chizeta/cnorxz
chizeta/cnorxz
0
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

update readme + multithread one test

Browse source
This commit is contained in:
Christian Zimmermann 2019-01-16 10:23:18 +01:00
parent 6d1682efa2
commit e1b6271a95
2 changed files with 10 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

9
README.md
View file

@ -100,6 +100,15 @@ res2(i1,i2) = res1(i1,i2) / fma(i2,i1)
Further examples will follow; you can also look in the test executable source files in `./src/test/*.cc`
# Multithreading
Multithreading can be enabled by the operation class member function `par()`, which will parallelize the top loop. E.g. for the operation in the example above:
```c++
/* Enable multithreading for the loop over i1 */
res1(i1,i2).par() = (ma_a(i1,i2,i4,i5) + ma_b(i4,i3)).c(i4);
```
Caution: If you want to multithread operations over functional multi arrays, you have to make sure by yourself that involved functors are either static or const (const member variables).
# Planned features
Basically the structure of the operation routines allows in some cases (at least in those, where it is possible for simple c++ for loops) vectorization and multi threading. So far the necessary implamentations are not done, but it is planned to enable these features.

2
src/tests/op_unit_test.cc
View file

@ -457,7 +457,7 @@ namespace {
std::cout << "MultiArray time: " << static_cast<double>( end - begin ) / CLOCKS_PER_SEC
<< std::endl;
res2(delta, deltap) += ma(delta, alpha, alpha, beta, beta, gamma, gamma, deltap).c(alpha).c(beta).c(gamma);
res2(delta, deltap).par() += ma(delta, alpha, alpha, beta, beta, gamma, gamma, deltap).c(alpha).c(beta).c(gamma);
std::vector<double> vres(4*4);