On Mon, Jun 16, 2008 at 02:00:15PM +0200, Andi Kleen wrote: > > Took a closer look at numactl 2.0.0 and noticed some more problems. > Also did some numademo work and added a new benchmark (random pointer chase) > The main bug fix is for kernels without cpuset support and also some > smaller stuff. > > -Andi > > CHANGES entry: > > - Various numademo improvements: > * Fix random benchmark to use all specified memory > * Rename to random2 to signify it's different > * Optimize random benchmark by inlining random number generator fast path. > * Clear caches between runs for more stable results > * Add new random pointer chaser benchmark > * Compile benchmarks with gcc vectorization if available > * run numademo in regression test > - Add numa_exit_on_warn > - Fix no cpuset fallback in libnuma.c > - Install symlinks for the manpages of all new functions > - Make internal libnuma functions static > - Add copy_bitmask_to_bitmask() to numa.h > - Some cleanups > - Fix line reading in proc > - Add versions.ldscript to libnuma.so dependencies I've added your whole series of 17 patches to numactl/libnuma and posted it on ftp://oss.sgi.com/www/projects/libnuma/download as numactl-2.0.2-rc1.tar.gz (Plus the tweaks to numademo and numacompat1.h/numa.h that we've been discussing.) It runs the regression tests successfully for me, on ia64. I'll let it simmer there for a while for any further tweaks and afterthoughts. And for others' testing. -Cliff -- Cliff Wickman Silicon Graphics, Inc. cpw@xxxxxxx (651) 683-3824 -- To unsubscribe from this list: send the line "unsubscribe linux-numa" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
