Commits · cusolver · elpa / elpa

03 Dec, 2021 1 commit
- Minor correction · 380d9c98
  Soheil Soltani authored Dec 03, 2021
```
protect against CUDA kernel launch with zero blocks
```
  380d9c98
02 Dec, 2021 1 commit
- Minor correction · 1fad9cab
  Soheil Soltani authored Dec 02, 2021
```
correct the dimension of c_f pointer association argument
```
  1fad9cab
01 Dec, 2021 1 commit
- Implement Bcast of tmat2 via cuda_aware_mpi · c6a27cf9
  Soheil Soltani authored Dec 01, 2021
  
  c6a27cf9
16 Nov, 2021 4 commits
- Add lost code to check gpu_cholesky flag · 0895645c
  Andreas Marek authored Nov 16, 2021
  
  0895645c
- Merge branch 'master_pre_stage' into cusolver · 15ca2144
  Andreas Marek authored Nov 16, 2021
  
  15ca2144
- Do not use new gpu version on default · 6dbabc62
  Andreas Marek authored Nov 16, 2021
  
  6dbabc62
- Merge branch 'non_blocking_collectives' into master_pre_stage · d977966b
  Andreas Marek authored Nov 16, 2021
  
  d977966b
12 Nov, 2021 5 commits
- CUDA_AWARE_MPI for cholesky and invert_trm · aae38ff3
  Andreas Marek authored Nov 12, 2021
  
  aae38ff3
- Fix error introduces by previous merge · 0cdc0664
  Andreas Marek authored Nov 12, 2021
  
  0cdc0664
- More clear MPI code separation · 4570f64d
  Andreas Marek authored Nov 12, 2021
  
  4570f64d
- Remove debug prints · d7abf3ca
  Andreas Marek authored Nov 12, 2021
  
  d7abf3ca
- Bit of cleanup · bbee05dc
  Andreas Marek authored Nov 12, 2021
  
  bbee05dc
11 Nov, 2021 3 commits
- Gitlab CI: extensive tests · 9a8a13f9
  Andreas Marek authored Nov 11, 2021
  
  9a8a13f9
- PoC implementation of cholesky GPU version · 9b2989fd
  Andreas Marek authored Nov 11, 2021
  
  9b2989fd
- Move the memcpy outside the loop · 718acd73
  Soheil Soltani authored Nov 11, 2021
  
  718acd73
10 Nov, 2021 6 commits
- Optimzie GPU memory ops · 594efc58
  Soheil Soltani authored Nov 10, 2021
```
some redundant memory operations were removed
```
  594efc58
- Remove forgotten debug print · 7f6ee61f
  Andreas Marek authored Nov 10, 2021
  
  7f6ee61f
- cleanup of invert_trm · 8a6ac084
  Andreas Marek authored Nov 10, 2021
  
  8a6ac084
- module file for invert_trm · b3d93a90
  Andreas Marek authored Nov 10, 2021
  
  b3d93a90
- Move cholesky in new directory · 82e2d08c
  Andreas Marek authored Nov 10, 2021
  
  82e2d08c
- Add license headers to some files · 9fcde355
  Andreas Marek authored Nov 10, 2021
  
  9fcde355
09 Nov, 2021 6 commits
- Fix error after directory rename · 44e6ba0d
  Andreas Marek authored Nov 09, 2021
  
  44e6ba0d
- Fix error in cuda file · 0212e6af
  Andreas Marek authored Nov 09, 2021
  
  0212e6af
- Create subdir GPU and GPU/CUDA · 9a7bc93e
  Andreas Marek authored Nov 09, 2021
  
  9a7bc93e
- Rename folder elpa_invert_trm -> invert_trm · d618b941
  Andreas Marek authored Nov 09, 2021
  
  d618b941
- Fortran interfaced for CUDA Xpotrf · 92d2de6d
  Andreas Marek authored Nov 09, 2021
  
  92d2de6d
- start to add cusolver support for Xptrof · a64024d8
  Andreas Marek authored Nov 09, 2021
  
  a64024d8
08 Nov, 2021 1 commit
- Working GPU version of invert_trm · c22cab0d
  Andreas Marek authored Nov 08, 2021
  
  c22cab0d
03 Nov, 2021 4 commits
- [skip ci] Correct config option · 031b9e96
  Soheil Soltani authored Nov 03, 2021
```
correction is regarding the config option for setting the compute capability value for NVIDIA GPUs
```
  031b9e96
- correct argument order in cuda memcpy · 504ac860
  Andreas Marek authored Nov 03, 2021
  
  504ac860
- correct small typo in documentation · 57d288ae
  Andreas Marek authored Nov 03, 2021
  
  57d288ae
- Add precision macros for gpusolver_TRTRI · 2c81598c
  Andreas Marek authored Nov 03, 2021
  
  2c81598c
30 Oct, 2021 2 commits
- check error code of trtri · 8fd19bb6
  Andreas Marek authored Oct 30, 2021
  
  8fd19bb6
- Fix bug in resort_ev · 3c8e31cf
  Andreas Marek authored Oct 29, 2021
  
  3c8e31cf
29 Oct, 2021 1 commit
- Skeleton for GPU port of invert_trm · 83064ff3
  Andreas Marek authored Oct 29, 2021
  
  83064ff3
28 Oct, 2021 3 commits
- Add gpusolver layer · 2fe553df
  Andreas Marek authored Oct 28, 2021
  
  2fe553df
- Change preprocessor ifdefs · 05368314
  Andreas Marek authored Oct 28, 2021
  
  05368314
- optionally link against Nivdias cusolver library · 19d731b6
  Andreas Marek authored Oct 28, 2021
```
- Must be used with --with-cusolver
- needs cuda 11.4 or newer
- at the moment only for Xtrtri
```
  19d731b6
27 Oct, 2021 1 commit
- new autotuning with more options to tune · 50d58f47
  Andreas Marek authored Oct 27, 2021
  
  50d58f47
13 Oct, 2021 1 commit
- Remove forgotten print statements · 5915ea28
  Andreas Marek authored Oct 13, 2021
  
  5915ea28