Skip to content

Conversation

@MrBurmark
Copy link
Member

Change some parts of the Sycl resource to be more consistent with other resources.

These involve breaking changes and should not be merged until the next major release.

Now Sycl::get_queue returns a reference instead of a pointer
static sycl::queue& get_a_queue(const sycl::context* syclContext,
int num)
{
static constexpr int num_queues = 16;
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should we make something like this a configure option so it's easier to change if necessary for different architectures? Or, is it expected to always be 16?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

That's not a bad idea.

@MrBurmark
Copy link
Member Author

This adds a user settable default resource per thread, which may also be useful to have for the other GPU backends. It adds a bit more overhead as the global is not a thread_local instead.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants