Gordon Brown

We would be interested in having FFT support for Nvidia GPUs in oneMKL. Once this is supported in the open source oneMKL we could look into introducing support for this...

@ericlars now that https://github.com/oneapi-src/oneMKL/pull/215 has been merged the issues resulting from the multiple CUDA stream implementation in DPC++ should be resolved, would you be able to review this again?

Thanks for pointing this out, usually I present this one, I can see this being difficult to follow so I'll update it with comments.

I agree, I've been thinking this myself, we should perhaps have a hello world SYCL example, where we don't go into too much detail but describe the broad concepts.

This was the original exercise for demonstrating coalesced memory access, it's still an interesting exercises so might be worth adding it back in again, not sure where it would fit...

Yeah I think this ticket can be closed, I think it might be worth re-introducing the matrix transpose example, though we should maybe create a separate ticket for this,

So the idea here was after introducing the buffer and USM models to give some motivation for why you would choose one or the other, but yeah it does mention...

Hi @illuhad, that's a good idea, for the IWOCL tutorial we're using the `iwocl21` branch, on this branch we just have the original USM exercise as well as some additional...

That's great! Yeah that's a good point, these exercises originally came from a GPU programming tutorial so they were more GPU focused, I've put up a PR that switches to...

Yeah I think this makes sense.