threading.py -> winthreading.c : Call for benchmark code
Ype Kingma
ykingma at accessforall.nl
Thu Dec 20 13:43:30 EST 2001
More information about the Python-list mailing list
Thu Dec 20 13:43:30 EST 2001
- Previous message (by thread): threading.py -> winthreading.c : Call for benchmark code
- Next message (by thread): threading.py -> winthreading.c : Call for benchmark code
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Jason, You might like this: http://www.bagley.org/~doug/shootout/bench/prodcons/ Have fun, Ype P.S. In jython I would recommend: http://g.oswego.edu/dl/classes/EDU/oswego/cs/dl/util/concurrent/intro.html Jason Orendorff wrote: > > Everyone, > > I am reimplementing parts of threading.py in non-portable > Win32-only C. Lock and RLock are already working. Tests show: > winthreading.Lock is about 25% faster than threading.Lock. > winthreading.RLock is about 26 times as fast as threading.Lock! > > But I'm not sure these numbers translate into real-world gains. > So: I'm looking for Python code that shows how these objects > (Lock, RLock, Condition, Semaphore, Event) are used in the wild. > > If you're interested in a 26x faster RLock, please send me > real-world Python code that could benefit. > > Thanks in advance, > Jason Orendorff > > [Technical note: I discovered something neat. Everything I do, > or check, or set, or increment in C code is protected by the global > interpreter lock. This eliminates a lot of the complexity of writing > good threading code, because I don't have to worry about doing > anything atomically! I can avoid functions like InterlockedIncrement() > and InterlockedExchange(). The resulting code is clearer and faster.]
- Previous message (by thread): threading.py -> winthreading.c : Call for benchmark code
- Next message (by thread): threading.py -> winthreading.c : Call for benchmark code
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Python-list mailing list