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 | 19.02 | 2.7.26 | 6.80 |
73 +--------------+-----------------------+------------------+
74 | 18.11 | 2.4.6 | 6.01 |
75 +--------------+-----------------------+------------------+
76 | 18.08 | 2.4.6 | 6.01 |
77 +--------------+-----------------------+------------------+
78 | 18.05 | 2.4.6 | 6.01 |
79 +--------------+-----------------------+------------------+
80 | 18.02 | 2.4.3 | 6.01 |
81 +--------------+-----------------------+------------------+
82 | 17.11 | 2.1.26 | 6.01 |
83 +--------------+-----------------------+------------------+
84 | 17.08 | 2.0.19 | 6.01 |
85 +--------------+-----------------------+------------------+
86 | 17.05 | 1.5.23 | 5.05 |
87 +--------------+-----------------------+------------------+
88 | 17.02 | 1.5.23 | 5.05 |
89 +--------------+-----------------------+------------------+
90 | 16.11 | 1.5.23 | 5.05 |
91 +--------------+-----------------------+------------------+
92 | 16.07 | 1.4.25 | 5.04 |
93 +--------------+-----------------------+------------------+
94 | 16.04 | 1.4.25 | 5.02 |
95 +--------------+-----------------------+------------------+
97 Pre-Installation Configuration
98 ------------------------------
103 The following options can be modified in the ``config`` file.
104 Please note that enabling debugging options may affect system performance.
106 - ``CONFIG_RTE_LIBRTE_I40E_PMD`` (default ``y``)
108 Toggle compilation of the ``librte_pmd_i40e`` driver.
110 - ``CONFIG_RTE_LIBRTE_I40E_DEBUG_*`` (default ``n``)
112 Toggle display of generic debugging messages.
114 - ``CONFIG_RTE_LIBRTE_I40E_RX_ALLOW_BULK_ALLOC`` (default ``y``)
116 Toggle bulk allocation for RX.
118 - ``CONFIG_RTE_LIBRTE_I40E_INC_VECTOR`` (default ``n``)
120 Toggle the use of Vector PMD instead of normal RX/TX path.
121 To enable vPMD for RX, bulk allocation for Rx must be allowed.
123 - ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC`` (default ``n``)
125 Toggle to use a 16-byte RX descriptor, by default the RX descriptor is 32 byte.
127 - ``CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_PF`` (default ``64``)
129 Number of queues reserved for PF.
131 - ``CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_VM`` (default ``4``)
133 Number of queues reserved for each VMDQ Pool.
135 Runtime Config Options
136 ~~~~~~~~~~~~~~~~~~~~~~
138 - ``Reserved number of Queues per VF`` (default ``4``)
140 The number of reserved queue per VF is determined by its host PF. If the
141 PCI address of an i40e PF is aaaa:bb.cc, the number of reserved queues per
142 VF can be configured with EAL parameter like -w aaaa:bb.cc,queue-num-per-vf=n.
143 The value n can be 1, 2, 4, 8 or 16. If no such parameter is configured, the
144 number of reserved queues per VF is 4 by default. If VF request more than
145 reserved queues per VF, PF will able to allocate max to 16 queues after a VF
149 - ``Support multiple driver`` (default ``disable``)
151 There was a multiple driver support issue during use of 700 series Ethernet
152 Adapter with both Linux kernel and DPDK PMD. To fix this issue, ``devargs``
153 parameter ``support-multi-driver`` is introduced, for example::
155 -w 84:00.0,support-multi-driver=1
157 With the above configuration, DPDK PMD will not change global registers, and
158 will switch PF interrupt from IntN to Int0 to avoid interrupt conflict between
159 DPDK and Linux Kernel.
161 - ``Support VF Port Representor`` (default ``not enabled``)
163 The i40e PF PMD supports the creation of VF port representors for the control
164 and monitoring of i40e virtual function devices. Each port representor
165 corresponds to a single virtual function of that device. Using the ``devargs``
166 option ``representor`` the user can specify which virtual functions to create
167 port representors for on initialization of the PF PMD by passing the VF IDs of
168 the VFs which are required.::
170 -w DBDF,representor=[0,1,4]
172 Currently hot-plugging of representor ports is not supported so all required
173 representors must be specified on the creation of the PF.
175 - ``Use latest supported vector`` (default ``disable``)
177 Latest supported vector path may not always get the best perf so vector path was
178 recommended to use only on later platform. But users may want the latest vector path
179 since it can get better perf in some real work loading cases. So ``devargs`` param
180 ``use-latest-supported-vec`` is introduced, for example::
182 -w 84:00.0,use-latest-supported-vec=1
184 Vector RX Pre-conditions
185 ~~~~~~~~~~~~~~~~~~~~~~~~
186 For Vector RX it is assumed that the number of descriptor rings will be a power
187 of 2. With this pre-condition, the ring pointer can easily scroll back to the
188 head after hitting the tail without a conditional check. In addition Vector RX
189 can use this assumption to do a bit mask using ``ring_size - 1``.
191 Driver compilation and testing
192 ------------------------------
194 Refer to the document :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
198 SR-IOV: Prerequisites and sample Application Notes
199 --------------------------------------------------
201 #. Load the kernel module:
203 .. code-block:: console
207 Check the output in dmesg:
209 .. code-block:: console
211 i40e 0000:83:00.1 ens802f0: renamed from eth0
213 #. Bring up the PF ports:
215 .. code-block:: console
219 #. Create VF device(s):
221 Echo the number of VFs to be created into the ``sriov_numvfs`` sysfs entry
226 .. code-block:: console
228 echo 2 > /sys/devices/pci0000:00/0000:00:03.0/0000:81:00.0/sriov_numvfs
231 #. Assign VF MAC address:
233 Assign MAC address to the VF using iproute2 utility. The syntax is:
235 .. code-block:: console
237 ip link set <PF netdev id> vf <VF id> mac <macaddr>
241 .. code-block:: console
243 ip link set ens802f0 vf 0 mac a0:b0:c0:d0:e0:f0
245 #. Assign VF to VM, and bring up the VM.
246 Please see the documentation for the *I40E/IXGBE/IGB Virtual Function Driver*.
250 Follow instructions available in the document
251 :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
256 .. code-block:: console
259 EAL: PCI device 0000:83:00.0 on NUMA socket 1
260 EAL: probe driver: 8086:1572 rte_i40e_pmd
261 EAL: PCI memory mapped at 0x7f7f80000000
262 EAL: PCI memory mapped at 0x7f7f80800000
263 PMD: eth_i40e_dev_init(): FW 5.0 API 1.5 NVM 05.00.02 eetrack 8000208a
264 Interactive-mode selected
265 Configuring Port 0 (socket 0)
268 PMD: i40e_dev_rx_queue_setup(): Rx Burst Bulk Alloc Preconditions are
269 satisfied.Rx Burst Bulk Alloc function will be used on port=0, queue=0.
272 Port 0: 68:05:CA:26:85:84
273 Checking link statuses...
274 Port 0 Link Up - speed 10000 Mbps - full-duplex
280 Sample Application Notes
281 ------------------------
286 Vlan filter only works when Promiscuous mode is off.
288 To start ``testpmd``, and add vlan 10 to port 0:
290 .. code-block:: console
292 ./app/testpmd -l 0-15 -n 4 -- -i --forward-mode=mac
295 testpmd> set promisc 0 off
296 testpmd> rx_vlan add 10 0
302 The Flow Director works in receive mode to identify specific flows or sets of flows and route them to specific queues.
303 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.
305 The default input set of each flow type is::
307 ipv4-other : src_ip_address, dst_ip_address
308 ipv4-frag : src_ip_address, dst_ip_address
309 ipv4-tcp : src_ip_address, dst_ip_address, src_port, dst_port
310 ipv4-udp : src_ip_address, dst_ip_address, src_port, dst_port
311 ipv4-sctp : src_ip_address, dst_ip_address, src_port, dst_port,
313 ipv6-other : src_ip_address, dst_ip_address
314 ipv6-frag : src_ip_address, dst_ip_address
315 ipv6-tcp : src_ip_address, dst_ip_address, src_port, dst_port
316 ipv6-udp : src_ip_address, dst_ip_address, src_port, dst_port
317 ipv6-sctp : src_ip_address, dst_ip_address, src_port, dst_port,
319 l2_payload : ether_type
321 The flex payload is selected from offset 0 to 15 of packet's payload by default, while it is masked out from matching.
323 Start ``testpmd`` with ``--disable-rss`` and ``--pkt-filter-mode=perfect``:
325 .. code-block:: console
327 ./app/testpmd -l 0-15 -n 4 -- -i --disable-rss --pkt-filter-mode=perfect \
328 --rxq=8 --txq=8 --nb-cores=8 --nb-ports=1
330 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:
332 .. code-block:: console
334 testpmd> flow_director_filter 0 mode IP add flow ipv4-udp \
335 src 2.2.2.3 32 dst 2.2.2.5 32 vlan 0 flexbytes () \
336 fwd pf queue 1 fd_id 1
338 Check the flow director status:
340 .. code-block:: console
342 testpmd> show port fdir 0
344 ######################## FDIR infos for port 0 ####################
346 SUPPORTED FLOW TYPE: ipv4-frag ipv4-tcp ipv4-udp ipv4-sctp ipv4-other
347 ipv6-frag ipv6-tcp ipv6-udp ipv6-sctp ipv6-other
350 max_len: 16 payload_limit: 480
351 payload_unit: 2 payload_seg: 3
352 bitmask_unit: 2 bitmask_num: 2
355 src_ipv4: 0x00000000,
356 dst_ipv4: 0x00000000,
359 src_ipv6: 0x00000000,0x00000000,0x00000000,0x00000000,
360 dst_ipv6: 0x00000000,0x00000000,0x00000000,0x00000000
361 FLEX PAYLOAD SRC OFFSET:
362 L2_PAYLOAD: 0 1 2 3 4 5 6 ...
363 L3_PAYLOAD: 0 1 2 3 4 5 6 ...
364 L4_PAYLOAD: 0 1 2 3 4 5 6 ...
366 ipv4-udp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
367 ipv4-tcp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
368 ipv4-sctp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
369 ipv4-other: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
370 ipv4-frag: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
371 ipv6-udp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
372 ipv6-tcp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
373 ipv6-sctp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
374 ipv6-other: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
375 ipv6-frag: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
376 l2_payload: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
377 guarant_count: 1 best_count: 0
378 guarant_space: 512 best_space: 7168
385 Delete all flow director rules on a port:
387 .. code-block:: console
389 testpmd> flush_flow_director 0
394 The Intel® Ethernet 700 Series support a feature called
397 A Virtual Ethernet Bridge (VEB) is an IEEE Edge Virtual Bridging (EVB) term
398 for functionality that allows local switching between virtual endpoints within
399 a physical endpoint and also with an external bridge/network.
401 A "Floating" VEB doesn't have an uplink connection to the outside world so all
402 switching is done internally and remains within the host. As such, this
403 feature provides security benefits.
405 In addition, a Floating VEB overcomes a limitation of normal VEBs where they
406 cannot forward packets when the physical link is down. Floating VEBs don't need
407 to connect to the NIC port so they can still forward traffic from VF to VF
408 even when the physical link is down.
410 Therefore, with this feature enabled VFs can be limited to communicating with
411 each other but not an outside network, and they can do so even when there is
412 no physical uplink on the associated NIC port.
414 To enable this feature, the user should pass a ``devargs`` parameter to the
417 -w 84:00.0,enable_floating_veb=1
419 In this configuration the PMD will use the floating VEB feature for all the
420 VFs created by this PF device.
422 Alternatively, the user can specify which VFs need to connect to this floating
423 VEB using the ``floating_veb_list`` argument::
425 -w 84:00.0,enable_floating_veb=1,floating_veb_list=1;3-4
427 In this example ``VF1``, ``VF3`` and ``VF4`` connect to the floating VEB,
428 while other VFs connect to the normal VEB.
430 The current implementation only supports one floating VEB and one regular
431 VEB. VFs can connect to a floating VEB or a regular VEB according to the
432 configuration passed on the EAL command line.
434 The floating VEB functionality requires a NIC firmware version of 5.0
437 Dynamic Device Personalization (DDP)
438 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
440 The Intel® Ethernet 700 Series except for the Intel Ethernet Connection
441 X722 support a feature called "Dynamic Device Personalization (DDP)",
442 which is used to configure hardware by downloading a profile to support
443 protocols/filters which are not supported by default. The DDP
444 functionality requires a NIC firmware version of 6.0 or greater.
446 Current implementation supports GTP-C/GTP-U/PPPoE/PPPoL2TP,
447 steering can be used with rte_flow API.
449 GTPv1 package is released, and it can be downloaded from
450 https://downloadcenter.intel.com/download/27587.
452 PPPoE package is released, and it can be downloaded from
453 https://downloadcenter.intel.com/download/28040.
455 Load a profile which supports GTP and store backup profile:
457 .. code-block:: console
459 testpmd> ddp add 0 ./gtp.pkgo,./backup.pkgo
461 Delete a GTP profile and restore backup profile:
463 .. code-block:: console
465 testpmd> ddp del 0 ./backup.pkgo
467 Get loaded DDP package info list:
469 .. code-block:: console
471 testpmd> ddp get list 0
473 Display information about a GTP profile:
475 .. code-block:: console
477 testpmd> ddp get info ./gtp.pkgo
479 Input set configuration
480 ~~~~~~~~~~~~~~~~~~~~~~~
481 Input set for any PCTYPE can be configured with user defined configuration,
482 For example, to use only 48bit prefix for IPv6 src address for IPv6 TCP RSS:
484 .. code-block:: console
486 testpmd> port config 0 pctype 43 hash_inset clear all
487 testpmd> port config 0 pctype 43 hash_inset set field 13
488 testpmd> port config 0 pctype 43 hash_inset set field 14
489 testpmd> port config 0 pctype 43 hash_inset set field 15
491 Queue region configuration
492 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
493 The Intel® Ethernet 700 Series supports a feature of queue regions
494 configuration for RSS in the PF, so that different traffic classes or
495 different packet classification types can be separated to different
496 queues in different queue regions. There is an API for configuration
497 of queue regions in RSS with a command line. It can parse the parameters
498 of the region index, queue number, queue start index, user priority, traffic
499 classes and so on. Depending on commands from the command line, it will call
500 i40e private APIs and start the process of setting or flushing the queue
501 region configuration. As this feature is specific for i40e only private
502 APIs are used. These new ``test_pmd`` commands are as shown below. For
503 details please refer to :doc:`../testpmd_app_ug/index`.
505 .. code-block:: console
507 testpmd> set port (port_id) queue-region region_id (value) \
508 queue_start_index (value) queue_num (value)
509 testpmd> set port (port_id) queue-region region_id (value) flowtype (value)
510 testpmd> set port (port_id) queue-region UP (value) region_id (value)
511 testpmd> set port (port_id) queue-region flush (on|off)
512 testpmd> show port (port_id) queue-region
514 Limitations or Known issues
515 ---------------------------
517 MPLS packet classification
518 ~~~~~~~~~~~~~~~~~~~~~~~~~~
520 For firmware versions prior to 5.0, MPLS packets are not recognized by the NIC.
521 The L2 Payload flow type in flow director can be used to classify MPLS packet
522 by using a command in testpmd like:
524 testpmd> flow_director_filter 0 mode IP add flow l2_payload ether \
525 0x8847 flexbytes () fwd pf queue <N> fd_id <M>
527 With the NIC firmware version 5.0 or greater, some limited MPLS support
528 is added: Native MPLS (MPLS in Ethernet) skip is implemented, while no
529 new packet type, no classification or offload are possible. With this change,
530 L2 Payload flow type in flow director cannot be used to classify MPLS packet
531 as with previous firmware versions. Meanwhile, the Ethertype filter can be
532 used to classify MPLS packet by using a command in testpmd like:
534 testpmd> ethertype_filter 0 add mac_ignr 00:00:00:00:00:00 ethertype \
537 16 Byte RX Descriptor setting on DPDK VF
538 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
540 Currently the VF's RX descriptor mode is decided by PF. There's no PF-VF
541 interface for VF to request the RX descriptor mode, also no interface to notify
542 VF its own RX descriptor mode.
543 For all available versions of the i40e driver, these drivers don't support 16
544 byte RX descriptor. If the Linux i40e kernel driver is used as host driver,
545 while DPDK i40e PMD is used as the VF driver, DPDK cannot choose 16 byte receive
546 descriptor. The reason is that the RX descriptor is already set to 32 byte by
547 the i40e kernel driver. That is to say, user should keep
548 ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC=n`` in config file.
549 In the future, if the Linux i40e driver supports 16 byte RX descriptor, user
550 should make sure the DPDK VF uses the same RX descriptor mode, 16 byte or 32
551 byte, as the PF driver.
553 The same rule for DPDK PF + DPDK VF. The PF and VF should use the same RX
554 descriptor mode. Or the VF RX will not work.
556 Receive packets with Ethertype 0x88A8
557 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
559 Due to the FW limitation, PF can receive packets with Ethertype 0x88A8
560 only when floating VEB is disabled.
562 Incorrect Rx statistics when packet is oversize
563 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
565 When a packet is over maximum frame size, the packet is dropped.
566 However, the Rx statistics, when calling `rte_eth_stats_get` incorrectly
567 shows it as received.
569 VF & TC max bandwidth setting
570 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
572 The per VF max bandwidth and per TC max bandwidth cannot be enabled in parallel.
573 The behavior is different when handling per VF and per TC max bandwidth setting.
574 When enabling per VF max bandwidth, SW will check if per TC max bandwidth is
575 enabled. If so, return failure.
576 When enabling per TC max bandwidth, SW will check if per VF max bandwidth
577 is enabled. If so, disable per VF max bandwidth and continue with per TC max
580 TC TX scheduling mode setting
581 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
583 There're 2 TX scheduling modes for TCs, round robin and strict priority mode.
584 If a TC is set to strict priority mode, it can consume unlimited bandwidth.
585 It means if APP has set the max bandwidth for that TC, it comes to no
587 It's suggested to set the strict priority mode for a TC that is latency
588 sensitive but no consuming much bandwidth.
590 VF performance is impacted by PCI extended tag setting
591 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
593 To reach maximum NIC performance in the VF the PCI extended tag must be
594 enabled. The DPDK i40e PF driver will set this feature during initialization,
595 but the kernel PF driver does not. So when running traffic on a VF which is
596 managed by the kernel PF driver, a significant NIC performance downgrade has
597 been observed (for 64 byte packets, there is about 25% line-rate downgrade for
598 a 25GbE device and about 35% for a 40GbE device).
600 For kernel version >= 4.11, the kernel's PCI driver will enable the extended
601 tag if it detects that the device supports it. So by default, this is not an
602 issue. For kernels <= 4.11 or when the PCI extended tag is disabled it can be
603 enabled using the steps below.
605 #. Get the current value of the PCI configure register::
607 setpci -s <XX:XX.X> a8.w
611 value = value | 0x100
613 #. Set the PCI configure register with new value::
615 setpci -s <XX:XX.X> a8.w=<value>
620 The VF vlan strip function is only supported in the i40e kernel driver >= 2.1.26.
625 DCB works only when RSS is enabled.
627 Global configuration warning
628 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
630 I40E PMD will set some global registers to enable some function or set some
631 configure. Then when using different ports of the same NIC with Linux kernel
632 and DPDK, the port with Linux kernel will be impacted by the port with DPDK.
633 For example, register I40E_GL_SWT_L2TAGCTRL is used to control L2 tag, i40e
634 PMD uses I40E_GL_SWT_L2TAGCTRL to set vlan TPID. If setting TPID in port A
635 with DPDK, then the configuration will also impact port B in the NIC with
636 kernel driver, which don't want to use the TPID.
637 So PMD reports warning to clarify what is changed by writing global register.
639 High Performance of Small Packets on 40GbE NIC
640 ----------------------------------------------
642 As there might be firmware fixes for performance enhancement in latest version
643 of firmware image, the firmware update might be needed for getting high performance.
644 Check the Intel support website for the latest firmware updates.
645 Users should consult the release notes specific to a DPDK release to identify
646 the validated firmware version for a NIC using the i40e driver.
648 Use 16 Bytes RX Descriptor Size
649 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
651 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.
652 Configuration of ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC`` in config files can be changed to use 16 bytes size RX descriptors.
654 Example of getting best performance with l3fwd example
655 ------------------------------------------------------
657 The following is an example of running the DPDK ``l3fwd`` sample application to get high performance with a
658 server with Intel Xeon processors and Intel Ethernet CNA XL710.
660 The example scenario is to get best performance with two Intel Ethernet CNA XL710 40GbE ports.
661 See :numref:`figure_intel_perf_test_setup` for the performance test setup.
663 .. _figure_intel_perf_test_setup:
665 .. figure:: img/intel_perf_test_setup.*
667 Performance Test Setup
670 1. Add two Intel Ethernet CNA XL710 to the platform, and use one port per card to get best performance.
671 The reason for using two NICs is to overcome a PCIe v3.0 limitation since it cannot provide 80GbE bandwidth
672 for two 40GbE ports, but two different PCIe v3.0 x8 slot can.
673 Refer to the sample NICs output above, then we can select ``82:00.0`` and ``85:00.0`` as test ports::
675 82:00.0 Ethernet [0200]: Intel XL710 for 40GbE QSFP+ [8086:1583]
676 85:00.0 Ethernet [0200]: Intel XL710 for 40GbE QSFP+ [8086:1583]
678 2. Connect the ports to the traffic generator. For high speed testing, it's best to use a hardware traffic generator.
680 3. Check the PCI devices numa node (socket id) and get the cores number on the exact socket id.
681 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
683 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
684 cores from different cores (e.g core18 and core19).
686 4. Bind these two ports to igb_uio.
688 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
689 will be required, and each queue pair will need a dedicated CPU core for receiving/transmitting packets.
691 6. The DPDK sample application ``l3fwd`` will be used for performance testing, with using two ports for bi-directional forwarding.
692 Compile the ``l3fwd sample`` with the default lpm mode.
694 7. The command line of running l3fwd would be something like the following::
696 ./l3fwd -l 18-21 -n 4 -w 82:00.0 -w 85:00.0 \
697 -- -p 0x3 --config '(0,0,18),(0,1,19),(1,0,20),(1,1,21)'
699 This means that the application uses core 18 for port 0, queue pair 0 forwarding, core 19 for port 0, queue pair 1 forwarding,
700 core 20 for port 1, queue pair 0 forwarding, and core 21 for port 1, queue pair 1 forwarding.
702 8. Configure the traffic at a traffic generator.
704 * Start creating a stream on packet generator.
706 * Set the Ethernet II type to 0x0800.
708 Tx bytes affected by the link status change
709 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
711 For firmware versions prior to 6.01 for X710 series and 3.33 for X722 series, the tx_bytes statistics data is affected by
712 the link down event. Each time the link status changes to down, the tx_bytes decreases 110 bytes.