Ole Schütt

Since #170 the [dashboard test](https://dashboard.cp2k.org/archive/aiida/index.html) has been failing even more. Do you know for when the next aiida-core release is planned?

> I have a [branch](https://github.com/abussy/cp2k/tree/dbt_offload) where I experimented with offloading DBCSR calls to DBM (see cp_dbcsr_multiplication.F). ... From my tests, this seems to be fairly affordable, but certainly not ideal....

I also came across this issue and it currently prevents us from using `-Werror` for our C code: https://github.com/cp2k/cp2k/blob/20eb57b76425229f786a46f502396a5328512b8a/cmake/CompilerConfiguration.cmake#L91-L95

While I believe that DeePMD in principle supports HIP, there are currently no [pre-build binaries](https://github.com/deepmodeling/deepmd-kit/releases) for it. Generally, our DeePMD integration could use [some help](https://github.com/cp2k/cp2k/pull/3301#issuecomment-2224002909) because their core developers are...

Is there anything we can do about this? Otherwise, we should just add another red entry to our [compiler page](https://www.cp2k.org/dev:compiler_support).

> Is it possible to deal with OOM situations gracefully? It depends ;-) Many methods have a hard requirement and there is no way to perform the calculation with less...

I'm glad ulimit did the trick. As for our production container, those have been moved to a separate repository: https://github.com/cp2k/cp2k-containers

I'm closing this issue as the patch has been picked up by Spack and the master branch was fixed via #3685.

I don't see how #3435 could cause a floating point exception because it only added some build options. Most likely the issue existed before, which means we'll need a lot...

Support for DeepMD 3.0.1 and PyTorch was added via #3893 after the 2025.1 release. So, give the git master branch a try.