[Bug c++/61075] parallel std::accumulate reduct type cannot be different than the iterated type

[Bug c++/61075] parallel std::accumulate reduct type cannot be different than the iterated type

[denes.matetelki at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61075

Denes Matetelki <denes.matetelki at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
               Host|                            |Linux  3.13.5-gentoo #10
                   |                            |SMP Fri Apr 25 16:12:35
                   |                            |CEST 2014 x86_64 Intel(R)
                   |                            |Xeon(R) CPU W3690 @ 3.47GHz
                   |                            |GenuineIntel GNU/Linux

--- Comment #1 from Denes Matetelki <denes.matetelki at gmail dot com> ---
#include <vector>
#include <numeric>
#include <iostream>

class Custom
{
public:
  Custom(int i) : m_i(i) {}
  int getI() const { return m_i; }

private:
  int m_i;
};

int main (int /*argc*/, char* /*argv*/[])
{
  std::vector<int> v(10, 1);
  const int v_sum = std::accumulate(v.begin(), v.end(), 0,
                                    [](int sum, int i) { return sum + i; } );


  std::vector<Custom> v2(10, Custom(1));
  const int v2_sum = std::accumulate(v2.begin(), v2.end(), 0,
                                    [](int sum, const Custom& c) { return sum +
c.getI(); } );

  return 0;
}

Compiling with parallel mode:

parallel_accumulate.cpp:33:93:   required from here
/usr/lib/gcc/x86_64-pc-linux-gnu/4.8.2/include/g++-v4/parallel/par_loop.h:102:15:
error: cannot convert
'std::__iterator_traits<__gnu_cxx::__normal_iterator<Custom*,
std::vector<Custom> >, true>::value_type {aka Custom}' to 'int' in
initialization
      __reduct = new _Result(__f(__o, __begin + __start));

[Bug libstdc++/61075] parallel std::accumulate reduct type cannot be different than the iterated type

[denes.matetelki at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61075

--- Comment #2 from Denes Matetelki <denes.matetelki at gmail dot com> ---
g++ -v

Using built-in specs.
COLLECT_GCC=/usr/x86_64-pc-linux-gnu/gcc-bin/4.8.2/g++
COLLECT_LTO_WRAPPER=/usr/libexec/gcc/x86_64-pc-linux-gnu/4.8.2/lto-wrapper
Target: x86_64-pc-linux-gnu
Configured with: /var/tmp/portage/sys-devel/gcc-4.8.2/work/gcc-4.8.2/configure
--host=x86_64-pc-linux-gnu --build=x86_64-pc-linux-gnu --prefix=/usr
--bindir=/usr/x86_64-pc-linux-gnu/gcc-bin/4.8.2
--includedir=/usr/lib/gcc/x86_64-pc-linux-gnu/4.8.2/include
--datadir=/usr/share/gcc-data/x86_64-pc-linux-gnu/4.8.2
--mandir=/usr/share/gcc-data/x86_64-pc-linux-gnu/4.8.2/man
--infodir=/usr/share/gcc-data/x86_64-pc-linux-gnu/4.8.2/info
--with-gxx-include-dir=/usr/lib/gcc/x86_64-pc-linux-gnu/4.8.2/include/g++-v4
--with-python-dir=/share/gcc-data/x86_64-pc-linux-gnu/4.8.2/python
--enable-languages=c,c++,fortran --enable-obsolete --enable-secureplt
--disable-werror --with-system-zlib --enable-nls --without-included-gettext
--enable-checking=release --with-bugurl=https://bugs.gentoo.org/
--with-pkgversion='Gentoo 4.8.2 p1.3r1, pie-0.5.8r1' --enable-libstdcxx-time
--enable-shared --enable-threads=posix --enable-__cxa_atexit
--enable-clocale=gnu --enable-multilib --with-multilib-list=m32,m64
--disable-altivec --disable-fixed-point --enable-targets=all --disable-libgcj
--enable-libgomp --disable-libmudflap --disable-libssp --enable-lto
--without-cloog
Thread model: posix
gcc version 4.8.2 (Gentoo 4.8.2 p1.3r1, pie-0.5.8r1)

[Bug libstdc++/61075] parallel std::accumulate reduct type cannot be different than the iterated type

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61075

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-05-06
     Ever confirmed|0                           |1

[Bug libstdc++/61075] parallel std::accumulate reduct type cannot be different than the iterated type

[denes.matetelki at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61075

--- Comment #4 from Denes Matetelki <denes.matetelki at gmail dot com> ---
Thank you for the reply, Jonathan.

I understand your reasoning and not sure if my desires has much impact in the
future of GCC.

I'm suprised that the same source code cannot be compiled with parallel mode.
It would be ugly to branch with #ifdef _GLIBCXX_PARALLEL. 
Also, I feel it should be allowed for the user to create a custom labda to add
up custom types, just like in the single threaded mode.
The std::experimental::reduce is experimental, how shall I solve the problem
now? The world accumulate shows the intention, shall I use my own
implementation of reduce till that time: a parallel for_each then a single
threaded for to sum up?

[Bug libstdc++/61075] parallel std::accumulate reduct type cannot be different than the iterated type

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61075

--- Comment #5 from Jonathan Wakely <redi at gcc dot gnu.org> ---
(In reply to Denes Matetelki from comment #4)
> I'm suprised that the same source code cannot be compiled with parallel
> mode. It would be ugly to branch with #ifdef _GLIBCXX_PARALLEL. 
> Also, I feel it should be allowed for the user to create a custom labda to
> add up custom types, just like in the single threaded mode.

It is allowed, you just need to meet the additional requirement that Custom is
convertible to int, for example add this member function:

  explicit operator int() const { return m_i; }


N.B. the parallel mode is unlikely to see any new work or improvements. Instead
we are more likely to focus on implementing the Paralellism TS (of which N3850
is the current draft).

[Bug libstdc++/61075] parallel std::accumulate reduct type cannot be different than the iterated type

[denes.matetelki at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61075

Denes Matetelki <denes.matetelki at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |INVALID

--- Comment #6 from Denes Matetelki <denes.matetelki at gmail dot com> ---
I see. Thank you for your help, 

I'll implement the convert operator.

[Bug libstdc++/61075] parallel std::accumulate reduct type cannot be different than the iterated type

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61075

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|INVALID                     |WONTFIX

--- Comment #7 from Jonathan Wakely <redi at gcc dot gnu.org> ---
I'm going to set this to WONTFIX instead.

It is a real bug in the parallel mode, but we're not going to fix it because
doing so isn't really possible, and the parallel mode is not being actively
developed.

[Bug libstdc++/61075] parallel std::accumulate reduct type cannot be different than the iterated type

[denes.matetelki at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61075

--- Comment #8 from Denes Matetelki <denes.matetelki at gmail dot com> ---
Just an observation:

The contained type also need to declare and define ctor(int).

Which can be tricky, if it is a template class and:

template <typename T>
class Custom
{
public:
  Custom(T t) : m_t(t) {}
  Custom(int i) : m_t() {}
  explicit operator int() const { return convertToInt(m_t); }

private:
  T m_t;
};

Custom<int> c; // ambigous ctor(int)