I have included indicative performance test numbers from a netperf
TCP_CRR test, tested on top of the kernel side lock contention patches
sent out separately.
Jarno Rajahalme (7):
56.8k (datapath lock contention patches)
meta-flow: Fix setting MFF_IP_FRAG.
57k ofproto: Lock for vlan splinters only if have them.
58k ofproto: Optimize the case of a repeated learn action execution.
ofproto: Move 'rule->used' to the provider.
lib/ovs-thread: ovs_spinlock
59k ofproto-dpif: Use ovs_spinlock for stats.
60k meta-flow: Add mf_mask_subfield_wc()
lib/learn.c | 4 +-
lib/meta-flow.c | 160 ++++++++++++++++++++++++++++++++++++++-
lib/meta-flow.h | 4 +-
lib/nx-match.c | 15 +---
lib/ovs-thread.c | 56 +++++++++-----
lib/ovs-thread.h | 26 +++++++
ofproto/ofproto-dpif-xlate.c | 27 +++----
ofproto/ofproto-dpif.c | 169 ++++++++++++++++++++++++++----------------
ofproto/ofproto-provider.h | 17 +++--
ofproto/ofproto.c | 107 +++++++++++++++++++-------
tests/ofproto-dpif.at | 4 +-
11 files changed, 437 insertions(+), 152 deletions(-)
--
1.7.10.4
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
