Jim Lewis

We need to consider the fp config flag [CL_FP_CORRECTLY_ROUNDED_DIVIDE_SQRT](https://registry.khronos.org/OpenCL/specs/3.0-unified/html/OpenCL_API.html#CL_FP_CORRECTLY_ROUNDED_DIVIDE_SQRT) as well. For this, I don't _think_ starting with a correctly rounded reciprocal or inverse sqrt is sufficient without some extra...

I'm expecting that some existing Vk implementations do already provide (or at least have access to native functionality to provide) correctly rounded Divide or Square Root implementations. The idea is...

I should clarify that most of the reductions look fine, for the tests that aren't specifically testing max alloc size functionality. The main one I'm concerned about is `test_api_min_max`, where...

@chemis01 do you mind if I split off the infrastructure required for the #1318 fix into a separate PR (with attribution of course)? Or you can do it if you'd...

> > @chemis01 do you mind if I split off the infrastructure required for the #1318 fix into a separate PR (with attribution of course)? Or you can do it...