On Thu, 27 Jun 2019 at 12:08, Bryce Adelstein Lelbach aka wash <brycelelbach@gmail.com> wrote:

Thanks Billy and Jonathan. This is really interesting data to have. I hadn't realized how big of an impact this was!

I'm very appreciative for the data here too; at the very least, it's super interesting! It also certainly underlines caution received in this thread about marrying the C++98 algorithms with the C++17 ones.

I'm sceptical about there being a usable, non-intrusive, and universal way to marry the two, but will give it some thought.

Perhaps the initial proposal should only cover the C++98 numeric algorithms to lay the groundwork, and have a sequel look at the C++17 ones?

Just covering the C++98 algorithms will be moderately sized.

On Thu, Jun 27, 2019 at 2:25 AM Jonathan Wakely <cxx@kayari.org> wrote:

On Tue, 25 Jun 2019 at 19:25, Billy O'Neal (VC LIBS) via Lib-Ext <lib-ext@lists.isocpp.org> wrote:

To put some numbers to that, on our (MSVC++’s) implementation:

Roughly comparable numbers using GCC (where the reduce_par algo uses Intel's PSTL and TBB), so I think it's safe to say that std::reduce is valuable even in the non-parallel case:

_______________________________________________
Parallel mailing list
Parallel@lists.isocpp.org
Subscription: http://lists.isocpp.org/mailman/listinfo.cgi/parallel
Link to this post: http://lists.isocpp.org/parallel/2019/06/2753.php

--
Bryce Adelstein Lelbach aka wash
CUDA Core C++ Libraries (Thrust, CUB, libcu++) Lead @ NVIDIA
ISO C++ Library Evolution Incubator Chair
ISO C++ Tooling Chair
CppCon and C++Now Program Chair
CUDA Convert and Reformed AVX Junkie

Sleep is for the weak
--
_______________________________________________
Lib-Ext mailing list
Lib-Ext@lists.isocpp.org
Subscription: http://lists.isocpp.org/mailman/listinfo.cgi/lib-ext
Link to this post: http://lists.isocpp.org/lib-ext/2019/06/12008.php