick (more) sensible names for reduction inames. Remove non-ND FEM quadrature. · 9ad444e6
  Andreas Klöckner authored Nov 11, 2011
  
  9ad444e6