In the stress test of NVMe TCP (ARM platform, 6 nvme disks), we see nvmf_tcp_req_set_state() takes quite some CPU cycles (about 2%~3% of the nvmf_tgt process, ranking 6) moving TCP request structure between different queues. And after some analyzes, we think these actions can be saved. With this change we get 1%~1.5% performance gain overall. Change-Id: Ifd2f5609e4d99cab9fea06e773b461ded6320e93 Signed-off-by: Rui Chang <rui.chang@arm.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5667 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: Broadcom CI Community-CI: Mellanox Build Bot Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> |
||
---|---|---|
.. | ||
.gitignore | ||
Makefile | ||
tcp_ut.c |