+23
−5
Loading
Currently, for systems with multiple nodes and >100 cpu threads the time needed for map_cpus() to complete is ~4s. The bottleneck here is the is_cpu_online() which performs quite a lot of steps to yield such a simple information. Introduce new wrapper functions that will simply read from cpu's online attribute with a special case for cpu0 using the old routine. This drops the time needed to complete the map, on the same system, from ~4s to ~0.6s Change-Id: I66e004f5bb7505e47d70c487fe74a1b727856f58 Signed-off-by:Michal Berger <michal.berger@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/21030 Reviewed-by:
Konrad Sztyber <konrad.sztyber@intel.com> Reviewed-by:
Karol Latecki <karol.latecki@intel.com> Tested-by:
SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by:
Jim Harris <jim.harris@samsung.com>