Comments on: Stackless: You got your coroutines in my subroutines. http://jessenoller.com/2009/02/23/stackless-you-got-your-coroutines-in-my-subroutines/ python, programming and other things Mon, 02 Jan 2012 09:21:00 +0000 hourly 1 By: Adam Olsen http://jessenoller.com/2009/02/23/stackless-you-got-your-coroutines-in-my-subroutines/comment-page-1/#comment-139123 Adam Olsen Tue, 03 Mar 2009 12:49:52 +0000 http://jessenoller.com/?p=495#comment-139123 That's true for a 32-bit box with default stack sizes. Programs using a large number of threads frequently use much lower stack sizes (a factor of 100 is possible, depending on what the threads do). 64-bit CPUs make this moot however, as they have 64-bit pointers and currently support up to 48-bits of address space, meaning about 25 million threads. That’s true for a 32-bit box with default stack sizes. Programs using a large number of threads frequently use much lower stack sizes (a factor of 100 is possible, depending on what the threads do). 64-bit CPUs make this moot however, as they have 64-bit pointers and currently support up to 48-bits of address space, meaning about 25 million threads.

]]> By: jnoller http://jessenoller.com/2009/02/23/stackless-you-got-your-coroutines-in-my-subroutines/comment-page-1/#comment-66974 jnoller Tue, 03 Mar 2009 12:04:14 +0000 http://jessenoller.com/?p=495#comment-66974 Approved Approved

]]> By: Adam Olsen http://jessenoller.com/2009/02/23/stackless-you-got-your-coroutines-in-my-subroutines/comment-page-1/#comment-66973 Adam Olsen Tue, 03 Mar 2009 07:49:52 +0000 http://jessenoller.com/?p=495#comment-66973 That's true for a 32-bit box with default stack sizes. Programs using a large number of threads frequently use much lower stack sizes (a factor of 100 is possible, depending on what the threads do). 64-bit CPUs make this moot however, as they have 64-bit pointers and currently support up to 48-bits of address space, meaning about 25 million threads. That’s true for a 32-bit box with default stack sizes. Programs using a large number of threads frequently use much lower stack sizes (a factor of 100 is possible, depending on what the threads do). 64-bit CPUs make this moot however, as they have 64-bit pointers and currently support up to 48-bits of address space, meaning about 25 million threads.

]]> By: jnoller http://jessenoller.com/2009/02/23/stackless-you-got-your-coroutines-in-my-subroutines/comment-page-1/#comment-66916 jnoller Mon, 02 Mar 2009 12:41:13 +0000 http://jessenoller.com/?p=495#comment-66916 Actually, that's incorrect. I know for a fact given I have a machine<br>running with over 7325 threads active. I had to raise the ulimit a bit<br>because they're chewing on sockets. This is on ubuntu hardy heron,<br>FWIW. Actually, that’s incorrect. I know for a fact given I have a machine
running with over 7325 threads active. I had to raise the ulimit a bit
because they’re chewing on sockets. This is on ubuntu hardy heron,
FWIW.

]]> By: fijal http://jessenoller.com/2009/02/23/stackless-you-got-your-coroutines-in-my-subroutines/comment-page-1/#comment-66915 fijal Mon, 02 Mar 2009 08:15:34 +0000 http://jessenoller.com/?p=495#comment-66915 Note that there is one frustrating limitation with threads - on linux, by default you cannot have more than 372 threads.<br>period. That's because each thread consumes 8M virtual memory and you run out of addresss space.<br><br>Cheers,<br>fijal Note that there is one frustrating limitation with threads — on linux, by default you cannot have more than 372 threads.
period. That’s because each thread consumes 8M virtual memory and you run out of addresss space.

Cheers,
fijal

]]> By: jnoller http://jessenoller.com/2009/02/23/stackless-you-got-your-coroutines-in-my-subroutines/comment-page-1/#comment-66223 jnoller Tue, 24 Feb 2009 14:39:19 +0000 http://jessenoller.com/?p=495#comment-66223 I've fixed it - I misread something, I apologize. I’ve fixed it — I misread something, I apologize.

]]> By: Richard Tew http://jessenoller.com/2009/02/23/stackless-you-got-your-coroutines-in-my-subroutines/comment-page-1/#comment-66222 Richard Tew Tue, 24 Feb 2009 13:07:44 +0000 http://jessenoller.com/?p=495#comment-66222 Regarding "Some platforms simply aren’t supported, or are only partially supported due to this." <br><br>Either platforms are supported or they are not. If there is no stack switching code for a given platform, then Stackless will compile as normal Python. There is no such thing as partial support. Regarding “Some platforms simply aren’t supported, or are only partially supported due to this.”

Either platforms are supported or they are not. If there is no stack switching code for a given platform, then Stackless will compile as normal Python. There is no such thing as partial support.

]]>