Brice, I just tried that but the output is the same. As Samuel pointed out, it looks like /proc/cpuinfo is totally bogus for this kernel. Comparing it with what it "should" be emitting (according to http://kbase.redhat.com/faq/docs/DOC-7715), it doesn't report more than one under "cpu cores", and core id is always 255!
Kind regards, Dan -----Original Message----- From: hwloc-users-boun...@open-mpi.org [mailto:hwloc-users-boun...@open-mpi.org] On Behalf Of Brice Goglin Sent: Thursday, January 28, 2010 11:32 AM To: Hardware locality user list Subject: Re: [hwloc-users] No caches or hiearchy on RHEL 4.7 or 4.8 Dan Eaton wrote: > I couldn't compile trunk (rather, autogen.sh failed because it requires > Autoconf > 2.63), so I just pulled the nightly build hwloc-1.0a1r1682 and > applied the patch there. Does that sound reasonable? > Yes. By the way, there's an undocumented environment variable to force hwloc to read /proc/cpuinfo instead of sysfs. Try setting HWLOC_LINUX_USE_CPUINFO=1 before running the old lstopo (without the x86 backend), it might be better. Brice _______________________________________________ hwloc-users mailing list hwloc-us...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/hwloc-users
