Piotr Balcer
Piotr Balcer
> > I would expect to remove the legacy adapter after a single release of them coexisting. We wouldn't want to implement features in both for very long. > >...
e2e tests show *many* regressions. edit: the fact that all UR tests pass, while a lot of SYCL tests are failing, indiactes to me that we have a pretty big...
I need to take a look at #1063, but there isn't really any issue of just using the version variable as a sort of size (like in your option 3)....
> the "version" here could be an int or size of structure, and we are trying to figure out here which option is the best I don't think it matters....
Copying structs like this will never just work. Not with size or version. It needs to be updated after such a copy. The answer here is documentation. > Size is...
> We need size - user might get osprovider ops, duplicate them, and then i.e change one function to his (i.e to implement some extra logging). to duplicate struct user...
100% agreed with @vinser52
> Failed Tests (1): SYCL :: Regression/queue_submitted_kernels_oom.cpp Eh, I tried adding margin to the memory use check to make it works across all adapters, but not sure if it's possible....
>As a final solution, should I copy this header and paste it into the L0 adapters subdirectory? Yes, I think we should do that for all experimental level-zero APIs we...
> `urEnqueueGraph` tests are failing with `Segmentation fault from GPU at 0x80d5ffa50000, ctx_id: 1 (CCS) type: 1 (WriteAccessViolation), level: 0 (PTE), access: 1 (Write), banned: 1, aborting`. Those pass for...