| 8154be41 | 21-Jan-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
changed names of all GPU related functions, macros and variables that involve CUSP to use the string CUSP instead of the misleading CUDA
Hg-commit: 3264fa4049fd83eceb12c85feb980ef207360220 |
| a04f6461 | 19-Jan-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
updated users manual to do Schur complement preconditioner correct, finally I hope
Hg-commit: 3c53022524fbcd678904406f828dd1b3e6ed1544 |
| 435f959e | 17-Jan-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
Added MatView_SchurComplement() to view parts of matrix
Changed notation of PCView_Fieldsplit_Schur() to use A00, A01, A10, A11 notation instead of A, B, C, D
Hg-commit: 01877cdd11f782ec507ac3b082a8
Added MatView_SchurComplement() to view parts of matrix
Changed notation of PCView_Fieldsplit_Schur() to use A00, A01, A10, A11 notation instead of A, B, C, D
Hg-commit: 01877cdd11f782ec507ac3b082a8d6835bc2106d
show more ...
|
| c0adfefe | 17-Jan-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
hmm, forgot what I changed
Hg-commit: de55859da821036fe3032491b4caa4c8478f4edc |
| 6ce1633c | 16-Jan-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
added -pc_fieldsplit_stokes that automatically sets up Stokes solver with fieldsplit
Hg-commit: 93cf08337e0b89b643012c89924a66d5eb23e435 |
| e48cb861 | 15-Jan-2011 |
Hong Zhang <hzhang@mcs.anl.gov> |
bugfix of PCFactorSetUpMatSolverPackage()
add more test cases to ksp/examples/tutorials/ex52.c
Hg-commit: 699004b073abd3f70377bb965361283c0c3d9bf2 |
| 5a596e32 | 14-Jan-2011 |
Victor Minden <victorminden@gmail.com> |
commit after merge
Hg-commit: b9722e2c7aadbb358853c2635963a81e630e4ddb |
| f8260c8f | 14-Jan-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
changed PCFactorSetMatSolverPackage() to NOT require the Mat before being called, now it just keeps a copy of the package name and uses it later
added PCFactorSetUpMatSolverPackage() actually calls M
changed PCFactorSetMatSolverPackage() to NOT require the Mat before being called, now it just keeps a copy of the package name and uses it later
added PCFactorSetUpMatSolverPackage() actually calls MatGetFactor() so there now exists a factor matrix and PCFactorGetMatrix() can be called and then options set from the program.
Hg-commit: 7f7727b0030448abcd98a3d596d64ed83143f965
show more ...
|
| 8381b885 | 12-Jan-2011 |
Victor Minden <victorminden@gmail.com> |
added functions to set tolerance for bicgstabCUDA pc
Hg-commit: 1821b6f44bbb1a4bc34a59c118a3a116b28f8826 |
| 1a11c84d | 06-Jan-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
commit after merge
Hg-commit: 9eece74820683a3df266a7877d68c4eac77883cd |
| eb4736aa | 06-Jan-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
added ML_free() in MatWrapML_MPIAIJ() on gordering to prevent memory leak
Hg-commit: 5cfc9f44a5c1c84db14136f0df0edcf48046f93e |
| b709dedd | 06-Jan-2011 |
Victor Minden <victorminden@gmail.com> |
added framework for bicgstabcuda pc from cusp
removed old unneccesary code from sacuda pc
Hg-commit: f9d72302fefab6a4f284f29393c711349d2030ae |
| 0f11a792 | 05-Jan-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
updates to MATLAB interface
fix for Pastix fix
Hg-commit: 128f4b9629ba436a67a36180ed2f73c5394ad502 |
| 54379054 | 04-Jan-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
commit after merge
Hg-commit: e18f5cef7f51a3e72b47a5b55e4b408a7664a8e4 |
| 5e7ef714 | 04-Jan-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
added PCRedistributeGetKSP()
Hg-commit: 7979d255c25c706a54e3d24bfb09c9945e3d325f |
| 079416a2 | 04-Jan-2011 |
Jed Brown <jed@59A2.org> |
Fortran bindings for PCRedundantGetKSP() need to include "petscksp.h", why is bfort not smart enough to pick up direct includes?
Hg-commit: d2b300ce8a853af044faf90a0161016264d4dfff |
| 83ab6a24 | 04-Jan-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
changed PCREDUNDANT to actually use the KSP in the object! Now -redundant_ksp_type gmres for example works
Hg-commit: 9791ff14682958a0f4a5fdcb00bfb1afee0eb564 |
| e3c5b3ba | 26-Dec-2010 |
Barry Smith <bsmith@mcs.anl.gov> |
changed Matlab spelling to MATLAB some general MATLAB cleanup
Hg-commit: efcb24f191dfedc98b283d9e63e7859032e8ef35 |
| 41733a59 | 17-Dec-2010 |
Hong Zhang <hzhang@mcs.anl.gov> |
bugfix of PCFactorGetMatrix(): add missing 'else'
Hg-commit: 01727f8d47e03c53a36211e2c2da0cf459e0488f |
| d203961c | 15-Dec-2010 |
Barry Smith <bsmith@mcs.anl.gov> |
have PCFactorGetMatrix() generate error instead of silently fail if no factor exists.
Hg-commit: 90797865d34e4643f8a5e35d9d0e98bfbc548b28 |
| 9cfaa89b | 12-Dec-2010 |
Barry Smith <bsmith@mcs.anl.gov> |
added Matlab PetscOptionsSetValue() and
developer notes on PCSetInitialGuessNonzero()
Hg-commit: 5002a2a1c6352b05d17a90a5cb0e73986b8055fe |
| a3c4e3ec | 09-Dec-2010 |
Jed Brown <jed@59A2.org> |
Protect gesvd
Hg-commit: 905b7fba915ea1971461db7186b0d46544a81bc3 |
| 723b1ed7 | 08-Dec-2010 |
Barry Smith <bsmith@mcs.anl.gov> |
more fixes to configure for CUDA
Hg-commit: ad0d8d99353c8e3cb4de6cbac132d207ecbd9d0c |
| c3518bce | 06-Dec-2010 |
Dmitry Karpeev <karpeev@mcs.anl.gov> |
Added GASM example ex8g to testexamples_C.
Hg-commit: 44f4869c3a9d0ee3fcf89bcee9b9c5876d05c6e1 |
| eec7e3fa | 06-Dec-2010 |
Dmitry Karpeev <karpeev@mcs.anl.gov> |
rank-straddling GASM is trickier than I had thought. ex8g seems to work now.
There is an inherent serialization of subdomain solves imposed by the current PCGASMCreateSubdomains2D algorithm.
Subdoma
rank-straddling GASM is trickier than I had thought. ex8g seems to work now.
There is an inherent serialization of subdomain solves imposed by the current PCGASMCreateSubdomains2D algorithm.
Subdomains should be globally reordered, to remove serialization. This can be done, but I haven't done it yet.
In general, the burden is on the user to provide a good subdomain ordering.
Hg-commit: 51d041259f83c4fcb9b36890def11afa295ad5ca
show more ...
|