Performance comparison, ULE vs 4BSD and AMD64 vs i386

Brooks Davis brooks at one-eyed-alien.net
Wed Feb 25 10:45:21 PST 2004


On Wed, Feb 25, 2004 at 10:32:34AM -0800, David O'Brien wrote:
> On Wed, Feb 25, 2004 at 11:07:54AM -0500, Kenneth Culver wrote:
> > The buildworld problem could just be because it takes longer for the 
> > compiler to
> > generate amd64 code. In fact, I'm almost willing to bet that's the case 
> > since
> > amd64 has 2x the GPR's that x86 does. It's likely that it's harder to 
> > optimize
> > for it. Does anyone who knows compilers care to comment?
> 
> s/harder/slower/
> 
> It is defineately easier to optimize for amd64 because if its increased #
> of registers.  But I'm not sure even "slower" is a valid claim -- on the
> i386 the compiler has to do a lot of time figuring out the best way to do
> the spill code (when, where).

The other big thing with gcc for amd64 is that we're still more in
the "run correctly" then the the "run fast" stage in terms of the
development cycle.  With a compiler, "produces fast/correct code" beats
"run fast" any day.  The code generator has been widly used for less
then a year.  Compared to the life history of the i386 code generator
that's pretty short and the user base is still small relative to the
installed base of i386 machines.  Even if generating amd64 code is
easier then generating i386 code, it's probably still a bit early to
expect the compiler to do it quickly.

-- Brooks

-- 
Any statement of the form "X is the one, true Y" is FALSE.
PGP fingerprint 655D 519C 26A7 82E7 2529  9BF0 5D8E 8BE9 F238 1AD4
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-amd64/attachments/20040225/d20796ac/attachment.bin


More information about the freebsd-amd64 mailing list