On Sat, Sep 19, 1998 at 08:57:34PM -0600, Alex Rousskov wrote:
> Care to benchmark the performance of a Unix pipe compared to a
> shared memory with locks? Any known studies that have done that?
Not that I know of - this isn't the reason I want to do this. I want
to do this because it makes it possible for 4 or more processors on a
big box to manipulate common data structures effectively, pipes are
typically end to end.
The situation is much more complex than this, I'll write it up and
post it later this week - I can;t be bothered right now as I feel
like crap (hang over).
> I would be interested to see if there is a big difference for
> *Squid IPC traffic patterns*. In case memory locks win big,
> somebody might want to patch things to use [probably
> less-portable?] shared memory primitives. If not, there will be no
> reason for extra coding.
Its more than just locking - something can be done lock-free. Its
highly non-portable, and involves assembly primitives for different
compilers and architectures.
> Just a thought; before the list will be filled with long
> discussions on shared memory pros and cons. :)
The change is only worth doing if it buys us something... and I'm not
convinced it does, it just makes ideas clearer in my mind.
-cw
Received on Tue Jul 29 2003 - 13:15:54 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:11:56 MST