Well inspite of all the "this talk is not about.." - I still found this to be tremendously helpful in understanding python behavior in a multi-core world. <div><br></div><div>Sure, we all prefer message passing, blah blah blah, but for a lot of web services when you're running a whole stack of forked python interpreters to maintain some degree of parallism every once in a while you ask yourself "where did all my memory go?" because each process is taking 30-100 Megs, and you start to wonder "what would I get from threads?"<br>
<br></div><div>The last bit about GIL contention on multi-core machines is pretty interesting too - I'm not much into systems stuff, but I really wonder if there's room for improvement there.</div><div><br></div><div>
Alec</div><div><br><div class="gmail_quote">On Fri, Jun 12, 2009 at 12:28 PM, Benjamin Sergeant <span dir="ltr"><<a href="mailto:bsergean@gmail.com">bsergean@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<a href="http://www.dabeaz.com/python/GIL.pdf" target="_blank">http://www.dabeaz.com/python/GIL.pdf</a></blockquote></div><br></div>