Setting thread priorities

[John Nagle]

There's no way to set thread priorities within Python, is there?
We have some threads that go compute-bound, and would like to
reduce their priority slightly so the other operations, like
accessing the database and servicing queries, aren't slowed
as much.

John Nagle

[Gerald Kaszuba]

Re: Setting thread priorities

Hi John

On May 13, 4:46 pm, John Nagle <n...@animats.c omwrote:

There's no way to set thread priorities within Python, is there?

Not exactly. You can however use the ctypes module to access the o/s
methods of pthread_setsche dparam() for UNIX and SetThreadPriori ty()
for Windows.

I'm not sure why this hasn't been implemented in Python.

Gerald

http://geraldkaszuba.com/

[Grant Edwards]

Re: Setting thread priorities

On 2007-05-13, Gerald Kaszuba <gerald.kaszuba @gmail.comwrote :

Hi John
>
On May 13, 4:46 pm, John Nagle <n...@animats.c omwrote:

> There's no way to set thread priorities within Python, is there?

>
Not exactly. You can however use the ctypes module to access the o/s
methods of pthread_setsche dparam() for UNIX and SetThreadPriori ty()
for Windows.
>
I'm not sure why this hasn't been implemented in Python.

AFAICT, Python's threading paradigm seems to be based on the
assumption that alls threads spend most of their time blocking
on I/O or some other event. In that case priorities don't
matter. Priorities only matter if multiple threads are ready to
run.

--
Grant Edwards
grante@visi.com

[Nick Vatamaniuc]

Re: Setting thread priorities

On May 13, 2:46 am, John Nagle <n...@animats.c omwrote:

There's no way to set thread priorities within Python, is there?
We have some threads that go compute-bound, and would like to
reduce their priority slightly so the other operations, like
accessing the database and servicing queries, aren't slowed
as much.
>
John Nagle

John,

You can implicitly create a priority system if you let some threads
yield more to other threads. Say if your one thread has to have a
higher priority, you would make other threads sleep longer. Of course,
as someone mentioned, Python threads are mostly there to take
advantage of I/O blocking and not to take advantage of multiple CPU
cores (it can't do that at the moment anyway).

Check out the small snippet of code I wrote below. Each thread is
initialized with some priority value between 0.0 (the lowest) and up.
Then look at the run trace and notice that on average the 0.75
priority thread is called more often than the 1.0 priority.

Hope this helped,
-Nick Vatamaniuc

>>from threading import Thread
>>from time import sleep
>>class Worker(Thread):

...: def __init__(self,p ri):
...: Thread.__init__ (self)
...: self.pri=pri
...: def run(self):
...: for i in range(20):
...: sleep(1.0*self. pri)
...: print " -thread with priority:",self .pri

>>w1=Worker(1.0 ); w2=Worker(0.75)

>>w1.start(); w2.start()

>> -thread with priority: 0.75

-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 0.75
-thread with priority: 1.0
-thread with priority: 1.0
-thread with priority: 1.0
-thread with priority: 1.0
-thread with priority: 1.0

>>>