>>>>> "Michael" == Michael K Edwards <[EMAIL PROTECTED]> writes:
Michael> On 5/18/05, Hubert Chan <[EMAIL PROTECTED]> wrote: >> The configure flags for compiling libgc are: --disable-threads >> --disable-shared --enable-cplusplus (obviously with --disable-threads >> toggled for the different tests). I'm also running on a >> single-processor machine. Michael> Not a hyperthreaded Xeon, for instance? No, I'm running on a Pentium M. Michael> Are you running a UP kernel so that (kernel-level) spinlocks Michael> are no-ops? I'm running on a kernel that has SMP support disabled. (The kernel .config contains "# CONFIG_SMP is not set".) Kernel preemption is turned on, though; I don't know if that makes a difference. Michael> Note also that the libgc implementation of threading primitives Michael> on Linux was historically tied pretty closely to linuxthreads, Michael> and parsed "cpu" lines out of /proc/stat to find out whether to Michael> use UP or SMP tunings. I haven't looked at the current code Michael> yet, but you might want to make sure you are compiling Michael> specifically for NPTL and do not have USE_SPIN_LOCK defined Michael> (which forces the use of a libgc spinlock implementation that Michael> loses on NPTL). Well, I'm trying to compare the single-threaded libgc against the stock libgc that is in Debian, so for my baseline, I'm not going to do anything that Ryan doesn't do in his package (other than, of course, static linking, to make the comparison fair). -- Hubert Chan <[EMAIL PROTECTED]> - http://www.uhoreg.ca/ PGP/GnuPG key: 1024D/124B61FA Fingerprint: 96C5 012F 5F74 A5F7 1FF7 5291 AF29 C719 124B 61FA Key available at wwwkeys.pgp.net. Encrypted e-mail preferred. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
