| e4af1fa5 | 20-Sep-2022 |
James Wright <james@jameswright.xyz> |
ci: Add color for diffs (#1067) |
| 9538871e | 17-Sep-2022 |
James Wright <james@jameswright.xyz> |
fluids: Fix Ek_tot underflow bug (#1069) |
| 175f00a6 | 14-Sep-2022 |
James Wright <james@jameswright.xyz> |
fluids: Refactor STG wall distance variable
- Needless shortening. - Confused me when debugging even though I wrote the code myself... |
| d97dc904 | 14-Sep-2022 |
James Wright <james@jameswright.xyz> |
fluids: Fix STG Ek_tot calculation bug
When the spectrum is small enough, Ek_tot can be zero, causing a divide-by-zero NaN, which then corrupts the calculation of the velocity fluctuations later dow
fluids: Fix STG Ek_tot calculation bug
When the spectrum is small enough, Ek_tot can be zero, causing a divide-by-zero NaN, which then corrupts the calculation of the velocity fluctuations later down the line.
Originally found during OLCF Crusher Hackathon 4
show more ...
|
| 417419bc | 15-Sep-2022 |
Jed Brown <jed@jedbrown.org> |
Merge pull request #1040 from CEED/jed/emscripten-build
emscripten build |
| db7c0df8 | 12-Sep-2022 |
Jeremy L Thompson <jeremy@jeremylt.org> |
ci - add color for diffs |
| e3e35afd | 12-Aug-2022 |
Jed Brown <jed@jedbrown.org> |
Makefile: add custom executable suffixes for emscripten
emcc looks for the output suffix .wasm when linking executables.
This now builds cleanly:
$ emmake make STATIC=1 OPT=-O build/ex2-surface.wa
Makefile: add custom executable suffixes for emscripten
emcc looks for the output suffix .wasm when linking executables.
This now builds cleanly:
$ emmake make STATIC=1 OPT=-O build/ex2-surface.wasm [...] $ wasmer run build/ex2-surface.wasm -- -s 32767 [...] Exact mesh surface area : 6 Computed mesh surface area : 5.9548265118955 Surface area error : -0.04517348810445
Increasing the size causes calloc to fail, likely to do some default memory limits that I haven't worked out yet.
Emscripten docs:
https://emscripten.org/docs/compiling/Building-Projects.html#building-projects
show more ...
|
| dacf7c88 | 12-Aug-2022 |
Jed Brown <jed@jedbrown.org> |
Makefile: skip -Wl,-rpath when using STATIC=1 |
| a0b0bf66 | 12-Sep-2022 |
Jeremy L Thompson <jeremy@jeremylt.org> |
Merge pull request #1066 from CEED/jeremy/lazy-qf
Lazy access of qf kernel name and path |
| ca5eadf8 | 12-Sep-2022 |
Jeremy L Thompson <jeremy@jeremylt.org> |
qf - lazy access of qf kernel name and path |
| dc64899e | 06-Sep-2022 |
Yohann <dudouit1@llnl.gov> |
Change the initialization logic for `useCollograd`. (#1021)
* Change the initialization logic for `useCollograd`.
* Guard useCollograd for 3D only.
* Propagate `useCollograd` change to `hip-ge
Change the initialization logic for `useCollograd`. (#1021)
* Change the initialization logic for `useCollograd`.
* Guard useCollograd for 3D only.
* Propagate `useCollograd` change to `hip-gen`.
* Update backends/hip-gen/ceed-hip-gen-operator-build.cpp
* Propagate changes to `hig-gen`.
* Revert redimensioning of `r_tt`.
Co-authored-by: Jed Brown <jed@jedbrown.org>
show more ...
|
| dbac3b9f | 06-Sep-2022 |
Natalie Beams <246972+nbeams@users.noreply.github.com> |
Fix t568 bug (#1065)
* Add multi-component apply QFunction to t568
* junit.py: Add t568 back to GPU backend testing |
| 4a64a794 | 06-Sep-2022 |
Jed Brown <jed@jedbrown.org> |
Merge pull request #1061 from CEED/jed/noether-cuda
ci: use noether for CUDA |
| 92cdb9d4 | 06-Sep-2022 |
Jed Brown <jed@jedbrown.org> |
ci: patch occa-1.1.0 for spurious sys/sysctl.h |
| 963d8636 | 06-Sep-2022 |
Jed Brown <jed@jedbrown.org> |
t568-operator: fix input order and make test handle nan; skip on broken backends
if (nan < constant) is always false. This test had a memory error causing it to always compute nan in both the libcee
t568-operator: fix input order and make test handle nan; skip on broken backends
if (nan < constant) is always false. This test had a memory error causing it to always compute nan in both the libceed and reference values. This conditional ensures that output will print in such cases.
This test is BROKEN on /gpu/cuda/gen and /gpu/hip/*. This will need to be addressed in a future issue.
Reported-by: Natalie Beams <nbeams@icl.utk.edu>
show more ...
|
| 68c150f8 | 02-Sep-2022 |
Jed Brown <jed@jedbrown.org> |
ci: use Noether for CUDA, handle CUDA on Debian CUDA_DIR=/usr
fix XSMM description |
| 5dfaedb8 | 02-Sep-2022 |
Jed Brown <jed@jedbrown.org> |
examples/petsc: fix missing PetscFunctionBegin |
| 07d14e58 | 05-Sep-2022 |
Leila Ghaffari <49916147+LeilaGhaffari@users.noreply.github.com> |
Fluids - Blasius: Some Minor Modifications (#1063)
* Fluids - blasius: default T_wall = T_inf = 288
* Fluids - blasius: break when Chebyshev solve diverges
* Fluids - blasius: add prefix for C
Fluids - Blasius: Some Minor Modifications (#1063)
* Fluids - blasius: default T_wall = T_inf = 288
* Fluids - blasius: break when Chebyshev solve diverges
* Fluids - blasius: add prefix for Chebyshev snes solve
* Fluids - blasius: N_Chebyshev -> n_chebyshev
* Update examples/fluids/tests-output/blasius_test.yaml
Co-authored-by: Jed Brown <jed@jedbrown.org>
* Fluids - blasius: Modify the parameters in the regression test
* examples/fluids: Blasius solution is variable density
We use the ideal gas relation that density ~ 1/enthalpy when held at
constant pressure.
* examples/fluids: Blasius slower growth factor
* examples/fluids: fix Chebyshev on GPU
and close memory leak
Co-authored-by: Jed Brown <jed@jedbrown.org>
show more ...
|
| bb4ed8c7 | 02-Sep-2022 |
Jeremy L Thompson <jeremy@jeremylt.org> |
Merge pull request #1058 from CEED/jeremy/noether-ci
LIBXSMM CI Update |
| 1a48f341 | 02-Sep-2022 |
Jeremy L Thompson <jeremy@jeremylt.org> |
ci - newer xsmm for ci |
| 2dc3fb5f | 31-Aug-2022 |
abdelfattah83 <36712794+abdelfattah83@users.noreply.github.com> |
Icl/magma ntgemm (#1060)
* tuning data and driver for the non-tensor gemm
* header
* update magma non-tensor sgemm/dgemm to use the gemm selector
* add cpp files for the magma backend
*
Icl/magma ntgemm (#1060)
* tuning data and driver for the non-tensor gemm
* header
* update magma non-tensor sgemm/dgemm to use the gemm selector
* add cpp files for the magma backend
* minor fix
* define CEED_INTERN for every function instead of a block definition
* include tuning data for CUDA or HIP only
* recent tuning data for a100 and mi250x
* style
* remove unused declarations
* expand tuning data for v100 and mi100
* switch to std array instead of std vector for individual records
* choose between gfx90a and gfx908 for HIP
* bug fix: choose between magma and vendor blas in non-batch mode
* style
show more ...
|
| 01005eab | 30-Aug-2022 |
Jeremy L Thompson <jeremy@jeremylt.org> |
Merge pull request #1053 from CEED/natalie/fix-magma-jit-mem
Fix small memory leaks in JIT source code management |
| 11a69ce0 | 30-Aug-2022 |
James Wright <james@jameswright.xyz> |
fluids: Fix time label for strong STG #1057 |
| 6739d7b6 | 30-Aug-2022 |
Jeremy L Thompson <jeremy@jeremylt.org> |
Merge pull request #1059 from CEED/leila/fix-blasius-test
Fluids - Fix blasius test |
| 2561c73a | 30-Aug-2022 |
Leila Ghaffari <Leila.Ghaffari@colorado.edu> |
Fluids - fix blasius test |