[PATCH 3/3] Try to perform OpenCL type initialization in C++

Neil Henning llvm at neil-henning.co.uk
Tue May 20 02:15:23 PDT 2014


Just a note: SYCL doesn't require OpenCL++ - it parses plain ol' C++ 
code and generates IR/binaries (SPIR or otherwise) to offload the 
workloads where possible.

My company (Codeplay) has an implementation of the provisional SYCL 
specification that does use some patches on tip Clang, but these do not 
modify any code in Clang that parses C++.

-Neil.

On 19/05/2014 16:17, David Tweed wrote:
> Hi,
>
> So in a personal communication, James has pointed out that there are
> some subtle differences between C and C++ specifications.
>
> What's the semantics for OpenCL++ to be? Is it basically "version of
> C++ plus whatever modifications langOpts.openCL adds to clang" (ie, in
> case of divergence between C and C++ there's no attempt to be
> consistent with OpenCL, but to pick whatever C++ produces)?
>
> Particularly since Khronos haven't yet decided upon the what official
> OpenCL C++ support will look like, that seems like a reasonable
> position but I guess it ought to be documented that that's what it's
> intended semantics are.
>
>
> On Mon, May 19, 2014 at 3:02 PM, Adam Strzelecki <ono at java.pl> wrote:
>>> Is there a spec for OpenCL++ that I've missed somewhere? I can't seem to find anything with a google.
>> There is SYCL: Abstraction layer for leveraging C++ and OpenCL
>> https://www.khronos.org/opencl/sycl
>>
>> I intentionally haven't mentioned SYCL, because these patches do not bring anything in regards SYCL API. The point of these patches was to be able to generate proper OpenCL SPIR code from C++ code, which basically makes compiling existing C++ programs much much easier.
>>
>> OpenCL++ is somehow transitional between SYCL which is C++ plus some objective API wrapping OpenCL calls, so you can think of it as:
>>
>>          OpenCL & C++ => OpenCL++
>>          OpenCL++ & SYCL API => SYCL Spec
>>
>> Also AMD's OpenCL SDK allows C++ usage in OpenCL, but that has nothing to do my patches.
>>
>> Regards,
>> --
>> Adam
>> _______________________________________________
>> cfe-commits mailing list
>> cfe-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
>
>




More information about the cfe-commits mailing list