-
@sbacchio @urbach @Finkenrath
I think I found the reason why the threading doesn't work right now... in DDalphaAMG interface:
```
#ifdef OMP
if(mg_omp_num_threads
-
The final residual of a WILSON inversion is wrong when switching from my test version for JSC to the current HEAD commits of the devel branches.
=== qphix ===
kostrzewa/qphix@ccc4542613ae0879703b6…
-
@sbacchio, @Finkenrath, @urbach : I would like to merge the production version of the ```DDalphaAMG_nd``` branch with my ```qphix_hmc_devel``` branch to then proceed to pull-request into etmc/master.…
-
@plabus @martin-ueding @urbach
This issue is to serve as a forum-like place, in addition to the [Kanban](https://github.com/orgs/etmc/projects/1), to track progress on the qphix interface. Mainly, t…
-
For testing purposes (such as for tuning DDalphaAMG or QPHIX), it is quite helpful to be able to disable the writing of propagators to avoid problems with using too much space or having to manually de…
-
Tasks:
- [x] Contact HLRS support about the issue.
- [x] Build libxml2 manually with Cray wrapped Intel compiler.
- [x] Build libxml2 with system GCC.
- [x] Use different version of Intel compil…
-
I've just observed a rather curious failure of the Richardson solver (using CG as the underlying algorithm with my MMdag addition).
For the double-single solver In the strong scaling limit with a l…
-
Since the MG mods have been merged into QPhiX, all solvers have been equipped with an additional flag to switch between absolute and relative residual, the latter being the default. In the QPhiX inter…
-
The clover_product stuff does not compile with half precision due to missing ```rep```'s in various places. I had already solved this in this in the tmLQCD version of these, see https://github.com/etm…
-
In some test programs, there is the following interesting code:
```c++
// Stupid compiler thing
#undef SEEK_SET
#undef SEEK_CUR
#undef SEEK_END
```
What is it about? Is perhaps the include …