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 Pre-Installation Configuration
62 ------------------------------
67 The following options can be modified in the ``config`` file.
68 Please note that enabling debugging options may affect system performance.
70 - ``CONFIG_RTE_LIBRTE_I40E_PMD`` (default ``y``)
72 Toggle compilation of the ``librte_pmd_i40e`` driver.
74 - ``CONFIG_RTE_LIBRTE_I40E_DEBUG_*`` (default ``n``)
76 Toggle display of generic debugging messages.
78 - ``CONFIG_RTE_LIBRTE_I40E_RX_ALLOW_BULK_ALLOC`` (default ``y``)
80 Toggle bulk allocation for RX.
82 - ``CONFIG_RTE_LIBRTE_I40E_INC_VECTOR`` (default ``n``)
84 Toggle the use of Vector PMD instead of normal RX/TX path.
85 To enable vPMD for RX, bulk allocation for Rx must be allowed.
87 - ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC`` (default ``n``)
89 Toggle to use a 16-byte RX descriptor, by default the RX descriptor is 32 byte.
91 - ``CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_PF`` (default ``64``)
93 Number of queues reserved for PF.
95 - ``CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_VM`` (default ``4``)
97 Number of queues reserved for each VMDQ Pool.
99 - ``CONFIG_RTE_LIBRTE_I40E_ITR_INTERVAL`` (default ``-1``)
101 Interrupt Throttling interval.
104 Runtime Config Options
105 ~~~~~~~~~~~~~~~~~~~~~~
107 - ``Number of Queues per VF`` (default ``4``)
109 The number of queue per VF is determined by its host PF. If the PCI address
110 of an i40e PF is aaaa:bb.cc, the number of queues per VF can be configured
111 with EAL parameter like -w aaaa:bb.cc,queue-num-per-vf=n. The value n can be
112 1, 2, 4, 8 or 16. If no such parameter is configured, the number of queues
113 per VF is 4 by default.
115 - ``Support multiple driver`` (default ``disable``)
117 There was a multiple driver support issue during use of 700 series Ethernet
118 Adapter with both Linux kernel and DPDK PMD. To fix this issue, ``devargs``
119 parameter ``support-multi-driver`` is introduced, for example::
121 -w 84:00.0,support-multi-driver=1
123 With the above configuration, DPDK PMD will not change global registers, and
124 will switch PF interrupt from IntN to Int0 to avoid interrupt conflict between
125 DPDK and Linux Kernel.
127 - ``Support VF Port Representor`` (default ``not enabled``)
129 The i40e PF PMD supports the creation of VF port representors for the control
130 and monitoring of i40e virtual function devices. Each port representor
131 corresponds to a single virtual function of that device. Using the ``devargs``
132 option ``representor`` the user can specify which virtual functions to create
133 port representors for on initialization of the PF PMD by passing the VF IDs of
134 the VFs which are required.::
136 -w DBDF,representor=[0,1,4]
138 Currently hot-plugging of representor ports is not supported so all required
139 representors must be specified on the creation of the PF.
141 Driver compilation and testing
142 ------------------------------
144 Refer to the document :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
148 SR-IOV: Prerequisites and sample Application Notes
149 --------------------------------------------------
151 #. Load the kernel module:
153 .. code-block:: console
157 Check the output in dmesg:
159 .. code-block:: console
161 i40e 0000:83:00.1 ens802f0: renamed from eth0
163 #. Bring up the PF ports:
165 .. code-block:: console
169 #. Create VF device(s):
171 Echo the number of VFs to be created into the ``sriov_numvfs`` sysfs entry
176 .. code-block:: console
178 echo 2 > /sys/devices/pci0000:00/0000:00:03.0/0000:81:00.0/sriov_numvfs
181 #. Assign VF MAC address:
183 Assign MAC address to the VF using iproute2 utility. The syntax is:
185 .. code-block:: console
187 ip link set <PF netdev id> vf <VF id> mac <macaddr>
191 .. code-block:: console
193 ip link set ens802f0 vf 0 mac a0:b0:c0:d0:e0:f0
195 #. Assign VF to VM, and bring up the VM.
196 Please see the documentation for the *I40E/IXGBE/IGB Virtual Function Driver*.
200 Follow instructions available in the document
201 :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
206 .. code-block:: console
209 EAL: PCI device 0000:83:00.0 on NUMA socket 1
210 EAL: probe driver: 8086:1572 rte_i40e_pmd
211 EAL: PCI memory mapped at 0x7f7f80000000
212 EAL: PCI memory mapped at 0x7f7f80800000
213 PMD: eth_i40e_dev_init(): FW 5.0 API 1.5 NVM 05.00.02 eetrack 8000208a
214 Interactive-mode selected
215 Configuring Port 0 (socket 0)
218 PMD: i40e_dev_rx_queue_setup(): Rx Burst Bulk Alloc Preconditions are
219 satisfied.Rx Burst Bulk Alloc function will be used on port=0, queue=0.
222 Port 0: 68:05:CA:26:85:84
223 Checking link statuses...
224 Port 0 Link Up - speed 10000 Mbps - full-duplex
230 Sample Application Notes
231 ------------------------
236 Vlan filter only works when Promiscuous mode is off.
238 To start ``testpmd``, and add vlan 10 to port 0:
240 .. code-block:: console
242 ./app/testpmd -l 0-15 -n 4 -- -i --forward-mode=mac
245 testpmd> set promisc 0 off
246 testpmd> rx_vlan add 10 0
252 The Flow Director works in receive mode to identify specific flows or sets of flows and route them to specific queues.
253 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.
255 The default input set of each flow type is::
257 ipv4-other : src_ip_address, dst_ip_address
258 ipv4-frag : src_ip_address, dst_ip_address
259 ipv4-tcp : src_ip_address, dst_ip_address, src_port, dst_port
260 ipv4-udp : src_ip_address, dst_ip_address, src_port, dst_port
261 ipv4-sctp : src_ip_address, dst_ip_address, src_port, dst_port,
263 ipv6-other : src_ip_address, dst_ip_address
264 ipv6-frag : src_ip_address, dst_ip_address
265 ipv6-tcp : src_ip_address, dst_ip_address, src_port, dst_port
266 ipv6-udp : src_ip_address, dst_ip_address, src_port, dst_port
267 ipv6-sctp : src_ip_address, dst_ip_address, src_port, dst_port,
269 l2_payload : ether_type
271 The flex payload is selected from offset 0 to 15 of packet's payload by default, while it is masked out from matching.
273 Start ``testpmd`` with ``--disable-rss`` and ``--pkt-filter-mode=perfect``:
275 .. code-block:: console
277 ./app/testpmd -l 0-15 -n 4 -- -i --disable-rss --pkt-filter-mode=perfect \
278 --rxq=8 --txq=8 --nb-cores=8 --nb-ports=1
280 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:
282 .. code-block:: console
284 testpmd> flow_director_filter 0 mode IP add flow ipv4-udp \
285 src 2.2.2.3 32 dst 2.2.2.5 32 vlan 0 flexbytes () \
286 fwd pf queue 1 fd_id 1
288 Check the flow director status:
290 .. code-block:: console
292 testpmd> show port fdir 0
294 ######################## FDIR infos for port 0 ####################
296 SUPPORTED FLOW TYPE: ipv4-frag ipv4-tcp ipv4-udp ipv4-sctp ipv4-other
297 ipv6-frag ipv6-tcp ipv6-udp ipv6-sctp ipv6-other
300 max_len: 16 payload_limit: 480
301 payload_unit: 2 payload_seg: 3
302 bitmask_unit: 2 bitmask_num: 2
305 src_ipv4: 0x00000000,
306 dst_ipv4: 0x00000000,
309 src_ipv6: 0x00000000,0x00000000,0x00000000,0x00000000,
310 dst_ipv6: 0x00000000,0x00000000,0x00000000,0x00000000
311 FLEX PAYLOAD SRC OFFSET:
312 L2_PAYLOAD: 0 1 2 3 4 5 6 ...
313 L3_PAYLOAD: 0 1 2 3 4 5 6 ...
314 L4_PAYLOAD: 0 1 2 3 4 5 6 ...
316 ipv4-udp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
317 ipv4-tcp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
318 ipv4-sctp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
319 ipv4-other: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
320 ipv4-frag: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
321 ipv6-udp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
322 ipv6-tcp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
323 ipv6-sctp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
324 ipv6-other: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
325 ipv6-frag: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
326 l2_payload: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
327 guarant_count: 1 best_count: 0
328 guarant_space: 512 best_space: 7168
335 Delete all flow director rules on a port:
337 .. code-block:: console
339 testpmd> flush_flow_director 0
344 The Intel® Ethernet 700 Series support a feature called
347 A Virtual Ethernet Bridge (VEB) is an IEEE Edge Virtual Bridging (EVB) term
348 for functionality that allows local switching between virtual endpoints within
349 a physical endpoint and also with an external bridge/network.
351 A "Floating" VEB doesn't have an uplink connection to the outside world so all
352 switching is done internally and remains within the host. As such, this
353 feature provides security benefits.
355 In addition, a Floating VEB overcomes a limitation of normal VEBs where they
356 cannot forward packets when the physical link is down. Floating VEBs don't need
357 to connect to the NIC port so they can still forward traffic from VF to VF
358 even when the physical link is down.
360 Therefore, with this feature enabled VFs can be limited to communicating with
361 each other but not an outside network, and they can do so even when there is
362 no physical uplink on the associated NIC port.
364 To enable this feature, the user should pass a ``devargs`` parameter to the
367 -w 84:00.0,enable_floating_veb=1
369 In this configuration the PMD will use the floating VEB feature for all the
370 VFs created by this PF device.
372 Alternatively, the user can specify which VFs need to connect to this floating
373 VEB using the ``floating_veb_list`` argument::
375 -w 84:00.0,enable_floating_veb=1,floating_veb_list=1;3-4
377 In this example ``VF1``, ``VF3`` and ``VF4`` connect to the floating VEB,
378 while other VFs connect to the normal VEB.
380 The current implementation only supports one floating VEB and one regular
381 VEB. VFs can connect to a floating VEB or a regular VEB according to the
382 configuration passed on the EAL command line.
384 The floating VEB functionality requires a NIC firmware version of 5.0
387 Dynamic Device Personalization (DDP)
388 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
390 The Intel® Ethernet 700 Series except for the Intel Ethernet Connection
391 X722 support a feature called "Dynamic Device Personalization (DDP)",
392 which is used to configure hardware by downloading a profile to support
393 protocols/filters which are not supported by default. The DDP
394 functionality requires a NIC firmware version of 6.0 or greater.
396 Current implementation supports GTP-C/GTP-U/PPPoE/PPPoL2TP,
397 steering can be used with rte_flow API.
399 Load a profile which supports GTP and store backup profile:
401 .. code-block:: console
403 testpmd> ddp add 0 ./gtp.pkgo,./backup.pkgo
405 Delete a GTP profile and restore backup profile:
407 .. code-block:: console
409 testpmd> ddp del 0 ./backup.pkgo
411 Get loaded DDP package info list:
413 .. code-block:: console
415 testpmd> ddp get list 0
417 Display information about a GTP profile:
419 .. code-block:: console
421 testpmd> ddp get info ./gtp.pkgo
423 Input set configuration
424 ~~~~~~~~~~~~~~~~~~~~~~~
425 Input set for any PCTYPE can be configured with user defined configuration,
426 For example, to use only 48bit prefix for IPv6 src address for IPv6 TCP RSS:
428 .. code-block:: console
430 testpmd> port config 0 pctype 43 hash_inset clear all
431 testpmd> port config 0 pctype 43 hash_inset set field 13
432 testpmd> port config 0 pctype 43 hash_inset set field 14
433 testpmd> port config 0 pctype 43 hash_inset set field 15
435 Queue region configuration
436 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
437 The Intel® Ethernet 700 Series supports a feature of queue regions
438 configuration for RSS in the PF, so that different traffic classes or
439 different packet classification types can be separated to different
440 queues in different queue regions. There is an API for configuration
441 of queue regions in RSS with a command line. It can parse the parameters
442 of the region index, queue number, queue start index, user priority, traffic
443 classes and so on. Depending on commands from the command line, it will call
444 i40e private APIs and start the process of setting or flushing the queue
445 region configuration. As this feature is specific for i40e only private
446 APIs are used. These new ``test_pmd`` commands are as shown below. For
447 details please refer to :doc:`../testpmd_app_ug/index`.
449 .. code-block:: console
451 testpmd> set port (port_id) queue-region region_id (value) \
452 queue_start_index (value) queue_num (value)
453 testpmd> set port (port_id) queue-region region_id (value) flowtype (value)
454 testpmd> set port (port_id) queue-region UP (value) region_id (value)
455 testpmd> set port (port_id) queue-region flush (on|off)
456 testpmd> show port (port_id) queue-region
458 Limitations or Known issues
459 ---------------------------
461 MPLS packet classification
462 ~~~~~~~~~~~~~~~~~~~~~~~~~~
464 For firmware versions prior to 5.0, MPLS packets are not recognized by the NIC.
465 The L2 Payload flow type in flow director can be used to classify MPLS packet
466 by using a command in testpmd like:
468 testpmd> flow_director_filter 0 mode IP add flow l2_payload ether \
469 0x8847 flexbytes () fwd pf queue <N> fd_id <M>
471 With the NIC firmware version 5.0 or greater, some limited MPLS support
472 is added: Native MPLS (MPLS in Ethernet) skip is implemented, while no
473 new packet type, no classification or offload are possible. With this change,
474 L2 Payload flow type in flow director cannot be used to classify MPLS packet
475 as with previous firmware versions. Meanwhile, the Ethertype filter can be
476 used to classify MPLS packet by using a command in testpmd like:
478 testpmd> ethertype_filter 0 add mac_ignr 00:00:00:00:00:00 ethertype \
481 16 Byte RX Descriptor setting on DPDK VF
482 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
484 Currently the VF's RX descriptor mode is decided by PF. There's no PF-VF
485 interface for VF to request the RX descriptor mode, also no interface to notify
486 VF its own RX descriptor mode.
487 For all available versions of the i40e driver, these drivers don't support 16
488 byte RX descriptor. If the Linux i40e kernel driver is used as host driver,
489 while DPDK i40e PMD is used as the VF driver, DPDK cannot choose 16 byte receive
490 descriptor. The reason is that the RX descriptor is already set to 32 byte by
491 the i40e kernel driver. That is to say, user should keep
492 ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC=n`` in config file.
493 In the future, if the Linux i40e driver supports 16 byte RX descriptor, user
494 should make sure the DPDK VF uses the same RX descriptor mode, 16 byte or 32
495 byte, as the PF driver.
497 The same rule for DPDK PF + DPDK VF. The PF and VF should use the same RX
498 descriptor mode. Or the VF RX will not work.
500 Receive packets with Ethertype 0x88A8
501 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
503 Due to the FW limitation, PF can receive packets with Ethertype 0x88A8
504 only when floating VEB is disabled.
506 Incorrect Rx statistics when packet is oversize
507 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
509 When a packet is over maximum frame size, the packet is dropped.
510 However, the Rx statistics, when calling `rte_eth_stats_get` incorrectly
511 shows it as received.
513 VF & TC max bandwidth setting
514 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
516 The per VF max bandwidth and per TC max bandwidth cannot be enabled in parallel.
517 The behavior is different when handling per VF and per TC max bandwidth setting.
518 When enabling per VF max bandwidth, SW will check if per TC max bandwidth is
519 enabled. If so, return failure.
520 When enabling per TC max bandwidth, SW will check if per VF max bandwidth
521 is enabled. If so, disable per VF max bandwidth and continue with per TC max
524 TC TX scheduling mode setting
525 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
527 There're 2 TX scheduling modes for TCs, round robin and strict priority mode.
528 If a TC is set to strict priority mode, it can consume unlimited bandwidth.
529 It means if APP has set the max bandwidth for that TC, it comes to no
531 It's suggested to set the strict priority mode for a TC that is latency
532 sensitive but no consuming much bandwidth.
534 VF performance is impacted by PCI extended tag setting
535 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
537 To reach maximum NIC performance in the VF the PCI extended tag must be
538 enabled. The DPDK i40e PF driver will set this feature during initialization,
539 but the kernel PF driver does not. So when running traffic on a VF which is
540 managed by the kernel PF driver, a significant NIC performance downgrade has
541 been observed (for 64 byte packets, there is about 25% line-rate downgrade for
542 a 25GbE device and about 35% for a 40GbE device).
544 For kernel version >= 4.11, the kernel's PCI driver will enable the extended
545 tag if it detects that the device supports it. So by default, this is not an
546 issue. For kernels <= 4.11 or when the PCI extended tag is disabled it can be
547 enabled using the steps below.
549 #. Get the current value of the PCI configure register::
551 setpci -s <XX:XX.X> a8.w
555 value = value | 0x100
557 #. Set the PCI configure register with new value::
559 setpci -s <XX:XX.X> a8.w=<value>
564 The VF vlan strip function is only supported in the i40e kernel driver >= 2.1.26.
569 DCB works only when RSS is enabled.
571 Global configuration warning
572 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
574 I40E PMD will set some global registers to enable some function or set some
575 configure. Then when using different ports of the same NIC with Linux kernel
576 and DPDK, the port with Linux kernel will be impacted by the port with DPDK.
577 For example, register I40E_GL_SWT_L2TAGCTRL is used to control L2 tag, i40e
578 PMD uses I40E_GL_SWT_L2TAGCTRL to set vlan TPID. If setting TPID in port A
579 with DPDK, then the configuration will also impact port B in the NIC with
580 kernel driver, which don't want to use the TPID.
581 So PMD reports warning to clarify what is changed by writing global register.
583 High Performance of Small Packets on 40GbE NIC
584 ----------------------------------------------
586 As there might be firmware fixes for performance enhancement in latest version
587 of firmware image, the firmware update might be needed for getting high performance.
588 Check the Intel support website for the latest firmware updates.
589 Users should consult the release notes specific to a DPDK release to identify
590 the validated firmware version for a NIC using the i40e driver.
592 Use 16 Bytes RX Descriptor Size
593 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
595 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.
596 Configuration of ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC`` in config files can be changed to use 16 bytes size RX descriptors.
598 High Performance and per Packet Latency Tradeoff
599 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
601 Due to the hardware design, the interrupt signal inside NIC is needed for per
602 packet descriptor write-back. The minimum interval of interrupts could be set
603 at compile time by ``CONFIG_RTE_LIBRTE_I40E_ITR_INTERVAL`` in configuration files.
604 Though there is a default configuration, the interval could be tuned by the
605 users with that configuration item depends on what the user cares about more,
606 performance or per packet latency.
608 Example of getting best performance with l3fwd example
609 ------------------------------------------------------
611 The following is an example of running the DPDK ``l3fwd`` sample application to get high performance with a
612 server with Intel Xeon processors and Intel Ethernet CNA XL710.
614 The example scenario is to get best performance with two Intel Ethernet CNA XL710 40GbE ports.
615 See :numref:`figure_intel_perf_test_setup` for the performance test setup.
617 .. _figure_intel_perf_test_setup:
619 .. figure:: img/intel_perf_test_setup.*
621 Performance Test Setup
624 1. Add two Intel Ethernet CNA XL710 to the platform, and use one port per card to get best performance.
625 The reason for using two NICs is to overcome a PCIe v3.0 limitation since it cannot provide 80GbE bandwidth
626 for two 40GbE ports, but two different PCIe v3.0 x8 slot can.
627 Refer to the sample NICs output above, then we can select ``82:00.0`` and ``85:00.0`` as test ports::
629 82:00.0 Ethernet [0200]: Intel XL710 for 40GbE QSFP+ [8086:1583]
630 85:00.0 Ethernet [0200]: Intel XL710 for 40GbE QSFP+ [8086:1583]
632 2. Connect the ports to the traffic generator. For high speed testing, it's best to use a hardware traffic generator.
634 3. Check the PCI devices numa node (socket id) and get the cores number on the exact socket id.
635 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
637 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
638 cores from different cores (e.g core18 and core19).
640 4. Bind these two ports to igb_uio.
642 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
643 will be required, and each queue pair will need a dedicated CPU core for receiving/transmitting packets.
645 6. The DPDK sample application ``l3fwd`` will be used for performance testing, with using two ports for bi-directional forwarding.
646 Compile the ``l3fwd sample`` with the default lpm mode.
648 7. The command line of running l3fwd would be something like the following::
650 ./l3fwd -l 18-21 -n 4 -w 82:00.0 -w 85:00.0 \
651 -- -p 0x3 --config '(0,0,18),(0,1,19),(1,0,20),(1,1,21)'
653 This means that the application uses core 18 for port 0, queue pair 0 forwarding, core 19 for port 0, queue pair 1 forwarding,
654 core 20 for port 1, queue pair 0 forwarding, and core 21 for port 1, queue pair 1 forwarding.
656 8. Configure the traffic at a traffic generator.
658 * Start creating a stream on packet generator.
660 * Set the Ethernet II type to 0x0800.