Thomas Grützmacher

Current NOT_IMPLEMENTED kernels

6

We should discuss what kernels we should implement before the release. Since I lost the overview, I looked in the code and searched for kernels which are marked as `GKO_NOT_IMPLEMENTED`....

Make the `Csr::automatical` strategy work with `gko::Executor`

3

Currently, we need either a `HipExecutor` or a `CudaExecutor` to create the `Csr::automatical` strategy. It would be a lot easier to have a constructor that just takes a `gko::Executor`, so...

Improving Csr::strategy_type

2

Currently, all strategies for CSR are tailored for CUDA, which is fine for the most part. However, also the names are CUDA specific (see `Csr::cusparse`), which should be changed in...

This issue is supposed to centralize all feature requests and list what features are currently worked upon. __If you have a request, please comment below with the use-case you have...

(CUDA) kernels inside namespace or suffix `_kernel`

8

In most cases, we append the suffix `_kernel` to a function to identify it as a function which is not executed on the host machine (currently only CUDA code). However,...

Add template-type functions to combine and merge type lists. As a result, the GINKGO_DPCPP_SINGLE_MODE only needs to be present once. Additionally, change the typen name ValueAndIndexType to ComplexAndPODTypes (because gko::size_type...