From mboxrd@z Thu Jan  1 00:00:00 1970
From: Jeffrey A Law <law@upchuck.cygnus.com>
To: Richard Henderson <rth@cygnus.com>
Cc: Jason Merrill <jason@cygnus.com>, Sylvain Pion <Sylvain.Pion@sophia.inria.fr>, egcs@egcs.cygnus.com
Subject: Re: C++ default copy ctor not optimal 
Date: Sun, 28 Feb 1999 22:53:00 -0000
Message-ID: <14555.919137968@upchuck>
References: <19990215171524.A19063@cygnus.com>
X-SW-Source: 1999-02n/msg00694.html
Message-ID: <19990228225300.RI8gtK6gmsywzGT3wQr3M-0Nn2O8xAFpfnODLQRLNMc@z>

  In message < 19990215171524.A19063@cygnus.com >you write:
  > On Sun, Feb 14, 1999 at 09:01:29PM -0800, Jason Merrill wrote:
  > > The backend should be able to determine that the struct it's copying is
  > > composed of FP values and use FP instructions for the copy.  I don't thin
  > k
  > > we can do that in general on the x86 because the FPU faults if you try to
  > > load an invalid FP value, but I may be remembering wrong.
  > 
  > The x86 fpu can load DImode values without faulting, and since
  > the frational part of the extended double register is 64-bits
  > wide, we don't lose bits.
But is it profitable?  Particularly in cases where the addresses are
not 64bit aligned?

jeff