[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1239322187-3764-2-git-send-email-andy.grover@oracle.com>
Date: Thu, 9 Apr 2009 17:09:36 -0700
From: Andy Grover <andy.grover@...cle.com>
To: netdev@...r.kernel.org
Cc: rds-devel@....oracle.com, Steve Wise <swise@...ngridcomputing.com>
Subject: [PATCH 01/12] RDS/IW+IB: Set recv ring low water mark to 1/2 full.
From: Steve Wise <swise@...ngridcomputing.com>
Currently the recv ring low water mark is 1/4 the depth. Performance
measurements show that this limits iWARP throughput by flow controlling
the rds-stress senders. Setting it to 1/2 seems to max the T3
performance. I tried even higher levels but that didn't help and it
started to increase the rds thread cpu utilization.
Signed-off-by: Steve Wise <swise@...ngridcomputing.com>
Signed-off-by: Andy Grover <andy.grover@...cle.com>
---
net/rds/ib_ring.c | 2 +-
net/rds/iw_ring.c | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/net/rds/ib_ring.c b/net/rds/ib_ring.c
index 99a6cca..ff97e8e 100644
--- a/net/rds/ib_ring.c
+++ b/net/rds/ib_ring.c
@@ -137,7 +137,7 @@ int rds_ib_ring_empty(struct rds_ib_work_ring *ring)
int rds_ib_ring_low(struct rds_ib_work_ring *ring)
{
- return __rds_ib_ring_used(ring) <= (ring->w_nr >> 2);
+ return __rds_ib_ring_used(ring) <= (ring->w_nr >> 1);
}
/*
diff --git a/net/rds/iw_ring.c b/net/rds/iw_ring.c
index d422d4b..da8e3b6 100644
--- a/net/rds/iw_ring.c
+++ b/net/rds/iw_ring.c
@@ -137,7 +137,7 @@ int rds_iw_ring_empty(struct rds_iw_work_ring *ring)
int rds_iw_ring_low(struct rds_iw_work_ring *ring)
{
- return __rds_iw_ring_used(ring) <= (ring->w_nr >> 2);
+ return __rds_iw_ring_used(ring) <= (ring->w_nr >> 1);
}
--
1.5.6.3
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists