Hi Tor,
I used to have build failures with a few benchmarks when using the
g77-era gcc compiler suite, version 3.4.6 (with the same glibc
version that you're using). I tried gcc 4.1.1 with the new fortran
tree, also with glibc 3.4.6, and all the benchmarks built fine (for
Alpha Linux). If you're curious, I can send you the various error and
warning logs that are pretty detailed about what fails where
(sometimes even why). I've also got the SPEC config files I used, if
that helps you.
I'm doing the build using the same cross compiler infrastructure
(crosstool-0.43) but on a centos 4.5 host. I believe the cross
compilers were generated with the default gcc 3.4.6 compiler, but
without taking another look at the crosstool logs I can't swear to it.
Best regards,
Dan
On Aug 26, 2007, at 3:29 PM, Steve Reinhardt wrote:
Hi Tor,
I don't have a lot of experience with this myself, so I'm
forwarding your request to the m5-users list.
Can anyone else out there help?
Steve
---------- Forwarded message ----------
From: Tor M.Aamodt <[EMAIL PROTECTED]>
Date: Aug 25, 2007 10:28 PM
Subject: SPEC 2000 alpha binary problems on M5
To: [EMAIL PROTECTED]
Cc: "Tor M. Aamodt" <[EMAIL PROTECTED]>
Hi Steve,
I'm having some trouble getting a few SPEC 2000 alpha binaries to
compile and run in system emulation mode on your M5 simulator
(m5-2.0b3). I've searched around and found the following
documentation which talks a bit about building alpha tools to get
alpha binaries:
< http://www.m5sim.org/wiki/index.php/Using_linux-
dist_to_Create_Disk_Images_and_Kernels_for_M5>
(a) RE: "gcc-3.4.3-glibc-2.3.5.dat" -- this does not seem to want
to build under a native gcc-4.1 system (which is what we happen to
have here). Seems that it is always hard to build older versions
of gcc with newer versions...
(b) RE: "however we have lately had some success with gcc-4.0.2
coupled with glibc-2.3.6." -- In contrast, I've successfully built
gcc-4.0.2 coupled with glibc-2.3.6 (using crosstool-0.43), but the
following SPEC INT 2000 benchmarks are having problems (with the
test inputs):
vpr
gcc
perlbmk
vortex
I'm wondering if I'm doing something wrong somewhere, or whether
you folks have seen problems with these same benchmarks under
gcc-4.0.2 coupled with glibc-2.3.6. Any insight?
Thanks,
Tor
--
Tor M. Aamodt, Ph.D., P.Eng.
Assistant Professor
Department of Electrical and Computer Engineering
University of British Columbia
Fred Kaiser Building, 2332 Main Mall, Office 4043
Vancouver, B.C. V6T 1Z4, CANADA
tel: +1.604.827.4116
fax: +1.604.822.5949
email: [EMAIL PROTECTED]
URL: http://www.ece.ubc.ca/~aamodt
_______________________________________________
m5-users mailing list
m5-users@m5sim.org
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
Daniel Alex Finkelstein
Ph.D. Candidate (CS)
Work: +1-718-260-3378
Mobile: +1-718-246-7421
Fax: +1-718-260-3609
Email: [EMAIL PROTECTED]
IM: steuben13 (Skype)
Polytechnic University
6 MetroTech Center
Brooklyn, NY 11201 USA
_______________________________________________
m5-users mailing list
m5-users@m5sim.org
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
