On Mon, 16 Jan 2023, Ahmed S. Darwish wrote: > On Mon, 16 Jan 2023, John Kacur wrote: > > > > On Thu, 12 Jan 2023, Sebastian Andrzej Siewior wrote: > > > > > numa_num_task_cpus() returns here > 100 CPUs while the system has only > > > 32 populated. The BIOS assumes that I can probably use larger CPUs (with > > > more cores) on the socket so the number of "configured CPUs" is rather > > > high. > > > For default configuration, only with the -S option, it makes sense to > > > use the current affinity instead looking at the number of possible CPUs > > > which could be brought online. It still depends on the affinity of the > > > created threads if the additional CPUs can be used. In a container setup > > > this may not be the case. > > > > > > Use sched_getaffinity() to figure out the number of possible CPUs. > ... > > > > In cyclictest we have > > > > if (num_threads == -1) > > num_threads = get_available_cpus(affinity_mask); > > > > > > But the num_threads can be set to -1 in OPT_THREADS too, > > not just for OPT_SMP > > A sane default for --smp, which is almost-always used in sane cyclictest > invocations, is still valuable IMHO. > > Thanks, > > -- > Ahmed S. Darwish > Linutronix GmbH > Okay, not sure how that adds to the conversation. I'm all for "sane" defaults for --smp too, but we care about other kinds of invocations too. As soon as I get a patch that doesn't break other things as well, I'll apply it. John
