Terry Cojean

I think one limit with this technic if I understand properly is that it requires two consistencies: 1. All executors will have a similar implementation so that they have the...

format!

Thanks for this, I think using google benchmark for these parts can make sense, since that's pretty much what it was designed for. I have a few remarks after a...

There are long lived (p)threads on every core. Tasks are just structures which have a code tied to it and data. The tasks are moved to the cores/pthreads by the...

I'm not sure what you mean, but the data can move between cores/threads depending on the scheduler. Whether there will be concurrent access depends on the code itself as well:...

As Tobias says, 1 should be relatively cheap to implement and manage, very much like some other mixins that we have. It makes code a bit less easy to understand,...

format!

format!

Thanks for the report! Sadly that part is one we did not test so well so far. We'll fix this.

> Should I just change the default for ginkgo for the moment while this branch is experimental, or do we want to keep this PR open to debug the rocprim...