Databases
StackOverflow2013
Postsdbo
POProblems with setting OMP_THREAD_LIMIT during runtime (c++ gcc 4.4.7)
 

Note that there are some explanatory texts on larger screens.

plurals
  1. POProblems with setting OMP_THREAD_LIMIT during runtime (c++ gcc 4.4.7)
    primarykey
    data
    text
    <p>Heluuu,</p> <p>I have a rather large program that I'm attempting to thread. So far, this has been succesful, and the basics are all working as intended.</p> <p>I now want to do some fancy work with cascading threads in nested mode. Essentially, I want the main parallel region to use any free threads in lower parallel regions.</p> <p>To detail the current system, the main parallel region starts 10 threads. I have 12 cores, so I can use 2 more threads. There is a second parallel region where some heavy computing happens, and I want the first two threads to reach this point to start a new team there, each with 2 threads. Every new entry to the lower parallel region after this will continue in serial. </p> <p>So, this should look like the following.<br> Main region: 10 threads started.<br> Lower region: 2 new threads started. </p> <p>Thread 1: 2 threads in lower region.<br> Thread 2: 2 threads in lower region.<br> Thread 3-10: 1 thread in lower region. </p> <p>Please keep in mind that these numbers are for the sake of clarity in providing a concrete description of my situation, and not the absolute and only case in which the program operates. </p> <p>The code:</p> <pre><code>main() { ... ... omp_set_num_threads(n); omp_set_dynamic(x); #pragma omp parallel { #pragma omp for for (int i = 0; i &lt; iterations; i++) { ... Compute(); ... } } } </code></pre> <p>And in Compute</p> <pre><code>bool Compute() { ... float nThreads = omp_get_thread_limit() - omp_get_num_threads(); nThreads = ceil(nThreads / omp_get_num_threads()); omp_set_num_threads((int)nThreads); #pragma omp parallel { ... #pragma omp for for (int i = 0; i &lt; nReductSize; i++) { ... } } } </code></pre> <p>Now, my problem is that setting the uppermost limit for the whole program (i.e. OMP_THREAD_LIMIT) only works from outside the program. Using </p> <pre><code>export OMP_THREAD_LIMIT=5 </code></pre> <p>from the bash command line works great. But I want to do it internally. So far, I've tried </p> <pre><code>putenv("OMP_THREAD_LIMIT=12"); setenv("OMP_THREAD_LIMIT", "12", 1); </code></pre> <p>but when I call omp_get_thread_limit() or getenv("OMP_THREAD_LIMIT") I get wacky return values. Even when I set the variable with export, calling getenv("OMP_THREAD_LIMIT"); returns 0.<br> So, I would ask for your help in this: How do I properly set OMP_THREAD_LIMIT at runtime?</p> <p>This is the main function where I set the thread defaults. It is executed well before any threading occurs:</p> <pre><code>#ifdef _OPENMP const char *name = "OMP_THREAD_LIMIT"; const char *value = "5"; int overwrite = 1; int success = setenv(name, value, overwrite); cout &lt;&lt; "Var set (0 is success): " &lt;&lt; success &lt;&lt; endl; #endif </code></pre> <p>Oh, and setenv reports success in setting the variable. </p> <p>Compiler says<br> gcc44 (GCC) 4.4.7 20120313 (Red Hat 4.4.7-1) </p> <p>Flags<br> CCFLAGS = -c -O0 -fopenmp -g -msse -msse2 -msse3 -mfpmath=sse -std=c++0x </p> <p>OpenMP version is 3.0.</p>
    singulars
    1. PO
      singulars
      1. PTAnswer
    1. This table or related slice is empty.
    1. PTQuestion
    1. USDess
    1. USDess
    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
    3. PO
      singulars
      1. PTAnswer
    1. This table or related slice is empty.
    1. COOpenMP will not see your 'putenv'/'setenv' if you did it during runtime (after openmp library is initialized).
      singulars
      1. POProblems with setting OMP_THREAD_LIMIT during runtime (c++ gcc 4.4.7)
      1. USosgx
    2. CO`OMP_THREAD_LIMIT` is supposed to enforce externally the upper limit on the size of the thread pool used by the OpenMP runtime. I as a user would be particularly upset if there are provisions in OpenMP that would allow the program to ignore the value of `OMP_THREAD_LIMIT` that I have set (e.g. to a value that best matches my hardware) and/or to change it to an arbitrary chosen by the application programmer value.
      singulars
      1. POProblems with setting OMP_THREAD_LIMIT during runtime (c++ gcc 4.4.7)
      1. USHristo Iliev
    3. COI can see how that might be troublesome. But if that is what you perceive me as attempting, you did not read past the first paragraph (the purpose is stated in paragraph 2). I want to use OMP_THREAD_LIMIT as the upper limit for the number of concurrent threads (as it is intended I believe), and disperse the threads available in the most efficient manner I can. Since OMP_THREAD_LIMIT does not appear to be set by default to any practical limitation, I would like to set it within the program to avoid users having to do this manually.
      singulars
      1. POProblems with setting OMP_THREAD_LIMIT during runtime (c++ gcc 4.4.7)
      1. USDess
 

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