Databases
StackOverflow2013
Postsdbo
PO
 

Note that there are some explanatory texts on larger screens.

plurals
  1. PO
    primarykey
    data
    text
    <p><code>multiprocessing</code> makes some bad choices about pickling. Don't get me wrong, it makes some good choices that enable it to pickle certain types so they can be used in a pool's map function. However, since we have <code>dill</code> that can do the pickling, multiprocessing's own pickling becomes a bit limiting. Actually, if <code>multiprocessing</code> were to use <code>pickle</code> instead of <code>cPickle</code>... and also drop some of it's own pickling overrides, then <code>dill</code> could take over and give a much more full serialization for <code>multiprocessing</code>.</p> <p>Until that happens, there's a fork of <code>multiprocessing</code> called <a href="https://github.com/uqfoundation/pathos" rel="noreferrer">pathos</a> (the release version is a bit stale, unfortunately) that removes the above limitations. Pathos also adds some nice features that multiprocessing doesn't have, like multi-args in the map function. Pathos is due for a release, after some mild updating -- mostly conversion to python 3.x.</p> <pre><code>Python 2.7.5 (default, Sep 30 2013, 20:15:49) [GCC 4.2.1 (Apple Inc. build 5566)] on darwin Type "help", "copyright", "credits" or "license" for more information. &gt;&gt;&gt; import dill &gt;&gt;&gt; from pathos.multiprocessing import ProcessingPool &gt;&gt;&gt; pool = ProcessingPool(nodes=4) &gt;&gt;&gt; result = pool.map(lambda x: x**2, range(10)) &gt;&gt;&gt; result [0, 1, 4, 9, 16, 25, 36, 49, 64, 81] </code></pre> <p>and just to show off a little of what <code>pathos.multiprocessing</code> can do...</p> <pre><code>&gt;&gt;&gt; def busy_add(x,y, delay=0.01): ... for n in range(x): ... x += n ... for n in range(y): ... y -= n ... import time ... time.sleep(delay) ... return x + y ... &gt;&gt;&gt; def busy_squared(x): ... import time, random ... time.sleep(2*random.random()) ... return x*x ... &gt;&gt;&gt; def squared(x): ... return x*x ... &gt;&gt;&gt; def quad_factory(a=1, b=1, c=0): ... def quad(x): ... return a*x**2 + b*x + c ... return quad ... &gt;&gt;&gt; square_plus_one = quad_factory(2,0,1) &gt;&gt;&gt; &gt;&gt;&gt; def test1(pool): ... print pool ... print "x: %s\n" % str(x) ... print pool.map.__name__ ... start = time.time() ... res = pool.map(squared, x) ... print "time to results:", time.time() - start ... print "y: %s\n" % str(res) ... print pool.imap.__name__ ... start = time.time() ... res = pool.imap(squared, x) ... print "time to queue:", time.time() - start ... start = time.time() ... res = list(res) ... print "time to results:", time.time() - start ... print "y: %s\n" % str(res) ... print pool.amap.__name__ ... start = time.time() ... res = pool.amap(squared, x) ... print "time to queue:", time.time() - start ... start = time.time() ... res = res.get() ... print "time to results:", time.time() - start ... print "y: %s\n" % str(res) ... &gt;&gt;&gt; def test2(pool, items=4, delay=0): ... _x = range(-items/2,items/2,2) ... _y = range(len(_x)) ... _d = [delay]*len(_x) ... print map ... res1 = map(busy_squared, _x) ... res2 = map(busy_add, _x, _y, _d) ... print pool.map ... _res1 = pool.map(busy_squared, _x) ... _res2 = pool.map(busy_add, _x, _y, _d) ... assert _res1 == res1 ... assert _res2 == res2 ... print pool.imap ... _res1 = pool.imap(busy_squared, _x) ... _res2 = pool.imap(busy_add, _x, _y, _d) ... assert list(_res1) == res1 ... assert list(_res2) == res2 ... print pool.amap ... _res1 = pool.amap(busy_squared, _x) ... _res2 = pool.amap(busy_add, _x, _y, _d) ... assert _res1.get() == res1 ... assert _res2.get() == res2 ... print "" ... &gt;&gt;&gt; def test3(pool): # test against a function that should fail in pickle ... print pool ... print "x: %s\n" % str(x) ... print pool.map.__name__ ... start = time.time() ... res = pool.map(square_plus_one, x) ... print "time to results:", time.time() - start ... print "y: %s\n" % str(res) ... &gt;&gt;&gt; def test4(pool, maxtries, delay): ... print pool ... m = pool.amap(busy_add, x, x) ... tries = 0 ... while not m.ready(): ... time.sleep(delay) ... tries += 1 ... print "TRY: %s" % tries ... if tries &gt;= maxtries: ... print "TIMEOUT" ... break ... print m.get() ... &gt;&gt;&gt; import time &gt;&gt;&gt; x = range(18) &gt;&gt;&gt; delay = 0.01 &gt;&gt;&gt; items = 20 &gt;&gt;&gt; maxtries = 20 &gt;&gt;&gt; from pathos.multiprocessing import ProcessingPool as Pool &gt;&gt;&gt; pool = Pool(nodes=4) &gt;&gt;&gt; test1(pool) &lt;pool ProcessingPool(ncpus=4)&gt; x: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17] map time to results: 0.0553691387177 y: [0, 1, 4, 9, 16, 25, 36, 49, 64, 81, 100, 121, 144, 169, 196, 225, 256, 289] imap time to queue: 7.91549682617e-05 time to results: 0.102381229401 y: [0, 1, 4, 9, 16, 25, 36, 49, 64, 81, 100, 121, 144, 169, 196, 225, 256, 289] amap time to queue: 7.08103179932e-05 time to results: 0.0489699840546 y: [0, 1, 4, 9, 16, 25, 36, 49, 64, 81, 100, 121, 144, 169, 196, 225, 256, 289] &gt;&gt;&gt; test2(pool, items, delay) &lt;built-in function map&gt; &lt;bound method ProcessingPool.map of &lt;pool ProcessingPool(ncpus=4)&gt;&gt; &lt;bound method ProcessingPool.imap of &lt;pool ProcessingPool(ncpus=4)&gt;&gt; &lt;bound method ProcessingPool.amap of &lt;pool ProcessingPool(ncpus=4)&gt;&gt; &gt;&gt;&gt; test3(pool) &lt;pool ProcessingPool(ncpus=4)&gt; x: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17] map time to results: 0.0523059368134 y: [1, 3, 9, 19, 33, 51, 73, 99, 129, 163, 201, 243, 289, 339, 393, 451, 513, 579] &gt;&gt;&gt; test4(pool, maxtries, delay) &lt;pool ProcessingPool(ncpus=4)&gt; TRY: 1 TRY: 2 TRY: 3 TRY: 4 TRY: 5 TRY: 6 TRY: 7 [0, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32, 34] </code></pre>
    singulars
    1. This table or related slice is empty.
    1. POWhat can multiprocessing and dill do together?
      singulars
      1. PTQuestion
    1. PTAnswer
    1. USlaike9m
    1. USMike McKerns
    plurals
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. POWhat can multiprocessing and dill do together?
      singulars
      1. PTQuestion
    1. This table or related slice is empty.
    1. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTUpMod
    2. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTUpMod
    3. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTUpMod
    1. COSo is the problem that multiprocessing uses `cPickle` rather than `pickle` causing `dill` to be unable to perform its usual magic?
      singulars
      1. PO
      1. USMRocklin
    2. COYes, primarily. More recent versions of multiprocessing try to pickle certain types, but some of the picklings are not as (cough) thorough as those that dill does. So, my fork does two things (1) replace cPickle with pickle, and (2) delete a few of multiprocessing's attempts at pickling. I've also added some features to my fork like the multi-args map, but if multiprocessing did the above two things... then that would enable dill magic to happen in multiprocessing.
      singulars
      1. PO
      1. USMike McKerns
    3. COI haven't yet figured out how to make a cPickle "override"... Multiprocessing is the primary reason to do so, at least for the versions using cPickle. In python 3.x, there is no cPickle, so there should be less of a limitation that multiprocessing imposes on itself.
      singulars
      1. PO
      1. USMike McKerns
 

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