Jython, GILs and object locking.
Pettersen, Bjorn S
BjornPettersen at fairisaac.com
Fri Oct 10 21:39:38 EDT 2003
More information about the Python-list mailing list
Fri Oct 10 21:39:38 EDT 2003
- Previous message (by thread): Jython, GILs and object locking.
- Next message (by thread): Jython, GILs and object locking.
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
> From: Jon Franz_antispam [mailto:jfranzXREMOVEX at neurokode.com] > > > Most of the I/O code that's built into python, already does this. It's > > written in C, and the C code releases the GIL before engaging in the > > operation. Or, I should add, so I have been informed by previous > > beatings of this horse carcass. > > I have yet to see such code run on a seperate cpu, but I may > be mistaken... > I'll make some test cases. Why would IO code called from _your_ thread run on a different cpu? Your thread is blocked, waiting for the IO to finish. While it is blocked, however, it does _not_ hold the gil -- thus other threads can run on both _your thread's_ processor, as well as any other processors you can get the os to schedule them on. Ergo, this is probably the case where the gil has the _least_ impact. -- bjorn
- Previous message (by thread): Jython, GILs and object locking.
- Next message (by thread): Jython, GILs and object locking.
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Python-list mailing list