Commit 305b96de authored by Martin Reinecke's avatar Martin Reinecke

add missing file

parent 15b62dd9
* This file is part of DUCC.
* DUCC is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
* DUCC is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* GNU General Public License for more details.
* You should have received a copy of the GNU General Public License
* along with DUCC; if not, write to the Free Software
* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
* DUCC is being developed at the Max-Planck-Institut fuer Astrophysik
* Copyright (C) 2020 Max-Planck-Society
* Author: Martin Reinecke
#include <algorithm>
#include "mr_util/infra/error_handling.h"
#include "mr_util/infra/mav.h"
namespace mr {
namespace detail_transpose {
using namespace std;
using shape_t=fmav_info::shape_t;
using stride_t=fmav_info::stride_t;
template<typename T1, typename T2> void rearrange(T1 &v, const T2 &idx)
T1 tmp(v);
for (size_t i=0; i<idx.size(); ++i)
v[i] = tmp[idx[i]];
auto prep(const fmav_info &in, const fmav_info &out)
MR_assert(in.shape()==out.shape(), "shape mismatch");
shape_t shp;
stride_t si, so;
for (size_t i=0; i<in.ndim(); ++i)
if (in.shape(i)!=1) // remove axes of length 1
// sort dimensions in order of descending output stride
vector<size_t> idx(shp.size());
iota(idx.begin(), idx.end(), 0);
sort(idx.begin(), idx.end(),
[&so](size_t i1, size_t i2) {return so[i1] > so[i2];});
rearrange(shp, idx);
rearrange(si, idx);
rearrange(so, idx);
// try merging dimensions
// [...]
if (shp.size()>1)
// move axis with smallest remaining input stride to second-to-last place
auto iminstr = min_element(si.begin(), si.end()-1) - si.begin();
size_t s2l = si.size()-2;
swap (shp[iminstr], shp[s2l]);
swap (si[iminstr], si[s2l]);
swap (so[iminstr], so[s2l]);
return make_tuple(shp, si, so);
template<typename T, typename Func> void sthelper1(const T *in, T *out,
size_t s0, ptrdiff_t sti0, ptrdiff_t sto0, Func func)
for (size_t i=0; i<s0; ++i, in+=sti0, out+=sto0)
func(*in, *out);
bool critical(ptrdiff_t s)
s = (s>=0) ? s : -s;
return (s>4096) && ((s&(s-1))==0);
template<typename T, typename Func> void sthelper2(const T *in, T *out,
size_t s0, size_t s1, ptrdiff_t sti0, ptrdiff_t sti1, ptrdiff_t sto0,
ptrdiff_t sto1, Func func)
if ((sti0<=sti1) && (sto0<=sto1)) // no need to block
for (size_t i=0; i<s1; ++i, in+=sti1, out+=sto1)
auto pi0=in;
auto po0=out;
for (size_t j=0; j<s0; ++j, pi0+=sti0, po0+=sto0)
func(*pi0, *po0);
if ((sti0>=sti1) && (sto0>=sto1)) // no need to block
for (size_t i=0; i<s0; ++i, in+=sti0, out+=sto0)
auto pi1=in;
auto po1=out;
for (size_t j=0; j<s1; ++j, pi1+=sti1, po1+=sto1)
func(*pi1, *po1);
// OK, we have to do a real transpose
// select blockig sizes depending on critical strides
bool crit0 = critical(sizeof(T)*sti0) || critical(sizeof(T)*sto0);
bool crit1 = critical(sizeof(T)*sti1) || critical(sizeof(T)*sto1);
size_t bs0 = crit0 ? 8 : 8;
size_t bs1 = crit1 ? 8 : 8;
// make sure that the smallest absolute stride goes in the innermost loop
if (min(abs(sti0),abs(sto0))<min(abs(sti1),abs(sto1)))
swap(bs0, bs1);
for (size_t ii0=0; ii0<s0; ii0+=bs0)
size_t ii0e = min(s0, ii0+bs0);
for (size_t ii1=0; ii1<s1; ii1+=bs1)
size_t ii1e = min(s1, ii1+bs1);
for (size_t i0=ii0; i0<ii0e; ++i0)
for (size_t i1=ii1; i1<ii1e; ++i1)
func(in[i0*sti0+i1*sti1], out[i0*sto0+i1*sto1]);
template<typename T, typename Func> void iter(const fmav<T> &in,
fmav<T> &out, size_t dim, ptrdiff_t idxin, ptrdiff_t idxout, Func func)
size_t ndim = in.ndim();
if (dim+2==ndim)
sthelper2(, out.vdata()+idxout, in.shape(ndim-2), in.shape(ndim-1),
in.stride(ndim-2), in.stride(ndim-1), out.stride(ndim-2),
out.stride(ndim-1), func);
for (size_t i=0; i<in.shape(dim); ++i)
iter(in, out, dim+1, idxin+i*in.stride(dim), idxout+i*out.stride(dim), func);
template<typename T, typename Func> void transpose(const fmav<T> &in,
fmav<T> &out, Func func)
auto [shp, si, so] = prep(in, out);
fmav<T> in2(in, shp, si), out2(out, shp, so);
if (in2.ndim()==1) // 1D, just iterate
sthelper1(, out2.vdata(), in2.shape(0), in2.stride(0), out2.stride(0), func);
iter(in2, out2, 0, 0, 0, func);
using detail_transpose::transpose;
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment