Conversation
|
Thanks so much, everyone. Regarding function naming, I would strongly prefer removing the z prefix from all function names. In the PLASMA package, the z prefix was conventionally used to denote double complex data types, and the remaining precisions were generated via Python scripts. However, in Julia, we rely on type inference and multiple dispatch, so there’s no need to encode the data type in the function name. Therefore, I recommend omitting the z prefix in NextLA for consistency with Julia’s design philosophy. I’ll make sure to test all functions and ensure they pass the test suite after the renaming. |
|
@fda-tome Please expand the testing matrix sizes and remove z prefix from all function names. |
|
@Rabab53 thank you for point out the mistakes, I corrected the names and made the sizes bigger |
This PR adds many triangular on top of square (TS), triangular on top of triangular (TT), and some more basic QR kernels with proper testing, this is heavily based on the work done by @Azura-yuwi and @AkhilAkkapelli.