Create new sparse matrix for R factor in sparse least square solve

[andreasnoack]

instead of using a view to avoid slow fallback in back substitution.

Fixes #30288

[andreasnoack]

@nanosoldier runbenchmarks("sparse solves", vs = ":master")

[nanosoldier]

Your benchmark job has completed, but no benchmarks were actually executed. Perhaps your tag predicate contains misspelled tags? cc @ararslan

[andreasnoack]

What is the predicate to run the sparse solves in the sparse benchmark group?

[KristofferC]

I don't think Nanosoldier has been updated with your recent PR anyway.

[ararslan]

Oh, do the benchmarks need to be updated and retuned? If so I can do that now.

[andreasnoack]

@nanosoldier runbenchmarks("sparse", vs = ":master")

[ararslan]

You need backticks. @nanosoldier runbenchmarks("sparse", vs=":master")

[andreasnoack]

Oops. Didn't see your comments before commenting again. I thought that nano just pulled from BaseBenchmarks master. Would that be impossible? It would be nice if you could easily add a benchmark and make use of it when you realize something isn't covered.

@ararslan please just cancel the latest run if possible.

[KristofferC]

The benchmarks also go through a "tuning" process.

[andreasnoack]

You need backticks.

I make that copy-paste error almost every time. However, the benchmarks should only be run once the new ones I added are actually included in nano's portfolio.

[ararslan]

please just cancel the latest run if possible.

It's a PITA so I'll just let it run. I'm retuning the benchmarks with your updates now and I'll trigger a new job here once everything is updated. EDIT: FYI it is not a fast process.

[nanosoldier]

Your benchmark job has completed - possible performance regressions were detected. A full report can be found here. cc @ararslan

[ararslan]

Things are all updated now. @nanosoldier runbenchmarks("sparse", vs=":master")

[nanosoldier]

Your benchmark job has completed - possible performance regressions were detected. A full report can be found here. cc @ararslan

[andreasnoack]

@ararslan Thanks for updating nano. The constructor regressions must be spurious and the extra memory usage in the sparse solves is expected since we now create a new sparse array instead of using a view.