Databases
StackOverflow2013
Postsdbo
POWhy does this parallel code in D scale so badly?
 

Note that there are some explanatory texts on larger screens.

plurals
  1. POWhy does this parallel code in D scale so badly?
    primarykey
    data
    text
    <p>Here is one experiment I performed comparing parallelism in C++ and D. I implemented an algorithm (a parallel label propagation scheme for community detection in networks) in both languages, using the same design: A parallel iterator gets a handle function (normally a closure) and applies it for every node in the graph. </p> <p>Here is the iterator in D, implemented using <code>taskPool</code> from <code>std.parallelism</code>:</p> <pre><code>/** * Iterate in parallel over all nodes of the graph and call handler (lambda closure). */ void parallelForNodes(F)(F handle) { foreach (node v; taskPool.parallel(std.range.iota(z))) { // call here handle(v); } } </code></pre> <p>And this is the handle function which is passed:</p> <pre><code> auto propagateLabels = (node v){ if (active[v] &amp;&amp; (G.degree(v) &gt; 0)) { integer[label] labelCounts; G.forNeighborsOf(v, (node w) { label lw = labels[w]; labelCounts[lw] += 1; // add weight of edge {v, w} }); // get dominant label label dominant; integer lcmax = 0; foreach (label l, integer lc; labelCounts) { if (lc &gt; lcmax) { dominant = l; lcmax = lc; } } if (labels[v] != dominant) { // UPDATE labels[v] = dominant; nUpdated += 1; // TODO: atomic update? G.forNeighborsOf(v, (node u) { active[u] = 1; }); } else { active[v] = 0; } } }; </code></pre> <p>The C++11 implementation is almost identical, but uses OpenMP for parallelization. So what do the scaling experiments show? </p> <p><img src="https://i.stack.imgur.com/2LloS.png" alt="scaling"></p> <p>Here I examine weak scaling, doubling the input graph size while also doubling the number of threads and measuring the running time. The ideal would be a straight line, but of course there is some overhead for parallelism. I use <code>defaultPoolThreads(nThreads)</code> in my main function to set the number of threads for the D program. The curve for C++ looks good, but the curve for D looks surprisingly bad. Am I doing something wrong w.r.t. D parallelism, or does this reflect badly on the scalability of parallel D programs?</p> <p>p.s. compiler flags</p> <p>for D: <code>rdmd -release -O -inline -noboundscheck</code></p> <p>for C++: <code>-std=c++11 -fopenmp -O3 -DNDEBUG</code></p> <p>pps. Something must be really wrong, because the D implementation is slower in parallel than sequentially:</p> <p><img src="https://i.stack.imgur.com/rtUih.png" alt="enter image description here"></p> <p>ppps. For the curious, here are the Mercurial clone urls for both implementations:</p> <ul> <li><a href="http://algohub.iti.kit.edu/parco/Prototypes/PLPd" rel="nofollow noreferrer">http://algohub.iti.kit.edu/parco/Prototypes/PLPd</a></li> <li><a href="http://algohub.iti.kit.edu/parco/Prototypes/PLPcpp" rel="nofollow noreferrer">http://algohub.iti.kit.edu/parco/Prototypes/PLPcpp</a></li> </ul>
    singulars
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. PTQuestion
    1. USclstaudt
    1. USclstaudt
    plurals
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. PO
      singulars
      1. PTAnswer
    2. PO
      singulars
      1. PTAnswer
    1. VO
      singulars
      1. POWhy does this parallel code in D scale so badly?
      1. This table or related slice is empty.
      1. VTUpMod
    2. VO
      singulars
      1. POWhy does this parallel code in D scale so badly?
      1. USKyle
      1. VTFavorite
    3. VO
      singulars
      1. POWhy does this parallel code in D scale so badly?
      1. USclstaudt
      1. VTFavorite
    1. COWhat does the performance look like if you did it without openmp?
      singulars
      1. POWhy does this parallel code in D scale so badly?
      1. USgreatwolf
    2. COFrom checking around it doesn't look like the dmd compiler currently supports openmp. It doesn't seem like an apples-to-apples comparsion to me if one version uses openmp and other version doesn't.
      singulars
      1. POWhy does this parallel code in D scale so badly?
      1. USgreatwolf
    3. CO@greatwolf Unless I misunderstand you, I believe you are missing the point. D does not have OpenMP, but it has the `std.parallelism` library, which does provide similar parallel constructs. In fact, the D program uses many cores when running.
      singulars
      1. POWhy does this parallel code in D scale so badly?
      1. USclstaudt
 

Querying!

 
Guidance
A row detail

Detail views are divided into sections. All the information in the data section comes from columns in the selected row. The other sections display data from other, related rows.

Related data can be related in a to-one or a to-many fashion. Captions of data related in a to-many fashion link to a list view showing a filtered view of the table.

Try moving around until you find a non-empty to-many entry and click on the label to get to one. You can move back to the root by clicking on the database name in the header.

SQuiL has stopped working due to an internal error.

If you are curious you may find further information in the browser console, which is accessible through the devtools (F12).

Reload