1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright(c) 2016 Intel Corporation.
7 The i40e PMD (librte_pmd_i40e) provides poll mode driver support for
8 10/25/40 Gbps Intel® Ethernet 700 Series Network Adapters based on
9 the Intel Ethernet Controller X710/XL710/XXV710 and Intel Ethernet
10 Connection X722 (only support part of features).
16 Features of the i40e PMD are:
18 - Multiple queues for TX and RX
19 - Receiver Side Scaling (RSS)
21 - Packet type information
25 - VLAN/QinQ stripping and inserting
29 - Port hardware statistics
31 - Link state information
33 - Mirror on port, VLAN and VSI
34 - Interrupt mode for RX
35 - Scattered and gather for TX and RX
36 - Vector Poll mode driver
41 - IEEE1588/802.1AS timestamping
42 - VF Daemon (VFD) - EXPERIMENTAL
43 - Dynamic Device Personalization (DDP)
44 - Queue region configuration
45 - Virtual Function Port Representors
50 - Identifying your adapter using `Intel Support
51 <http://www.intel.com/support>`_ and get the latest NVM/FW images.
53 - Follow the DPDK :ref:`Getting Started Guide for Linux <linux_gsg>` to setup the basic DPDK environment.
55 - To get better performance on Intel platforms, please follow the "How to get best performance with NICs on Intel platforms"
56 section of the :ref:`Getting Started Guide for Linux <linux_gsg>`.
58 - Upgrade the NVM/FW version following the `Intel® Ethernet NVM Update Tool Quick Usage Guide for Linux
59 <https://www-ssl.intel.com/content/www/us/en/embedded/products/networking/nvm-update-tool-quick-linux-usage-guide.html>`_ and `Intel® Ethernet NVM Update Tool: Quick Usage Guide for EFI <https://www.intel.com/content/www/us/en/embedded/products/networking/nvm-update-tool-quick-efi-usage-guide.html>`_ if needed.
61 Recommended Matching List
62 -------------------------
64 It is highly recommended to upgrade the i40e kernel driver and firmware to
65 avoid the compatibility issues with i40e PMD. Here is the suggested matching
66 list which has been tested and verified. The detailed information can refer
67 to chapter Tested Platforms/Tested NICs in release notes.
69 +--------------+-----------------------+------------------+
70 | DPDK version | Kernel driver version | Firmware version |
71 +==============+=======================+==================+
72 | 19.11 | 2.9.21 | 7.00 |
73 +--------------+-----------------------+------------------+
74 | 19.08 | 2.8.43 | 7.00 |
75 +--------------+-----------------------+------------------+
76 | 19.05 | 2.7.29 | 6.80 |
77 +--------------+-----------------------+------------------+
78 | 19.02 | 2.7.26 | 6.80 |
79 +--------------+-----------------------+------------------+
80 | 18.11 | 2.4.6 | 6.01 |
81 +--------------+-----------------------+------------------+
82 | 18.08 | 2.4.6 | 6.01 |
83 +--------------+-----------------------+------------------+
84 | 18.05 | 2.4.6 | 6.01 |
85 +--------------+-----------------------+------------------+
86 | 18.02 | 2.4.3 | 6.01 |
87 +--------------+-----------------------+------------------+
88 | 17.11 | 2.1.26 | 6.01 |
89 +--------------+-----------------------+------------------+
90 | 17.08 | 2.0.19 | 6.01 |
91 +--------------+-----------------------+------------------+
92 | 17.05 | 1.5.23 | 5.05 |
93 +--------------+-----------------------+------------------+
94 | 17.02 | 1.5.23 | 5.05 |
95 +--------------+-----------------------+------------------+
96 | 16.11 | 1.5.23 | 5.05 |
97 +--------------+-----------------------+------------------+
98 | 16.07 | 1.4.25 | 5.04 |
99 +--------------+-----------------------+------------------+
100 | 16.04 | 1.4.25 | 5.02 |
101 +--------------+-----------------------+------------------+
103 Pre-Installation Configuration
104 ------------------------------
109 The following options can be modified in the ``config`` file.
110 Please note that enabling debugging options may affect system performance.
112 - ``CONFIG_RTE_LIBRTE_I40E_PMD`` (default ``y``)
114 Toggle compilation of the ``librte_pmd_i40e`` driver.
116 - ``CONFIG_RTE_LIBRTE_I40E_DEBUG_*`` (default ``n``)
118 Toggle display of generic debugging messages.
120 - ``CONFIG_RTE_LIBRTE_I40E_RX_ALLOW_BULK_ALLOC`` (default ``y``)
122 Toggle bulk allocation for RX.
124 - ``CONFIG_RTE_LIBRTE_I40E_INC_VECTOR`` (default ``n``)
126 Toggle the use of Vector PMD instead of normal RX/TX path.
127 To enable vPMD for RX, bulk allocation for Rx must be allowed.
129 - ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC`` (default ``n``)
131 Toggle to use a 16-byte RX descriptor, by default the RX descriptor is 32 byte.
133 - ``CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_PF`` (default ``64``)
135 Number of queues reserved for PF.
137 - ``CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_VM`` (default ``4``)
139 Number of queues reserved for each VMDQ Pool.
141 Runtime Config Options
142 ~~~~~~~~~~~~~~~~~~~~~~
144 - ``Reserved number of Queues per VF`` (default ``4``)
146 The number of reserved queue per VF is determined by its host PF. If the
147 PCI address of an i40e PF is aaaa:bb.cc, the number of reserved queues per
148 VF can be configured with EAL parameter like -w aaaa:bb.cc,queue-num-per-vf=n.
149 The value n can be 1, 2, 4, 8 or 16. If no such parameter is configured, the
150 number of reserved queues per VF is 4 by default. If VF request more than
151 reserved queues per VF, PF will able to allocate max to 16 queues after a VF
155 - ``Support multiple driver`` (default ``disable``)
157 There was a multiple driver support issue during use of 700 series Ethernet
158 Adapter with both Linux kernel and DPDK PMD. To fix this issue, ``devargs``
159 parameter ``support-multi-driver`` is introduced, for example::
161 -w 84:00.0,support-multi-driver=1
163 With the above configuration, DPDK PMD will not change global registers, and
164 will switch PF interrupt from IntN to Int0 to avoid interrupt conflict between
165 DPDK and Linux Kernel.
167 - ``Support VF Port Representor`` (default ``not enabled``)
169 The i40e PF PMD supports the creation of VF port representors for the control
170 and monitoring of i40e virtual function devices. Each port representor
171 corresponds to a single virtual function of that device. Using the ``devargs``
172 option ``representor`` the user can specify which virtual functions to create
173 port representors for on initialization of the PF PMD by passing the VF IDs of
174 the VFs which are required.::
176 -w DBDF,representor=[0,1,4]
178 Currently hot-plugging of representor ports is not supported so all required
179 representors must be specified on the creation of the PF.
181 - ``Use latest supported vector`` (default ``disable``)
183 Latest supported vector path may not always get the best perf so vector path was
184 recommended to use only on later platform. But users may want the latest vector path
185 since it can get better perf in some real work loading cases. So ``devargs`` param
186 ``use-latest-supported-vec`` is introduced, for example::
188 -w 84:00.0,use-latest-supported-vec=1
190 - ``Enable validation for VF message`` (default ``not enabled``)
192 The PF counts messages from each VF. If in any period of seconds the message
193 statistic from a VF exceeds maximal limitation, the PF will ignore any new message
194 from that VF for some seconds.
195 Format -- "maximal-message@period-seconds:ignore-seconds"
198 -w 84:00.0,vf_msg_cfg=80@120:180
200 Vector RX Pre-conditions
201 ~~~~~~~~~~~~~~~~~~~~~~~~
202 For Vector RX it is assumed that the number of descriptor rings will be a power
203 of 2. With this pre-condition, the ring pointer can easily scroll back to the
204 head after hitting the tail without a conditional check. In addition Vector RX
205 can use this assumption to do a bit mask using ``ring_size - 1``.
207 Driver compilation and testing
208 ------------------------------
210 Refer to the document :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
214 SR-IOV: Prerequisites and sample Application Notes
215 --------------------------------------------------
217 #. Load the kernel module:
219 .. code-block:: console
223 Check the output in dmesg:
225 .. code-block:: console
227 i40e 0000:83:00.1 ens802f0: renamed from eth0
229 #. Bring up the PF ports:
231 .. code-block:: console
235 #. Create VF device(s):
237 Echo the number of VFs to be created into the ``sriov_numvfs`` sysfs entry
242 .. code-block:: console
244 echo 2 > /sys/devices/pci0000:00/0000:00:03.0/0000:81:00.0/sriov_numvfs
247 #. Assign VF MAC address:
249 Assign MAC address to the VF using iproute2 utility. The syntax is:
251 .. code-block:: console
253 ip link set <PF netdev id> vf <VF id> mac <macaddr>
257 .. code-block:: console
259 ip link set ens802f0 vf 0 mac a0:b0:c0:d0:e0:f0
261 #. Assign VF to VM, and bring up the VM.
262 Please see the documentation for the *I40E/IXGBE/IGB Virtual Function Driver*.
266 Follow instructions available in the document
267 :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
272 .. code-block:: console
275 EAL: PCI device 0000:83:00.0 on NUMA socket 1
276 EAL: probe driver: 8086:1572 rte_i40e_pmd
277 EAL: PCI memory mapped at 0x7f7f80000000
278 EAL: PCI memory mapped at 0x7f7f80800000
279 PMD: eth_i40e_dev_init(): FW 5.0 API 1.5 NVM 05.00.02 eetrack 8000208a
280 Interactive-mode selected
281 Configuring Port 0 (socket 0)
284 PMD: i40e_dev_rx_queue_setup(): Rx Burst Bulk Alloc Preconditions are
285 satisfied.Rx Burst Bulk Alloc function will be used on port=0, queue=0.
288 Port 0: 68:05:CA:26:85:84
289 Checking link statuses...
290 Port 0 Link Up - speed 10000 Mbps - full-duplex
296 Sample Application Notes
297 ------------------------
302 Vlan filter only works when Promiscuous mode is off.
304 To start ``testpmd``, and add vlan 10 to port 0:
306 .. code-block:: console
308 ./app/testpmd -l 0-15 -n 4 -- -i --forward-mode=mac
311 testpmd> set promisc 0 off
312 testpmd> rx_vlan add 10 0
318 The Flow Director works in receive mode to identify specific flows or sets of flows and route them to specific queues.
319 The Flow Director filters can match the different fields for different type of packet: flow type, specific input set per flow type and the flexible payload.
321 The default input set of each flow type is::
323 ipv4-other : src_ip_address, dst_ip_address
324 ipv4-frag : src_ip_address, dst_ip_address
325 ipv4-tcp : src_ip_address, dst_ip_address, src_port, dst_port
326 ipv4-udp : src_ip_address, dst_ip_address, src_port, dst_port
327 ipv4-sctp : src_ip_address, dst_ip_address, src_port, dst_port,
329 ipv6-other : src_ip_address, dst_ip_address
330 ipv6-frag : src_ip_address, dst_ip_address
331 ipv6-tcp : src_ip_address, dst_ip_address, src_port, dst_port
332 ipv6-udp : src_ip_address, dst_ip_address, src_port, dst_port
333 ipv6-sctp : src_ip_address, dst_ip_address, src_port, dst_port,
335 l2_payload : ether_type
337 The flex payload is selected from offset 0 to 15 of packet's payload by default, while it is masked out from matching.
339 Start ``testpmd`` with ``--disable-rss`` and ``--pkt-filter-mode=perfect``:
341 .. code-block:: console
343 ./app/testpmd -l 0-15 -n 4 -- -i --disable-rss --pkt-filter-mode=perfect \
344 --rxq=8 --txq=8 --nb-cores=8 --nb-ports=1
346 Add a rule to direct ``ipv4-udp`` packet whose ``dst_ip=2.2.2.5, src_ip=2.2.2.3, src_port=32, dst_port=32`` to queue 1:
348 .. code-block:: console
350 testpmd> flow_director_filter 0 mode IP add flow ipv4-udp \
351 src 2.2.2.3 32 dst 2.2.2.5 32 vlan 0 flexbytes () \
352 fwd pf queue 1 fd_id 1
354 Check the flow director status:
356 .. code-block:: console
358 testpmd> show port fdir 0
360 ######################## FDIR infos for port 0 ####################
362 SUPPORTED FLOW TYPE: ipv4-frag ipv4-tcp ipv4-udp ipv4-sctp ipv4-other
363 ipv6-frag ipv6-tcp ipv6-udp ipv6-sctp ipv6-other
366 max_len: 16 payload_limit: 480
367 payload_unit: 2 payload_seg: 3
368 bitmask_unit: 2 bitmask_num: 2
371 src_ipv4: 0x00000000,
372 dst_ipv4: 0x00000000,
375 src_ipv6: 0x00000000,0x00000000,0x00000000,0x00000000,
376 dst_ipv6: 0x00000000,0x00000000,0x00000000,0x00000000
377 FLEX PAYLOAD SRC OFFSET:
378 L2_PAYLOAD: 0 1 2 3 4 5 6 ...
379 L3_PAYLOAD: 0 1 2 3 4 5 6 ...
380 L4_PAYLOAD: 0 1 2 3 4 5 6 ...
382 ipv4-udp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
383 ipv4-tcp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
384 ipv4-sctp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
385 ipv4-other: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
386 ipv4-frag: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
387 ipv6-udp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
388 ipv6-tcp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
389 ipv6-sctp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
390 ipv6-other: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
391 ipv6-frag: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
392 l2_payload: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
393 guarant_count: 1 best_count: 0
394 guarant_space: 512 best_space: 7168
401 Delete all flow director rules on a port:
403 .. code-block:: console
405 testpmd> flush_flow_director 0
410 The Intel® Ethernet 700 Series support a feature called
413 A Virtual Ethernet Bridge (VEB) is an IEEE Edge Virtual Bridging (EVB) term
414 for functionality that allows local switching between virtual endpoints within
415 a physical endpoint and also with an external bridge/network.
417 A "Floating" VEB doesn't have an uplink connection to the outside world so all
418 switching is done internally and remains within the host. As such, this
419 feature provides security benefits.
421 In addition, a Floating VEB overcomes a limitation of normal VEBs where they
422 cannot forward packets when the physical link is down. Floating VEBs don't need
423 to connect to the NIC port so they can still forward traffic from VF to VF
424 even when the physical link is down.
426 Therefore, with this feature enabled VFs can be limited to communicating with
427 each other but not an outside network, and they can do so even when there is
428 no physical uplink on the associated NIC port.
430 To enable this feature, the user should pass a ``devargs`` parameter to the
433 -w 84:00.0,enable_floating_veb=1
435 In this configuration the PMD will use the floating VEB feature for all the
436 VFs created by this PF device.
438 Alternatively, the user can specify which VFs need to connect to this floating
439 VEB using the ``floating_veb_list`` argument::
441 -w 84:00.0,enable_floating_veb=1,floating_veb_list=1;3-4
443 In this example ``VF1``, ``VF3`` and ``VF4`` connect to the floating VEB,
444 while other VFs connect to the normal VEB.
446 The current implementation only supports one floating VEB and one regular
447 VEB. VFs can connect to a floating VEB or a regular VEB according to the
448 configuration passed on the EAL command line.
450 The floating VEB functionality requires a NIC firmware version of 5.0
453 Dynamic Device Personalization (DDP)
454 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
456 The Intel® Ethernet 700 Series except for the Intel Ethernet Connection
457 X722 support a feature called "Dynamic Device Personalization (DDP)",
458 which is used to configure hardware by downloading a profile to support
459 protocols/filters which are not supported by default. The DDP
460 functionality requires a NIC firmware version of 6.0 or greater.
462 Current implementation supports GTP-C/GTP-U/PPPoE/PPPoL2TP,
463 steering can be used with rte_flow API.
465 GTPv1 package is released, and it can be downloaded from
466 https://downloadcenter.intel.com/download/27587.
468 PPPoE package is released, and it can be downloaded from
469 https://downloadcenter.intel.com/download/28040.
471 Load a profile which supports GTP and store backup profile:
473 .. code-block:: console
475 testpmd> ddp add 0 ./gtp.pkgo,./backup.pkgo
477 Delete a GTP profile and restore backup profile:
479 .. code-block:: console
481 testpmd> ddp del 0 ./backup.pkgo
483 Get loaded DDP package info list:
485 .. code-block:: console
487 testpmd> ddp get list 0
489 Display information about a GTP profile:
491 .. code-block:: console
493 testpmd> ddp get info ./gtp.pkgo
495 Input set configuration
496 ~~~~~~~~~~~~~~~~~~~~~~~
497 Input set for any PCTYPE can be configured with user defined configuration,
498 For example, to use only 48bit prefix for IPv6 src address for IPv6 TCP RSS:
500 .. code-block:: console
502 testpmd> port config 0 pctype 43 hash_inset clear all
503 testpmd> port config 0 pctype 43 hash_inset set field 13
504 testpmd> port config 0 pctype 43 hash_inset set field 14
505 testpmd> port config 0 pctype 43 hash_inset set field 15
507 Queue region configuration
508 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
509 The Intel® Ethernet 700 Series supports a feature of queue regions
510 configuration for RSS in the PF, so that different traffic classes or
511 different packet classification types can be separated to different
512 queues in different queue regions. There is an API for configuration
513 of queue regions in RSS with a command line. It can parse the parameters
514 of the region index, queue number, queue start index, user priority, traffic
515 classes and so on. Depending on commands from the command line, it will call
516 i40e private APIs and start the process of setting or flushing the queue
517 region configuration. As this feature is specific for i40e only private
518 APIs are used. These new ``test_pmd`` commands are as shown below. For
519 details please refer to :doc:`../testpmd_app_ug/index`.
521 .. code-block:: console
523 testpmd> set port (port_id) queue-region region_id (value) \
524 queue_start_index (value) queue_num (value)
525 testpmd> set port (port_id) queue-region region_id (value) flowtype (value)
526 testpmd> set port (port_id) queue-region UP (value) region_id (value)
527 testpmd> set port (port_id) queue-region flush (on|off)
528 testpmd> show port (port_id) queue-region
530 Limitations or Known issues
531 ---------------------------
533 MPLS packet classification
534 ~~~~~~~~~~~~~~~~~~~~~~~~~~
536 For firmware versions prior to 5.0, MPLS packets are not recognized by the NIC.
537 The L2 Payload flow type in flow director can be used to classify MPLS packet
538 by using a command in testpmd like:
540 testpmd> flow_director_filter 0 mode IP add flow l2_payload ether \
541 0x8847 flexbytes () fwd pf queue <N> fd_id <M>
543 With the NIC firmware version 5.0 or greater, some limited MPLS support
544 is added: Native MPLS (MPLS in Ethernet) skip is implemented, while no
545 new packet type, no classification or offload are possible. With this change,
546 L2 Payload flow type in flow director cannot be used to classify MPLS packet
547 as with previous firmware versions. Meanwhile, the Ethertype filter can be
548 used to classify MPLS packet by using a command in testpmd like:
550 testpmd> ethertype_filter 0 add mac_ignr 00:00:00:00:00:00 ethertype \
553 16 Byte RX Descriptor setting on DPDK VF
554 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
556 Currently the VF's RX descriptor mode is decided by PF. There's no PF-VF
557 interface for VF to request the RX descriptor mode, also no interface to notify
558 VF its own RX descriptor mode.
559 For all available versions of the i40e driver, these drivers don't support 16
560 byte RX descriptor. If the Linux i40e kernel driver is used as host driver,
561 while DPDK i40e PMD is used as the VF driver, DPDK cannot choose 16 byte receive
562 descriptor. The reason is that the RX descriptor is already set to 32 byte by
563 the i40e kernel driver. That is to say, user should keep
564 ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC=n`` in config file.
565 In the future, if the Linux i40e driver supports 16 byte RX descriptor, user
566 should make sure the DPDK VF uses the same RX descriptor mode, 16 byte or 32
567 byte, as the PF driver.
569 The same rule for DPDK PF + DPDK VF. The PF and VF should use the same RX
570 descriptor mode. Or the VF RX will not work.
572 Receive packets with Ethertype 0x88A8
573 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
575 Due to the FW limitation, PF can receive packets with Ethertype 0x88A8
576 only when floating VEB is disabled.
578 Incorrect Rx statistics when packet is oversize
579 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
581 When a packet is over maximum frame size, the packet is dropped.
582 However, the Rx statistics, when calling `rte_eth_stats_get` incorrectly
583 shows it as received.
585 VF & TC max bandwidth setting
586 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
588 The per VF max bandwidth and per TC max bandwidth cannot be enabled in parallel.
589 The behavior is different when handling per VF and per TC max bandwidth setting.
590 When enabling per VF max bandwidth, SW will check if per TC max bandwidth is
591 enabled. If so, return failure.
592 When enabling per TC max bandwidth, SW will check if per VF max bandwidth
593 is enabled. If so, disable per VF max bandwidth and continue with per TC max
596 TC TX scheduling mode setting
597 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
599 There are 2 TX scheduling modes for TCs, round robin and strict priority mode.
600 If a TC is set to strict priority mode, it can consume unlimited bandwidth.
601 It means if APP has set the max bandwidth for that TC, it comes to no
603 It's suggested to set the strict priority mode for a TC that is latency
604 sensitive but no consuming much bandwidth.
606 VF performance is impacted by PCI extended tag setting
607 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
609 To reach maximum NIC performance in the VF the PCI extended tag must be
610 enabled. The DPDK i40e PF driver will set this feature during initialization,
611 but the kernel PF driver does not. So when running traffic on a VF which is
612 managed by the kernel PF driver, a significant NIC performance downgrade has
613 been observed (for 64 byte packets, there is about 25% line-rate downgrade for
614 a 25GbE device and about 35% for a 40GbE device).
616 For kernel version >= 4.11, the kernel's PCI driver will enable the extended
617 tag if it detects that the device supports it. So by default, this is not an
618 issue. For kernels <= 4.11 or when the PCI extended tag is disabled it can be
619 enabled using the steps below.
621 #. Get the current value of the PCI configure register::
623 setpci -s <XX:XX.X> a8.w
627 value = value | 0x100
629 #. Set the PCI configure register with new value::
631 setpci -s <XX:XX.X> a8.w=<value>
636 The VF vlan strip function is only supported in the i40e kernel driver >= 2.1.26.
641 DCB works only when RSS is enabled.
643 Global configuration warning
644 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
646 I40E PMD will set some global registers to enable some function or set some
647 configure. Then when using different ports of the same NIC with Linux kernel
648 and DPDK, the port with Linux kernel will be impacted by the port with DPDK.
649 For example, register I40E_GL_SWT_L2TAGCTRL is used to control L2 tag, i40e
650 PMD uses I40E_GL_SWT_L2TAGCTRL to set vlan TPID. If setting TPID in port A
651 with DPDK, then the configuration will also impact port B in the NIC with
652 kernel driver, which don't want to use the TPID.
653 So PMD reports warning to clarify what is changed by writing global register.
655 High Performance of Small Packets on 40GbE NIC
656 ----------------------------------------------
658 As there might be firmware fixes for performance enhancement in latest version
659 of firmware image, the firmware update might be needed for getting high performance.
660 Check the Intel support website for the latest firmware updates.
661 Users should consult the release notes specific to a DPDK release to identify
662 the validated firmware version for a NIC using the i40e driver.
664 Use 16 Bytes RX Descriptor Size
665 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
667 As i40e PMD supports both 16 and 32 bytes RX descriptor sizes, and 16 bytes size can provide helps to high performance of small packets.
668 Configuration of ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC`` in config files can be changed to use 16 bytes size RX descriptors.
670 Example of getting best performance with l3fwd example
671 ------------------------------------------------------
673 The following is an example of running the DPDK ``l3fwd`` sample application to get high performance with a
674 server with Intel Xeon processors and Intel Ethernet CNA XL710.
676 The example scenario is to get best performance with two Intel Ethernet CNA XL710 40GbE ports.
677 See :numref:`figure_intel_perf_test_setup` for the performance test setup.
679 .. _figure_intel_perf_test_setup:
681 .. figure:: img/intel_perf_test_setup.*
683 Performance Test Setup
686 1. Add two Intel Ethernet CNA XL710 to the platform, and use one port per card to get best performance.
687 The reason for using two NICs is to overcome a PCIe v3.0 limitation since it cannot provide 80GbE bandwidth
688 for two 40GbE ports, but two different PCIe v3.0 x8 slot can.
689 Refer to the sample NICs output above, then we can select ``82:00.0`` and ``85:00.0`` as test ports::
691 82:00.0 Ethernet [0200]: Intel XL710 for 40GbE QSFP+ [8086:1583]
692 85:00.0 Ethernet [0200]: Intel XL710 for 40GbE QSFP+ [8086:1583]
694 2. Connect the ports to the traffic generator. For high speed testing, it's best to use a hardware traffic generator.
696 3. Check the PCI devices numa node (socket id) and get the cores number on the exact socket id.
697 In this case, ``82:00.0`` and ``85:00.0`` are both in socket 1, and the cores on socket 1 in the referenced platform
699 Note: Don't use 2 logical cores on the same core (e.g core18 has 2 logical cores, core18 and core54), instead, use 2 logical
700 cores from different cores (e.g core18 and core19).
702 4. Bind these two ports to igb_uio.
704 5. As to Intel Ethernet CNA XL710 40GbE port, we need at least two queue pairs to achieve best performance, then two queues per port
705 will be required, and each queue pair will need a dedicated CPU core for receiving/transmitting packets.
707 6. The DPDK sample application ``l3fwd`` will be used for performance testing, with using two ports for bi-directional forwarding.
708 Compile the ``l3fwd sample`` with the default lpm mode.
710 7. The command line of running l3fwd would be something like the following::
712 ./l3fwd -l 18-21 -n 4 -w 82:00.0 -w 85:00.0 \
713 -- -p 0x3 --config '(0,0,18),(0,1,19),(1,0,20),(1,1,21)'
715 This means that the application uses core 18 for port 0, queue pair 0 forwarding, core 19 for port 0, queue pair 1 forwarding,
716 core 20 for port 1, queue pair 0 forwarding, and core 21 for port 1, queue pair 1 forwarding.
718 8. Configure the traffic at a traffic generator.
720 * Start creating a stream on packet generator.
722 * Set the Ethernet II type to 0x0800.
724 Tx bytes affected by the link status change
725 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
727 For firmware versions prior to 6.01 for X710 series and 3.33 for X722 series, the tx_bytes statistics data is affected by
728 the link down event. Each time the link status changes to down, the tx_bytes decreases 110 bytes.