This commit adds the multiple-thread flow insertion optimization description.
Signed-off-by: Suanming Mou <suanmi...@nvidia.com> Acked-by: Matan Azrad <ma...@nvidia.com> --- doc/guides/nics/mlx5.rst | 5 +++++ doc/guides/rel_notes/release_21_08.rst | 1 + 2 files changed, 6 insertions(+) diff --git a/doc/guides/nics/mlx5.rst b/doc/guides/nics/mlx5.rst index 5a3c4f198a..199beb2549 100644 --- a/doc/guides/nics/mlx5.rst +++ b/doc/guides/nics/mlx5.rst @@ -448,6 +448,11 @@ Limitations - 256 ports maximum. - 4M connections maximum. +- Multiple-thread flow insertion: + + - In order to achieve best insertion rate, application should manage the flows on the rte-lcore. + - Better to configure ``reclaim_mem_mode`` as 0 to accelerate the flow object allocate and release with cache. + Statistics ---------- diff --git a/doc/guides/rel_notes/release_21_08.rst b/doc/guides/rel_notes/release_21_08.rst index 6a902ef9ac..68496496e8 100644 --- a/doc/guides/rel_notes/release_21_08.rst +++ b/doc/guides/rel_notes/release_21_08.rst @@ -88,6 +88,7 @@ New Features * Added support for meter hierarchy. * Added devargs options ``allow_duplicate_pattern``. + * Optimize multiple-thread flow insertion rate. * **Added support for Marvell CNXK crypto driver.** -- 2.25.1
