diff options
| author | Eli Cohen <eli@dev.mellanox.co.il> | 2008-07-15 08:48:52 +0200 |
|---|---|---|
| committer | Roland Dreier <rolandd@cisco.com> | 2008-07-15 08:48:52 +0200 |
| commit | bc3a290b51aaefc6a6af2d6e6d52ed32387c416c (patch) | |
| tree | 94b3358177c589796224c818b80ca4fdc0944719 /drivers/infiniband | |
| parent | IPoIB/cm: Reduce connected mode TX object size (diff) | |
| download | linux-bc3a290b51aaefc6a6af2d6e6d52ed32387c416c.tar.xz linux-bc3a290b51aaefc6a6af2d6e6d52ed32387c416c.zip | |
IPoIB: Double default RX/TX ring sizes
Increase IPoIB ring sizes to twice their original sizes (RX: 128->256,
TX: 64->128) to act as a shock absorber for high traffic peaks. With
the current settings, we have seen cases that there are many calls to
netif_stop_queue(), which causes degradation in throughput. Also,
larger receive buffer sizes help IPoIB in CM mode to avoid experiencing
RNR NAK conditions due to insufficient receive buffers at the SRQ.
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Diffstat (limited to 'drivers/infiniband')
| -rw-r--r-- | drivers/infiniband/ulp/ipoib/ipoib.h | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/drivers/infiniband/ulp/ipoib/ipoib.h b/drivers/infiniband/ulp/ipoib/ipoib.h index 0281c8fecc90..b0ffc9abe8c0 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib.h +++ b/drivers/infiniband/ulp/ipoib/ipoib.h @@ -70,8 +70,8 @@ enum { IPOIB_CM_BUF_SIZE = IPOIB_CM_MTU + IPOIB_ENCAP_LEN, IPOIB_CM_HEAD_SIZE = IPOIB_CM_BUF_SIZE % PAGE_SIZE, IPOIB_CM_RX_SG = ALIGN(IPOIB_CM_BUF_SIZE, PAGE_SIZE) / PAGE_SIZE, - IPOIB_RX_RING_SIZE = 128, - IPOIB_TX_RING_SIZE = 64, + IPOIB_RX_RING_SIZE = 256, + IPOIB_TX_RING_SIZE = 128, IPOIB_MAX_QUEUE_SIZE = 8192, IPOIB_MIN_QUEUE_SIZE = 2, IPOIB_CM_MAX_CONN_QP = 4096, |