DPDK patches and discussions
 help / color / mirror / Atom feed
From: Thomas Monjalon <thomas@monjalon.net>
To: Morten Brørup <mb@smartsharesystems.com>
Cc: Ajit Khaparde <ajit.khaparde@broadcom.com>,
	"Ananyev, Konstantin" <konstantin.ananyev@intel.com>,
	Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>,
	dev@dpdk.org, "Yigit, Ferruh" <ferruh.yigit@intel.com>,
	david.marchand@redhat.com, "Richardson,
	Bruce" <bruce.richardson@intel.com>,
	olivier.matz@6wind.com, jerinj@marvell.com,
	viacheslavo@nvidia.com, honnappa.nagarahalli@arm.com,
	maxime.coquelin@redhat.com, stephen@networkplumber.org,
	hemant.agrawal@nxp.com, viacheslavo@nvidia.com
Subject: Re: [dpdk-dev] [PATCH 15/15] mbuf: move pool pointer in hotterfirst half
Date: Sat, 31 Oct 2020 21:40:46 +0100
Message-ID: <3458411.u7HY7OY5Un@thomas> (raw)
In-Reply-To: <98CBD80474FA8B44BF855DF32C47DC35C613CA@smartserver.smartshare.dk>

Thanks for the thoughts Morten.
I believe we need benchmarks of different scenarios with different drivers.


31/10/2020 19:20, Morten Brørup:
> Thomas,
> 
> Adding my thoughts to the already detailed feedback on this important patch...
> 
> The first cache line is not inherently "hotter" than the second. The hotness depends on their usage.
> 
> The mbuf cacheline1 marker has the following comment:
> /* second cache line - fields only used in slow path or on TX */
> 
> In other words, the second cache line is intended not to be touched in fast path RX.
> 
> I do not think this is true anymore. Not even with simple non-scattered RX. And regression testing probably didn't catch this, because the tests perform TX after RX, so the cache miss moved from TX to RX and became a cache hit in TX instead. (I may be wrong about this claim, but it's not important for the discussion.)
> 
> I think the right question for this patch is: Can we achieve this - not using the second cache line for fast path RX - again by putting the right fields in the first cache line?
> 
> Probably not in all cases, but perhaps for some...
> 
> Consider the application scenarios.
> 
> When a packet is received, one of three things happens to it:
> 1. It is immediately transmitted on one or more ports.
> 2. It is immediately discarded, e.g. by a firewall rule.
> 3. It is put in some sort of queue, e.g. a ring for the next pipeline stage, or in a QoS queue.
> 
> 1. If the packet is immediately transmitted, the m->tx_offload field in the second cache line will be touched by the application and TX function anyway, so we don't need to optimize the mbuf layout for this scenario.
> 
> 2. The second scenario touches m->pool no matter how it is implemented. The application can avoid touching m->next by using rte_mbuf_raw_free(), knowing that the mbuf came directly from RX and thus no other fields have been touched. In this scenario, we want m->pool in the first cache line.
> 
> 3. Now, let's consider the third scenario, where RX is followed by enqueue into a ring. If the application does nothing but put the packet into a ring, we don't need to move anything into the first cache line. But applications usually does more... So it is application specific what would be good to move to the first cache line:
> 
> A. If the application does not use segmented mbufs, and performs analysis and preparation for transmission in the initial pipeline stages, and only the last pipeline stage performs TX, we could move m->tx_offload to the first cache line, which would keep the second cache line cold until the actual TX happens in the last pipeline stage - maybe even after the packet has waited in a QoS queue for a long time, and its cache lines have gone cold.
> 
> B. If the application uses segmented mbufs on RX, it might make sense moving m->next to the first cache line. (We don't use segmented mbufs, so I'm not sure about this.)
> 
> 
> However, reality perhaps beats theory:
> 
> Looking at the E1000 PMD, it seems like even its non-scattered RX function, eth_igb_recv_pkts(), sets m->next. If it only kept its own free pool pre-initialized instead... I haven't investigated other PMDs, except briefly looking at the mlx5 PMD, and it seems like it doesn't touch m->next in RX.
> 
> I haven't looked deeper into how m->pool is being used by RX in PMDs, but I suppose that it isn't touched in RX.
> 
> <rant on>
> If only we had a performance test where RX was not immediately followed by TX, but the packets were passed through a large queue in-between, so RX cache misses were not free of charge because they transform TX cache misses into cache hits instead...
> <rant off>
> 
> Whatever you choose, I am sure that most applications will find it more useful than the timestamp. :-)




  reply	other threads:[~2020-10-31 20:41 UTC|newest]

Thread overview: 170+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-10-29  9:27 [dpdk-dev] [PATCH 00/15] remove mbuf timestamp Thomas Monjalon
2020-10-29  9:27 ` [dpdk-dev] [PATCH 01/15] eventdev: remove software Rx timestamp Thomas Monjalon
2020-10-29  9:27 ` [dpdk-dev] [PATCH 02/15] mbuf: add Rx timestamp dynamic flag Thomas Monjalon
2020-10-29  9:58   ` Andrew Rybchenko
2020-10-29 18:19     ` Ajit Khaparde
2020-10-29  9:27 ` [dpdk-dev] [PATCH 03/15] ethdev: register mbuf field and flags for timestamp Thomas Monjalon
2020-10-29 10:08   ` Andrew Rybchenko
2020-10-29 10:12     ` Thomas Monjalon
2020-10-29 10:33       ` Andrew Rybchenko
2020-10-29 10:46         ` Thomas Monjalon
2020-10-29  9:27 ` [dpdk-dev] [PATCH 04/15] latency: switch timestamp to dynamic mbuf field Thomas Monjalon
2020-10-29 10:13   ` Andrew Rybchenko
2020-10-29 10:40     ` Thomas Monjalon
2020-10-29 14:20   ` Pattan, Reshma
2020-10-29 16:15     ` Thomas Monjalon
2020-10-29  9:27 ` [dpdk-dev] [PATCH 05/15] net/ark: " Thomas Monjalon
2020-10-29  9:27 ` [dpdk-dev] [PATCH 06/15] net/dpaa2: " Thomas Monjalon
2020-10-29  9:27 ` [dpdk-dev] [PATCH 07/15] net/mlx5: fix dynamic mbuf offset lookup check Thomas Monjalon
2020-10-29  9:27 ` [dpdk-dev] [PATCH 08/15] net/mlx5: switch timestamp to dynamic mbuf field Thomas Monjalon
2020-10-29  9:27 ` [dpdk-dev] [PATCH 09/15] net/nfb: " Thomas Monjalon
2020-10-29  9:27 ` [dpdk-dev] [PATCH 10/15] net/octeontx2: " Thomas Monjalon
2020-10-29 11:02   ` Andrew Rybchenko
2020-10-29 11:34     ` Thomas Monjalon
2020-10-29 11:37       ` Andrew Rybchenko
2020-10-29 11:52     ` Slava Ovsiienko
2020-10-30 12:41       ` Jerin Jacob
2020-11-01 16:12         ` Thomas Monjalon
2020-11-01 20:00         ` Andrew Rybchenko
2020-10-29  9:27 ` [dpdk-dev] [PATCH 11/15] net/pcap: " Thomas Monjalon
2020-10-29  9:27 ` [dpdk-dev] [PATCH 12/15] app/testpmd: " Thomas Monjalon
2020-10-29 10:20   ` Andrew Rybchenko
2020-10-29 10:43     ` Thomas Monjalon
2020-10-29 10:52       ` Andrew Rybchenko
2020-10-29  9:27 ` [dpdk-dev] [PATCH 13/15] examples/rxtx_callbacks: switch timestamp to dynamic field Thomas Monjalon
2020-10-29 10:21   ` Andrew Rybchenko
2020-10-29 10:44     ` Thomas Monjalon
2020-10-29  9:27 ` [dpdk-dev] [PATCH 14/15] mbuf: remove deprecated timestamp field Thomas Monjalon
2020-10-29 10:23   ` Andrew Rybchenko
2020-10-29 18:18     ` Ajit Khaparde
2020-10-29 14:48   ` Kinsella, Ray
2020-10-29  9:27 ` [dpdk-dev] [PATCH 15/15] mbuf: move pool pointer in hotter first half Thomas Monjalon
2020-10-29 10:50   ` Andrew Rybchenko
2020-10-29 10:56     ` Thomas Monjalon
2020-10-29 14:15       ` Ananyev, Konstantin
2020-10-29 18:45         ` Ajit Khaparde
2020-10-31 18:20           ` [dpdk-dev] [PATCH 15/15] mbuf: move pool pointer in hotterfirst half Morten Brørup
2020-10-31 20:40             ` Thomas Monjalon [this message]
2020-11-01  9:12               ` Morten Brørup
2020-11-01 16:21                 ` Thomas Monjalon
2020-11-01 16:38                 ` Thomas Monjalon
2020-11-01 20:59                   ` Morten Brørup
2020-11-02 15:58                     ` Thomas Monjalon
2020-11-03 12:10                       ` Morten Brørup
2020-11-03 12:25                         ` Bruce Richardson
2020-11-03 13:46                           ` Morten Brørup
2020-11-03 13:50                             ` Bruce Richardson
2020-11-03 14:03                               ` Morten Brørup
2020-11-03 14:02                         ` Slava Ovsiienko
2020-11-03 15:03                           ` Morten Brørup
2020-11-04 15:00                             ` Olivier Matz
2020-11-05  0:25                               ` Ananyev, Konstantin
2020-11-05  9:04                                 ` Morten Brørup
2020-11-05  9:35                                 ` Morten Brørup
2020-11-05 10:29                                   ` Bruce Richardson
2020-10-29 14:42   ` [dpdk-dev] [PATCH 15/15] mbuf: move pool pointer in hotter first half Kinsella, Ray
2020-11-01 18:06 ` [dpdk-dev] [PATCH v2 00/14] remove mbuf timestamp Thomas Monjalon
2020-11-01 18:06   ` [dpdk-dev] [PATCH v2 01/14] eventdev: remove software Rx timestamp Thomas Monjalon
2020-11-01 18:06   ` [dpdk-dev] [PATCH v2 02/14] mbuf: add Rx timestamp dynamic flag Thomas Monjalon
2020-11-01 18:06   ` [dpdk-dev] [PATCH v2 03/14] ethdev: register mbuf field and flags for timestamp Thomas Monjalon
2020-11-02 15:39     ` Olivier Matz
2020-11-02 16:52       ` Thomas Monjalon
2020-11-01 18:06   ` [dpdk-dev] [PATCH v2 04/14] latency: switch timestamp to dynamic mbuf field Thomas Monjalon
2020-11-01 18:06   ` [dpdk-dev] [PATCH v2 05/14] net/ark: " Thomas Monjalon
2020-11-02 15:32     ` Olivier Matz
2020-11-02 16:10       ` Thomas Monjalon
2020-11-01 18:06   ` [dpdk-dev] [PATCH v2 06/14] net/dpaa2: " Thomas Monjalon
2020-11-01 18:06   ` [dpdk-dev] [PATCH v2 07/14] net/mlx5: fix dynamic mbuf offset lookup check Thomas Monjalon
2020-11-01 18:06   ` [dpdk-dev] [PATCH v2 08/14] net/mlx5: switch timestamp to dynamic mbuf field Thomas Monjalon
2020-11-02  5:08     ` Ruifeng Wang
2020-11-02 23:20     ` David Christensen
2020-11-01 18:06   ` [dpdk-dev] [PATCH v2 09/14] net/nfb: " Thomas Monjalon
2020-11-01 18:06   ` [dpdk-dev] [PATCH v2 10/14] net/octeontx2: " Thomas Monjalon
2020-11-01 18:28     ` Jerin Jacob
2020-11-02  9:38       ` Thomas Monjalon
2020-11-02 11:01         ` Thomas Monjalon
2020-11-01 18:06   ` [dpdk-dev] [PATCH v2 11/14] net/pcap: " Thomas Monjalon
2020-11-01 18:06   ` [dpdk-dev] [PATCH v2 12/14] app/testpmd: " Thomas Monjalon
2020-11-01 18:06   ` [dpdk-dev] [PATCH v2 13/14] examples/rxtx_callbacks: switch timestamp to dynamic field Thomas Monjalon
2020-11-01 18:06   ` [dpdk-dev] [PATCH v2 14/14] mbuf: remove deprecated timestamp field Thomas Monjalon
2020-11-02 15:41     ` Olivier Matz
2020-11-02 15:47       ` David Marchand
2020-11-02 15:49         ` Thomas Monjalon
2020-11-01 18:08   ` [dpdk-dev] [PATCH v2 00/14] remove mbuf timestamp Thomas Monjalon
2020-11-03  0:13 ` [dpdk-dev] [PATCH v3 00/16] " Thomas Monjalon
2020-11-03  0:13   ` [dpdk-dev] [PATCH v3 01/16] eventdev: remove software Rx timestamp Thomas Monjalon
2020-11-03  0:13   ` [dpdk-dev] [PATCH v3 02/16] mbuf: add Rx timestamp flag and helpers Thomas Monjalon
2020-11-03  9:33     ` Olivier Matz
2020-11-03  9:59       ` Thomas Monjalon
2020-11-03  0:13   ` [dpdk-dev] [PATCH v3 03/16] latency: switch Rx timestamp to dynamic mbuf field Thomas Monjalon
2020-11-03  0:13   ` [dpdk-dev] [PATCH v3 04/16] net/ark: " Thomas Monjalon
2020-11-03  0:13   ` [dpdk-dev] [PATCH v3 05/16] net/dpaa2: " Thomas Monjalon
2020-11-03  9:18     ` Hemant Agrawal
2020-11-03  0:13   ` [dpdk-dev] [PATCH v3 06/16] net/mlx5: fix dynamic mbuf offset lookup check Thomas Monjalon
2020-11-03  8:12     ` Slava Ovsiienko
2020-11-03  0:13   ` [dpdk-dev] [PATCH v3 07/16] net/mlx5: switch Rx timestamp to dynamic mbuf field Thomas Monjalon
2020-11-03  8:12     ` Slava Ovsiienko
2020-11-03  0:13   ` [dpdk-dev] [PATCH v3 08/16] net/nfb: " Thomas Monjalon
2020-11-03 10:20     ` Olivier Matz
2020-11-03  0:14   ` [dpdk-dev] [PATCH v3 09/16] net/octeontx2: " Thomas Monjalon
2020-11-03 10:52     ` Harman Kalra
2020-11-03 11:22       ` Thomas Monjalon
2020-11-03 12:21         ` Thomas Monjalon
2020-11-03 14:23           ` [dpdk-dev] [EXT] " Harman Kalra
2020-11-03  0:14   ` [dpdk-dev] [PATCH v3 10/16] net/pcap: " Thomas Monjalon
2020-11-03  0:14   ` [dpdk-dev] [PATCH v3 11/16] app/testpmd: " Thomas Monjalon
2020-11-03 10:23     ` Olivier Matz
2020-11-03  0:14   ` [dpdk-dev] [PATCH v3 12/16] examples/rxtx_callbacks: switch timestamp to dynamic field Thomas Monjalon
2020-11-03  0:14   ` [dpdk-dev] [PATCH v3 13/16] ethdev: add doxygen comment for Rx timestamp API Thomas Monjalon
2020-11-03  0:14   ` [dpdk-dev] [PATCH v3 14/16] mbuf: remove deprecated timestamp field Thomas Monjalon
2020-11-03  0:14   ` [dpdk-dev] [PATCH v3 15/16] mbuf: add Tx timestamp registration helper Thomas Monjalon
2020-11-03  0:14   ` [dpdk-dev] [PATCH v3 16/16] ethdev: include mbuf registration in Tx timestamp API Thomas Monjalon
2020-11-03  7:54     ` Slava Ovsiienko
2020-11-03  9:00   ` [dpdk-dev] [PATCH v3 00/16] remove mbuf timestamp David Marchand
2020-11-03 12:21 ` [dpdk-dev] [PATCH v4 " Thomas Monjalon
2020-11-03 12:21   ` [dpdk-dev] [PATCH v4 01/16] eventdev: remove software Rx timestamp Thomas Monjalon
2020-11-03 12:21   ` [dpdk-dev] [PATCH v4 02/16] mbuf: add Rx timestamp flag and helpers Thomas Monjalon
2020-11-03 12:34     ` Andrew Rybchenko
2020-11-03 12:21   ` [dpdk-dev] [PATCH v4 03/16] latency: switch Rx timestamp to dynamic mbuf field Thomas Monjalon
2020-11-03 12:21   ` [dpdk-dev] [PATCH v4 04/16] net/ark: " Thomas Monjalon
2020-11-03 12:37     ` Andrew Rybchenko
2020-11-03 13:08       ` Thomas Monjalon
2020-11-03 12:21   ` [dpdk-dev] [PATCH v4 05/16] net/dpaa2: " Thomas Monjalon
2020-11-03 12:21   ` [dpdk-dev] [PATCH v4 06/16] net/mlx5: fix dynamic mbuf offset lookup check Thomas Monjalon
2020-11-03 12:21   ` [dpdk-dev] [PATCH v4 07/16] net/mlx5: switch Rx timestamp to dynamic mbuf field Thomas Monjalon
2020-11-03 12:21   ` [dpdk-dev] [PATCH v4 08/16] net/nfb: " Thomas Monjalon
2020-11-03 12:21   ` [dpdk-dev] [PATCH v4 09/16] net/octeontx2: " Thomas Monjalon
2020-11-03 12:21   ` [dpdk-dev] [PATCH v4 10/16] net/pcap: " Thomas Monjalon
2020-11-03 12:22   ` [dpdk-dev] [PATCH v4 11/16] app/testpmd: " Thomas Monjalon
2020-11-03 12:22   ` [dpdk-dev] [PATCH v4 12/16] examples/rxtx_callbacks: switch timestamp to dynamic field Thomas Monjalon
2020-11-03 12:22   ` [dpdk-dev] [PATCH v4 13/16] ethdev: add doxygen comment for Rx timestamp API Thomas Monjalon
2020-11-03 12:40     ` Andrew Rybchenko
2020-11-03 12:22   ` [dpdk-dev] [PATCH v4 14/16] mbuf: remove deprecated timestamp field Thomas Monjalon
2020-11-03 12:22   ` [dpdk-dev] [PATCH v4 15/16] mbuf: add Tx timestamp registration helper Thomas Monjalon
2020-11-03 12:42     ` Andrew Rybchenko
2020-11-03 12:22   ` [dpdk-dev] [PATCH v4 16/16] ethdev: include mbuf registration in Tx timestamp API Thomas Monjalon
2020-11-03 12:45     ` Andrew Rybchenko
2020-11-03 14:09 ` [dpdk-dev] [PATCH v5 00/16] remove mbuf timestamp Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 01/16] eventdev: remove software Rx timestamp Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 02/16] mbuf: add Rx timestamp flag and helpers Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 03/16] latency: switch Rx timestamp to dynamic mbuf field Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 04/16] net/ark: " Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 05/16] net/dpaa2: " Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 06/16] net/mlx5: fix dynamic mbuf offset lookup check Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 07/16] net/mlx5: switch Rx timestamp to dynamic mbuf field Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 08/16] net/nfb: " Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 09/16] net/octeontx2: " Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 10/16] net/pcap: " Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 11/16] app/testpmd: " Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 12/16] examples/rxtx_callbacks: switch timestamp to dynamic field Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 13/16] ethdev: add doxygen comment for Rx timestamp API Thomas Monjalon
2020-11-03 19:07     ` Ajit Khaparde
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 14/16] mbuf: remove deprecated timestamp field Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 15/16] mbuf: add Tx timestamp registration helper Thomas Monjalon
2020-11-03 14:09   ` [dpdk-dev] [PATCH v5 16/16] ethdev: include mbuf registration in Tx timestamp API Thomas Monjalon
2020-11-03 14:17   ` [dpdk-dev] [PATCH v5 00/16] remove mbuf timestamp Olivier Matz
2020-11-03 14:44     ` Thomas Monjalon
2020-11-03 16:08   ` Stephen Hemminger
2020-11-03 16:20     ` Thomas Monjalon
2020-11-03 17:42       ` Stephen Hemminger
2020-11-03 17:55         ` Thomas Monjalon

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=3458411.u7HY7OY5Un@thomas \
    --to=thomas@monjalon.net \
    --cc=ajit.khaparde@broadcom.com \
    --cc=andrew.rybchenko@oktetlabs.ru \
    --cc=bruce.richardson@intel.com \
    --cc=david.marchand@redhat.com \
    --cc=dev@dpdk.org \
    --cc=ferruh.yigit@intel.com \
    --cc=hemant.agrawal@nxp.com \
    --cc=honnappa.nagarahalli@arm.com \
    --cc=jerinj@marvell.com \
    --cc=konstantin.ananyev@intel.com \
    --cc=maxime.coquelin@redhat.com \
    --cc=mb@smartsharesystems.com \
    --cc=olivier.matz@6wind.com \
    --cc=stephen@networkplumber.org \
    --cc=viacheslavo@nvidia.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

DPDK patches and discussions

This inbox may be cloned and mirrored by anyone:

	git clone --mirror https://inbox.dpdk.org/dev/0 dev/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 dev dev/ https://inbox.dpdk.org/dev \
		dev@dpdk.org
	public-inbox-index dev

Example config snippet for mirrors.
Newsgroup available over NNTP:
	nntp://inbox.dpdk.org/inbox.dpdk.dev


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git