Vulkan and OpenCL will merge into a single API

Re:Vulkan and OpenCL will merge into a single API by roel_v on Saturday May 20, 2017 @05:35PM (#14384470)

A bit more context after having re-read the OP: consider the following diagrams:

    - https://www.khronos.org/assets/uploads/apis/2016-spir-10.jpg
    - https://streamhpc.com/wp-content/uploads/2015/05/khronos-SPIR-V-flowchart-550x323.png

The core thing to understand is that (a bit simplified) both Vulkan and OpenCL (as of 2.2) run on SPIR-V IL. So in that context (with that common foundation for both API's in the near future - well, depending on your definition of 'near' I guess...), it makes sense that someone at Khronos might have said something about 'converging' - large parts of the ecosystem can be shared (compilers/transpilers, tools, ...), some parts of the API's might be 'aligned' more, ... But 'converging' is not 'OpenCL will be folded into the Vulkan API', which I think is an overhasty conclusion by the author of OP. If anything (at least, this was my impression from the conference), the big new things for OpenCL are on devices that are not GPU's (well, on FPGA's, to be more precise). This might require some specific support (for features) that don't have a place in Vulkan at all. That's not saying that the alluded to 'aligning' or 'converging' can't happen, but there is no One API To Rule Them All ('them' = graphics and compute).

So, to conclude: I see why the author would draw the conclusion he did, and if what he cites really are literal citations then maybe the Khronos communication could have been a bit more clear, but I don't think this conclusion is correct.

[ Parent | on HN ]

Re:Vulkan and OpenCL will merge into a single API by ScottMichaud on Sunday May 21, 2017 @12:20AM (#14385681)

Hello. I'm the author of the PC Perspective story that this post cites.

https://www.pcper.com/reviews/General-Tech/Breaking-OpenCL-M...

I am currently writing up another post, which is based on an interview I did with Tom Olson (Chairman of Vulkan) and Neil Trevett (President of Khronos Group) on Friday. While the implications of the announcement are smaller than I speculated (ex: the Khronos Group weren't concerned about Apple's ownership over OpenCL's trademarks and a few necessary patents) the announcement is correct.

Neil: "Congratulations. You win the prize for finding the extra press release hidden in the quote."

I'm still digesting the dense, ~40 minute discussion, but OpenCL's roadmap is to merge into Vulkan's. There's a bunch of reasons for it (ex: OpenCL pushes FP32 support onto hardware where it doesn't make sense, like DSPs and deep-learning applications) but I'll elaborate on that with the follow-up. Might take a few days though.

Also, here's Neil's presentation from IWOCL: https://www.khronos.org/developers/library/2017-iwocl

[ Parent | on HN ]

Re:Vulkan and OpenCL will merge into a single API by roel_v on Monday May 22, 2017 @03:58AM (#14391780)

Thanks for following up here, and sorry that my reply is late - too late to make any difference, probably.

Yes, I was at Neil's keynote that you link the slides for. I still do not agree with your conclusions, or maybe rather with how you frame them. OpenCL is not 'being merged into Vulkan'. Rather, 'the next version(s) of OpenCL will align more ('converge') with Vulkan'. For example, you claim 'The Khronos Group are converging OpenCL and Vulkan into a single API: Vulkan.'. I do not get that from either of the Khronos 'clarifications', nor from any of the presentations given last week, nor from any of the discussions I had with people from the various Khronos workgroups, including the OpenCL-related ones.

I've asked a few other people who were there how they perceived the message but haven't heard back yet. Maybe I'm just thick, wouldn't be the first time. But for now I do feel that you're spinning a more sensational story ('OpenCL will go away and be merged into Vulkan!' - on a meta-level, it's a sad reflection on my life that I consider that 'sensational'...) than what is actually going on.

[ Parent | on HN ]

Re:Vulkan and OpenCL will merge into a single API by ScottMichaud on Monday May 22, 2017 @03:57PM (#14396453)

I'm still writing up the interview that I did with them (Tom Olson and Neil Trevett) on Friday, after the initial post was published, but they do intend to merge OpenCL into Vulkan.

One thing they did want me to make clear, though, is that this is an OpenCL decision. Vulkan's roadmap isn't changing. While I'll elaborate in the follow-up post, it's to move OpenCL more low-level and abstract (ex: as I said above, FP32 makes no sense for many DSPs or deep-learning accelerators).

[ Parent | on HN ]

Re:Vulkan and OpenCL will merge into a single API by ScottMichaud on Wednesday May 24, 2017 @12:08PM (#14411069)

The interview is up:

https://www.pcper.com/reviews/Graphics-Cards/Follow-Neil-Tre...

[ Parent | on HN ]