Kévin Petit
Kévin Petit
We (Arm) return `CL_INVALID_VALUE` in that case. We'd support this becoming the specified behaviour.
Related to #10 and #11
Thanks for the initial review, that's very useful feedback! I don't think there are any irreconcilable differences so far. 1. I did notice this in the Intel extensions. I'm not...
I've added a list of outstanding issues to the PR description. We can link separate issues if some points require more discussion than reasonable to have here. I've also allowed...
I've just rebased the specification, added a dependency on https://github.com/KhronosGroup/OpenCL-Docs/pull/750 for per-format queries for max supported image dimensions and ticked the item in the list of outstanding issues. >I think...
I've rebased the specification draft and captured what I believe to be an exhaustive list of all the open topics in the PR description. We seem to be mostly aligned...
I've just bumped into this as well. I think we should make that change and also allow `CL_DEVICE_TYPE_CUSTOM` devices to be the default device in a platform. The specification currently...
Currently produces the following output: ``` Looking for orphan definitions in xml/cl.xml Enum 'CL_HALF_DIG' is defined but not used in any core version or extension! Enum 'CL_HALF_MANT_DIG' is defined but...
> Are you planning to fix most of these? Can do but I'll need a bit of help with vendor extensions. >All of the CL_HALF_XXX enums could be treated analogous...
https://github.com/KhronosGroup/OpenCL-Docs/pull/960 adds uses for all types and enums except vendor definitions and reservations.