Re: cost of -fnon-call-exceptions

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



forgot to reply the the list...

From: Ian Lance Taylor <iant@xxxxxxxxxx>
Subject: Re: cost of -fnon-call-exceptions
To: "Kenny Simpson" <theonetruekenny@xxxxxxxxx>
Date: Friday, September 10, 2010, 3:48 PM
Kenny Simpson <theonetruekenny@xxxxxxxxx>
writes:

> Meaning every memory reference is treated as though it were volatile ?
> Any rough idea how much this would hurt ?

Please reply to the mailing list, not just to me. 
Thanks.

It's not really accurate to say that every memory reference
is treated as volatile.  A memory reference which is not used
will still be removed from the program.  Also the compiler knows that many
memory references can not trap--e.g., references to variables stored on the
stack.  It would be more accurate to say that the compiler will not in
general reorder indirections through pointers.  How much that
hurts depends a lot on your processor.  On modern x86 other than Atom,
it doesn't hurt much, because the processors are not in-order and have
sophisticated memory reordering.  On Atom it will hurt, but how much
it will hurt depends on your application.  I don't think there is
any one answer.

Ian







[Index of Archives]     [Linux C Programming]     [Linux Kernel]     [eCos]     [Fedora Development]     [Fedora Announce]     [Autoconf]     [The DWARVES Debugging Tools]     [Yosemite Campsites]     [Yosemite News]     [Linux GCC]

  Powered by Linux