[dpdk-dev] [PATCH] net/mlx5: fix legacy non-inline multi-packet performance

[Viacheslav Ovsiienko <viacheslavo@mellanox.com>]

The legacy multi-packet write is the feature allowing to
put multiple packets into one transmitting descriptor,
this feature is supported by only NIC ConnectX-4LX.
The number of packets should be limited to provide
optimal size descriptor and better performance.

Fixes: 18a1c20044c0 ("net/mlx5: implement Tx burst template")

Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
---
 drivers/net/mlx5/mlx5_prm.h  | 2 ++
 drivers/net/mlx5/mlx5_rxtx.c | 4 +++-
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/drivers/net/mlx5/mlx5_prm.h b/drivers/net/mlx5/mlx5_prm.h
index 651006b..2bac6a7 100644
--- a/drivers/net/mlx5/mlx5_prm.h
+++ b/drivers/net/mlx5/mlx5_prm.h
@@ -99,6 +99,8 @@
  * in tx burst routine at the moment of freeing multiple mbufs.
  */
 #define MLX5_EMPW_MAX_PACKETS MLX5_TX_COMP_THRESH
+#define MLX5_MPW_MAX_PACKETS 6
+
 /*
  * Default packet length threshold to be inlined with
  * ordinary SEND. Inlining saves the MR key search
diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c
index 73c16de..25d915c 100644
--- a/drivers/net/mlx5/mlx5_rxtx.c
+++ b/drivers/net/mlx5/mlx5_rxtx.c
@@ -3888,7 +3888,9 @@ enum mlx5_txcmp_code {
 
 next_empw:
 		assert(NB_SEGS(loc->mbuf) == 1);
-		part = RTE_MIN(pkts_n, MLX5_EMPW_MAX_PACKETS);
+		part = RTE_MIN(pkts_n, MLX5_TXOFF_CONFIG(MPW) ?
+				       MLX5_MPW_MAX_PACKETS :
+				       MLX5_EMPW_MAX_PACKETS);
 		if (unlikely(loc->elts_free < part)) {
 			/* We have no enough elts to save all mbufs. */
 			if (unlikely(loc->elts_free < MLX5_EMPW_MIN_PACKETS))
-- 
1.8.3.1