Ajay wrote:
i have an application that runs on a pocket pc.
[...] my question is, in terms of memory and processing
power, how much overhead will creating threads cost.
I've never worked with Pocket-PC, but the answer is almost
certainly "not much". You can get reasonable bounds with simple
test programs. I'll include a Python thread creation timer
below.
Threading has improved vastly in recent years, and the current
versions of the popular operating systems now have excellent
thread support. As one might expect, out-dated ideas about
the cost of threads are still widespread.
--Bryan
import thread
import time
lock = thread.allocate_lock()
lst = [0]
def increment():
lock.acquire()
lst[0] += 1
lock.release()
def mass_thread(nthreads):
print "Running %d threads..." % nthreads
i = 0
benchmark = time.clock()
while i < nthreads:
try:
thread.start_new_thread(increment, ())
i += 1
except:
# thread.error is undocumented, so catch all
time.sleep(0.05)
go = 1
while go:
time.sleep(0.1)
lock.acquire()
if lst[0] == nthreads:
go = 0
lock.release()
benchmark = time.clock() - benchmark
print "All %s threads have run." % lst[0]
print "That took %f seconds." % benchmark
if __name__ == '__main__':
mass_thread(10000)
