This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: [RFC] Proposal to contribute Intel’s implementation of C++17 parallel algorithms
- From: Jonathan Wakely <jwakely at redhat dot com>
- To: Alexey Kukanov <Alexey dot Kukanov at intel dot com>
- Cc: libstdc++ at gcc dot gnu dot org, pekka jaaskelainen <pekka dot jaaskelainen at parmance dot com>, Thomas Rodgers <trodgers at redhat dot com>
- Date: Tue, 6 Mar 2018 16:46:59 -0500 (EST)
- Subject: Re: [RFC] Proposal to contribute Intel’s implementation of C++17 parallel algorithms
- Authentication-results: sourceware.org; auth=none
- References: <AB8EA9DFCD18FB439229633DA9B1F7A97B7C7191@IRSMSX104.ger.corp.intel.com> <20171129140354.GM31922@redhat.com> <AB8EA9DFCD18FB439229633DA9B1F7A97B7C749C@IRSMSX104.ger.corp.intel.com>
On 29/11/17 16:47 +0000, Kukanov, Alexey wrote:
>Thank you Jonathan for prompt response.
>
>Yes we plan to maintain Parallel STL for both C++ standard libraries as well as provide it with Intel Compiler and TBB (and on GitHub) for those who use older versions of the standard. Maintaining a single source would be much easier for us, so I very appreciate your willingness to be flexible.
>
>> The only other effort I know of to implement this for libstdc++ is by Pekka
>> Jääskeläinen and relies on GCC's offloading support
>
>I think it would be ideal if GCC offloading could be used as a backend for the implementation. We can adjust the backend API if that's needed to support different "executors".
>For start, I will study Pekka's presentation.
>
>> It's possible this is too late to be added to GCC 8 (as our dev phase for GCC 8
>> just finished). If it's entirely new C++17 code that doesn't affect the default
>> C++14 mode we might be able to bend the rules a bit. We'd need to move quickly
>> if we're going to do that though.
>
>When I analyzed which GCC version we can intercepted, I did not even hope for GCC 8 :)
>But if we could do that with the current implementation, that would be just great. I hope you or somebody else can look at our code, and decide if adding it to GCC 8 seems possible, and if so - what issues we should address.
Hi Alexey,
As you've probably guessed, we aren't going to include the Parallel
STL in GCC 8 now :-)
But we haven't forgotten about it, and I've CC'd Thomas Rodgers, who
is going to be doing most of the work to integrate it with Libstdc++.
He's already been experimenting with the code, and we've started the
necessary conversations to get approval for accepting the code into
GCC with its current licence.
I imagine you'll be hearing more from Thomas in the near future, as
he figures out what needs to be done to get the two code bases working
together.
Thanks again for the contribution!