Arturo Vargas

Use std::numeric_limits<double>::max() instead of inf()

1

I noticed sometimes we run into floating point issues using inf(). Could we consider using max() instead?

Adds option to perform discrete upwind global matrix free -- uses element assembly for the discrete upwind terms and PA for face lumping. Notes: I refactored the residual distribution classes...

PR https://github.com/LLNL/RAJA/pull/1665 introduces a new format for RAJA examples. Advantages include reduced code duplication while maintaining the ability to run a variety of backends when available. We can draw inspiration...

In a new example it may be useful to add a global indexing example to point users to. There are some for launch but a focused example may be useful...

Improving RAJA integration with perf tools

1

PR: https://github.com/LLNL/RAJA/pull/1662 propagates the kernel naming API to all back-ends. For CUDA and HIP it is well defined as we can use the ROCTX and NVTX tools to name our...

**Is your feature request related to a problem? Please describe.** Constructing a permuted layout can be cumbersome. For example the following will create a permuted layout and identify unit stride....

Based on feedback from new RAJA users, it would be nice to have: - [x] a link to the raja-suite-tutorial from raja docs - [ ] a warning about how...

hip/cuda_global_thread_{x,y,z} not supported in RAJA::kernel

1

We get the following error when trying: ``` must know one of block_size or grid_size ```

# Summary This PR tries out an experimental perfectly nested loop abstraction.