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 | 19.05 | 2.7.29 | 6.80 |
73 +--------------+-----------------------+------------------+
74 | 19.02 | 2.7.26 | 6.80 |
75 +--------------+-----------------------+------------------+
76 | 18.11 | 2.4.6 | 6.01 |
77 +--------------+-----------------------+------------------+
78 | 18.08 | 2.4.6 | 6.01 |
79 +--------------+-----------------------+------------------+
80 | 18.05 | 2.4.6 | 6.01 |
81 +--------------+-----------------------+------------------+
82 | 18.02 | 2.4.3 | 6.01 |
83 +--------------+-----------------------+------------------+
84 | 17.11 | 2.1.26 | 6.01 |
85 +--------------+-----------------------+------------------+
86 | 17.08 | 2.0.19 | 6.01 |
87 +--------------+-----------------------+------------------+
88 | 17.05 | 1.5.23 | 5.05 |
89 +--------------+-----------------------+------------------+
90 | 17.02 | 1.5.23 | 5.05 |
91 +--------------+-----------------------+------------------+
92 | 16.11 | 1.5.23 | 5.05 |
93 +--------------+-----------------------+------------------+
94 | 16.07 | 1.4.25 | 5.04 |
95 +--------------+-----------------------+------------------+
96 | 16.04 | 1.4.25 | 5.02 |
97 +--------------+-----------------------+------------------+
99 Pre-Installation Configuration
100 ------------------------------
105 The following options can be modified in the ``config`` file.
106 Please note that enabling debugging options may affect system performance.
108 - ``CONFIG_RTE_LIBRTE_I40E_PMD`` (default ``y``)
110 Toggle compilation of the ``librte_pmd_i40e`` driver.
112 - ``CONFIG_RTE_LIBRTE_I40E_DEBUG_*`` (default ``n``)
114 Toggle display of generic debugging messages.
116 - ``CONFIG_RTE_LIBRTE_I40E_RX_ALLOW_BULK_ALLOC`` (default ``y``)
118 Toggle bulk allocation for RX.
120 - ``CONFIG_RTE_LIBRTE_I40E_INC_VECTOR`` (default ``n``)
122 Toggle the use of Vector PMD instead of normal RX/TX path.
123 To enable vPMD for RX, bulk allocation for Rx must be allowed.
125 - ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC`` (default ``n``)
127 Toggle to use a 16-byte RX descriptor, by default the RX descriptor is 32 byte.
129 - ``CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_PF`` (default ``64``)
131 Number of queues reserved for PF.
133 - ``CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_VM`` (default ``4``)
135 Number of queues reserved for each VMDQ Pool.
137 Runtime Config Options
138 ~~~~~~~~~~~~~~~~~~~~~~
140 - ``Reserved number of Queues per VF`` (default ``4``)
142 The number of reserved queue per VF is determined by its host PF. If the
143 PCI address of an i40e PF is aaaa:bb.cc, the number of reserved queues per
144 VF can be configured with EAL parameter like -w aaaa:bb.cc,queue-num-per-vf=n.
145 The value n can be 1, 2, 4, 8 or 16. If no such parameter is configured, the
146 number of reserved queues per VF is 4 by default. If VF request more than
147 reserved queues per VF, PF will able to allocate max to 16 queues after a VF
151 - ``Support multiple driver`` (default ``disable``)
153 There was a multiple driver support issue during use of 700 series Ethernet
154 Adapter with both Linux kernel and DPDK PMD. To fix this issue, ``devargs``
155 parameter ``support-multi-driver`` is introduced, for example::
157 -w 84:00.0,support-multi-driver=1
159 With the above configuration, DPDK PMD will not change global registers, and
160 will switch PF interrupt from IntN to Int0 to avoid interrupt conflict between
161 DPDK and Linux Kernel.
163 - ``Support VF Port Representor`` (default ``not enabled``)
165 The i40e PF PMD supports the creation of VF port representors for the control
166 and monitoring of i40e virtual function devices. Each port representor
167 corresponds to a single virtual function of that device. Using the ``devargs``
168 option ``representor`` the user can specify which virtual functions to create
169 port representors for on initialization of the PF PMD by passing the VF IDs of
170 the VFs which are required.::
172 -w DBDF,representor=[0,1,4]
174 Currently hot-plugging of representor ports is not supported so all required
175 representors must be specified on the creation of the PF.
177 - ``Use latest supported vector`` (default ``disable``)
179 Latest supported vector path may not always get the best perf so vector path was
180 recommended to use only on later platform. But users may want the latest vector path
181 since it can get better perf in some real work loading cases. So ``devargs`` param
182 ``use-latest-supported-vec`` is introduced, for example::
184 -w 84:00.0,use-latest-supported-vec=1
186 Vector RX Pre-conditions
187 ~~~~~~~~~~~~~~~~~~~~~~~~
188 For Vector RX it is assumed that the number of descriptor rings will be a power
189 of 2. With this pre-condition, the ring pointer can easily scroll back to the
190 head after hitting the tail without a conditional check. In addition Vector RX
191 can use this assumption to do a bit mask using ``ring_size - 1``.
193 Driver compilation and testing
194 ------------------------------
196 Refer to the document :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
200 SR-IOV: Prerequisites and sample Application Notes
201 --------------------------------------------------
203 #. Load the kernel module:
205 .. code-block:: console
209 Check the output in dmesg:
211 .. code-block:: console
213 i40e 0000:83:00.1 ens802f0: renamed from eth0
215 #. Bring up the PF ports:
217 .. code-block:: console
221 #. Create VF device(s):
223 Echo the number of VFs to be created into the ``sriov_numvfs`` sysfs entry
228 .. code-block:: console
230 echo 2 > /sys/devices/pci0000:00/0000:00:03.0/0000:81:00.0/sriov_numvfs
233 #. Assign VF MAC address:
235 Assign MAC address to the VF using iproute2 utility. The syntax is:
237 .. code-block:: console
239 ip link set <PF netdev id> vf <VF id> mac <macaddr>
243 .. code-block:: console
245 ip link set ens802f0 vf 0 mac a0:b0:c0:d0:e0:f0
247 #. Assign VF to VM, and bring up the VM.
248 Please see the documentation for the *I40E/IXGBE/IGB Virtual Function Driver*.
252 Follow instructions available in the document
253 :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
258 .. code-block:: console
261 EAL: PCI device 0000:83:00.0 on NUMA socket 1
262 EAL: probe driver: 8086:1572 rte_i40e_pmd
263 EAL: PCI memory mapped at 0x7f7f80000000
264 EAL: PCI memory mapped at 0x7f7f80800000
265 PMD: eth_i40e_dev_init(): FW 5.0 API 1.5 NVM 05.00.02 eetrack 8000208a
266 Interactive-mode selected
267 Configuring Port 0 (socket 0)
270 PMD: i40e_dev_rx_queue_setup(): Rx Burst Bulk Alloc Preconditions are
271 satisfied.Rx Burst Bulk Alloc function will be used on port=0, queue=0.
274 Port 0: 68:05:CA:26:85:84
275 Checking link statuses...
276 Port 0 Link Up - speed 10000 Mbps - full-duplex
282 Sample Application Notes
283 ------------------------
288 Vlan filter only works when Promiscuous mode is off.
290 To start ``testpmd``, and add vlan 10 to port 0:
292 .. code-block:: console
294 ./app/testpmd -l 0-15 -n 4 -- -i --forward-mode=mac
297 testpmd> set promisc 0 off
298 testpmd> rx_vlan add 10 0
304 The Flow Director works in receive mode to identify specific flows or sets of flows and route them to specific queues.
305 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.
307 The default input set of each flow type is::
309 ipv4-other : src_ip_address, dst_ip_address
310 ipv4-frag : src_ip_address, dst_ip_address
311 ipv4-tcp : src_ip_address, dst_ip_address, src_port, dst_port
312 ipv4-udp : src_ip_address, dst_ip_address, src_port, dst_port
313 ipv4-sctp : src_ip_address, dst_ip_address, src_port, dst_port,
315 ipv6-other : src_ip_address, dst_ip_address
316 ipv6-frag : src_ip_address, dst_ip_address
317 ipv6-tcp : src_ip_address, dst_ip_address, src_port, dst_port
318 ipv6-udp : src_ip_address, dst_ip_address, src_port, dst_port
319 ipv6-sctp : src_ip_address, dst_ip_address, src_port, dst_port,
321 l2_payload : ether_type
323 The flex payload is selected from offset 0 to 15 of packet's payload by default, while it is masked out from matching.
325 Start ``testpmd`` with ``--disable-rss`` and ``--pkt-filter-mode=perfect``:
327 .. code-block:: console
329 ./app/testpmd -l 0-15 -n 4 -- -i --disable-rss --pkt-filter-mode=perfect \
330 --rxq=8 --txq=8 --nb-cores=8 --nb-ports=1
332 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:
334 .. code-block:: console
336 testpmd> flow_director_filter 0 mode IP add flow ipv4-udp \
337 src 2.2.2.3 32 dst 2.2.2.5 32 vlan 0 flexbytes () \
338 fwd pf queue 1 fd_id 1
340 Check the flow director status:
342 .. code-block:: console
344 testpmd> show port fdir 0
346 ######################## FDIR infos for port 0 ####################
348 SUPPORTED FLOW TYPE: ipv4-frag ipv4-tcp ipv4-udp ipv4-sctp ipv4-other
349 ipv6-frag ipv6-tcp ipv6-udp ipv6-sctp ipv6-other
352 max_len: 16 payload_limit: 480
353 payload_unit: 2 payload_seg: 3
354 bitmask_unit: 2 bitmask_num: 2
357 src_ipv4: 0x00000000,
358 dst_ipv4: 0x00000000,
361 src_ipv6: 0x00000000,0x00000000,0x00000000,0x00000000,
362 dst_ipv6: 0x00000000,0x00000000,0x00000000,0x00000000
363 FLEX PAYLOAD SRC OFFSET:
364 L2_PAYLOAD: 0 1 2 3 4 5 6 ...
365 L3_PAYLOAD: 0 1 2 3 4 5 6 ...
366 L4_PAYLOAD: 0 1 2 3 4 5 6 ...
368 ipv4-udp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
369 ipv4-tcp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
370 ipv4-sctp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
371 ipv4-other: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
372 ipv4-frag: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
373 ipv6-udp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
374 ipv6-tcp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
375 ipv6-sctp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
376 ipv6-other: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
377 ipv6-frag: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
378 l2_payload: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
379 guarant_count: 1 best_count: 0
380 guarant_space: 512 best_space: 7168
387 Delete all flow director rules on a port:
389 .. code-block:: console
391 testpmd> flush_flow_director 0
396 The Intel® Ethernet 700 Series support a feature called
399 A Virtual Ethernet Bridge (VEB) is an IEEE Edge Virtual Bridging (EVB) term
400 for functionality that allows local switching between virtual endpoints within
401 a physical endpoint and also with an external bridge/network.
403 A "Floating" VEB doesn't have an uplink connection to the outside world so all
404 switching is done internally and remains within the host. As such, this
405 feature provides security benefits.
407 In addition, a Floating VEB overcomes a limitation of normal VEBs where they
408 cannot forward packets when the physical link is down. Floating VEBs don't need
409 to connect to the NIC port so they can still forward traffic from VF to VF
410 even when the physical link is down.
412 Therefore, with this feature enabled VFs can be limited to communicating with
413 each other but not an outside network, and they can do so even when there is
414 no physical uplink on the associated NIC port.
416 To enable this feature, the user should pass a ``devargs`` parameter to the
419 -w 84:00.0,enable_floating_veb=1
421 In this configuration the PMD will use the floating VEB feature for all the
422 VFs created by this PF device.
424 Alternatively, the user can specify which VFs need to connect to this floating
425 VEB using the ``floating_veb_list`` argument::
427 -w 84:00.0,enable_floating_veb=1,floating_veb_list=1;3-4
429 In this example ``VF1``, ``VF3`` and ``VF4`` connect to the floating VEB,
430 while other VFs connect to the normal VEB.
432 The current implementation only supports one floating VEB and one regular
433 VEB. VFs can connect to a floating VEB or a regular VEB according to the
434 configuration passed on the EAL command line.
436 The floating VEB functionality requires a NIC firmware version of 5.0
439 Dynamic Device Personalization (DDP)
440 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
442 The Intel® Ethernet 700 Series except for the Intel Ethernet Connection
443 X722 support a feature called "Dynamic Device Personalization (DDP)",
444 which is used to configure hardware by downloading a profile to support
445 protocols/filters which are not supported by default. The DDP
446 functionality requires a NIC firmware version of 6.0 or greater.
448 Current implementation supports GTP-C/GTP-U/PPPoE/PPPoL2TP,
449 steering can be used with rte_flow API.
451 GTPv1 package is released, and it can be downloaded from
452 https://downloadcenter.intel.com/download/27587.
454 PPPoE package is released, and it can be downloaded from
455 https://downloadcenter.intel.com/download/28040.
457 Load a profile which supports GTP and store backup profile:
459 .. code-block:: console
461 testpmd> ddp add 0 ./gtp.pkgo,./backup.pkgo
463 Delete a GTP profile and restore backup profile:
465 .. code-block:: console
467 testpmd> ddp del 0 ./backup.pkgo
469 Get loaded DDP package info list:
471 .. code-block:: console
473 testpmd> ddp get list 0
475 Display information about a GTP profile:
477 .. code-block:: console
479 testpmd> ddp get info ./gtp.pkgo
481 Input set configuration
482 ~~~~~~~~~~~~~~~~~~~~~~~
483 Input set for any PCTYPE can be configured with user defined configuration,
484 For example, to use only 48bit prefix for IPv6 src address for IPv6 TCP RSS:
486 .. code-block:: console
488 testpmd> port config 0 pctype 43 hash_inset clear all
489 testpmd> port config 0 pctype 43 hash_inset set field 13
490 testpmd> port config 0 pctype 43 hash_inset set field 14
491 testpmd> port config 0 pctype 43 hash_inset set field 15
493 Queue region configuration
494 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
495 The Intel® Ethernet 700 Series supports a feature of queue regions
496 configuration for RSS in the PF, so that different traffic classes or
497 different packet classification types can be separated to different
498 queues in different queue regions. There is an API for configuration
499 of queue regions in RSS with a command line. It can parse the parameters
500 of the region index, queue number, queue start index, user priority, traffic
501 classes and so on. Depending on commands from the command line, it will call
502 i40e private APIs and start the process of setting or flushing the queue
503 region configuration. As this feature is specific for i40e only private
504 APIs are used. These new ``test_pmd`` commands are as shown below. For
505 details please refer to :doc:`../testpmd_app_ug/index`.
507 .. code-block:: console
509 testpmd> set port (port_id) queue-region region_id (value) \
510 queue_start_index (value) queue_num (value)
511 testpmd> set port (port_id) queue-region region_id (value) flowtype (value)
512 testpmd> set port (port_id) queue-region UP (value) region_id (value)
513 testpmd> set port (port_id) queue-region flush (on|off)
514 testpmd> show port (port_id) queue-region
516 Limitations or Known issues
517 ---------------------------
519 MPLS packet classification
520 ~~~~~~~~~~~~~~~~~~~~~~~~~~
522 For firmware versions prior to 5.0, MPLS packets are not recognized by the NIC.
523 The L2 Payload flow type in flow director can be used to classify MPLS packet
524 by using a command in testpmd like:
526 testpmd> flow_director_filter 0 mode IP add flow l2_payload ether \
527 0x8847 flexbytes () fwd pf queue <N> fd_id <M>
529 With the NIC firmware version 5.0 or greater, some limited MPLS support
530 is added: Native MPLS (MPLS in Ethernet) skip is implemented, while no
531 new packet type, no classification or offload are possible. With this change,
532 L2 Payload flow type in flow director cannot be used to classify MPLS packet
533 as with previous firmware versions. Meanwhile, the Ethertype filter can be
534 used to classify MPLS packet by using a command in testpmd like:
536 testpmd> ethertype_filter 0 add mac_ignr 00:00:00:00:00:00 ethertype \
539 16 Byte RX Descriptor setting on DPDK VF
540 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
542 Currently the VF's RX descriptor mode is decided by PF. There's no PF-VF
543 interface for VF to request the RX descriptor mode, also no interface to notify
544 VF its own RX descriptor mode.
545 For all available versions of the i40e driver, these drivers don't support 16
546 byte RX descriptor. If the Linux i40e kernel driver is used as host driver,
547 while DPDK i40e PMD is used as the VF driver, DPDK cannot choose 16 byte receive
548 descriptor. The reason is that the RX descriptor is already set to 32 byte by
549 the i40e kernel driver. That is to say, user should keep
550 ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC=n`` in config file.
551 In the future, if the Linux i40e driver supports 16 byte RX descriptor, user
552 should make sure the DPDK VF uses the same RX descriptor mode, 16 byte or 32
553 byte, as the PF driver.
555 The same rule for DPDK PF + DPDK VF. The PF and VF should use the same RX
556 descriptor mode. Or the VF RX will not work.
558 Receive packets with Ethertype 0x88A8
559 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
561 Due to the FW limitation, PF can receive packets with Ethertype 0x88A8
562 only when floating VEB is disabled.
564 Incorrect Rx statistics when packet is oversize
565 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
567 When a packet is over maximum frame size, the packet is dropped.
568 However, the Rx statistics, when calling `rte_eth_stats_get` incorrectly
569 shows it as received.
571 VF & TC max bandwidth setting
572 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
574 The per VF max bandwidth and per TC max bandwidth cannot be enabled in parallel.
575 The behavior is different when handling per VF and per TC max bandwidth setting.
576 When enabling per VF max bandwidth, SW will check if per TC max bandwidth is
577 enabled. If so, return failure.
578 When enabling per TC max bandwidth, SW will check if per VF max bandwidth
579 is enabled. If so, disable per VF max bandwidth and continue with per TC max
582 TC TX scheduling mode setting
583 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
585 There are 2 TX scheduling modes for TCs, round robin and strict priority mode.
586 If a TC is set to strict priority mode, it can consume unlimited bandwidth.
587 It means if APP has set the max bandwidth for that TC, it comes to no
589 It's suggested to set the strict priority mode for a TC that is latency
590 sensitive but no consuming much bandwidth.
592 VF performance is impacted by PCI extended tag setting
593 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
595 To reach maximum NIC performance in the VF the PCI extended tag must be
596 enabled. The DPDK i40e PF driver will set this feature during initialization,
597 but the kernel PF driver does not. So when running traffic on a VF which is
598 managed by the kernel PF driver, a significant NIC performance downgrade has
599 been observed (for 64 byte packets, there is about 25% line-rate downgrade for
600 a 25GbE device and about 35% for a 40GbE device).
602 For kernel version >= 4.11, the kernel's PCI driver will enable the extended
603 tag if it detects that the device supports it. So by default, this is not an
604 issue. For kernels <= 4.11 or when the PCI extended tag is disabled it can be
605 enabled using the steps below.
607 #. Get the current value of the PCI configure register::
609 setpci -s <XX:XX.X> a8.w
613 value = value | 0x100
615 #. Set the PCI configure register with new value::
617 setpci -s <XX:XX.X> a8.w=<value>
622 The VF vlan strip function is only supported in the i40e kernel driver >= 2.1.26.
627 DCB works only when RSS is enabled.
629 Global configuration warning
630 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
632 I40E PMD will set some global registers to enable some function or set some
633 configure. Then when using different ports of the same NIC with Linux kernel
634 and DPDK, the port with Linux kernel will be impacted by the port with DPDK.
635 For example, register I40E_GL_SWT_L2TAGCTRL is used to control L2 tag, i40e
636 PMD uses I40E_GL_SWT_L2TAGCTRL to set vlan TPID. If setting TPID in port A
637 with DPDK, then the configuration will also impact port B in the NIC with
638 kernel driver, which don't want to use the TPID.
639 So PMD reports warning to clarify what is changed by writing global register.
641 High Performance of Small Packets on 40GbE NIC
642 ----------------------------------------------
644 As there might be firmware fixes for performance enhancement in latest version
645 of firmware image, the firmware update might be needed for getting high performance.
646 Check the Intel support website for the latest firmware updates.
647 Users should consult the release notes specific to a DPDK release to identify
648 the validated firmware version for a NIC using the i40e driver.
650 Use 16 Bytes RX Descriptor Size
651 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
653 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.
654 Configuration of ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC`` in config files can be changed to use 16 bytes size RX descriptors.
656 Example of getting best performance with l3fwd example
657 ------------------------------------------------------
659 The following is an example of running the DPDK ``l3fwd`` sample application to get high performance with a
660 server with Intel Xeon processors and Intel Ethernet CNA XL710.
662 The example scenario is to get best performance with two Intel Ethernet CNA XL710 40GbE ports.
663 See :numref:`figure_intel_perf_test_setup` for the performance test setup.
665 .. _figure_intel_perf_test_setup:
667 .. figure:: img/intel_perf_test_setup.*
669 Performance Test Setup
672 1. Add two Intel Ethernet CNA XL710 to the platform, and use one port per card to get best performance.
673 The reason for using two NICs is to overcome a PCIe v3.0 limitation since it cannot provide 80GbE bandwidth
674 for two 40GbE ports, but two different PCIe v3.0 x8 slot can.
675 Refer to the sample NICs output above, then we can select ``82:00.0`` and ``85:00.0`` as test ports::
677 82:00.0 Ethernet [0200]: Intel XL710 for 40GbE QSFP+ [8086:1583]
678 85:00.0 Ethernet [0200]: Intel XL710 for 40GbE QSFP+ [8086:1583]
680 2. Connect the ports to the traffic generator. For high speed testing, it's best to use a hardware traffic generator.
682 3. Check the PCI devices numa node (socket id) and get the cores number on the exact socket id.
683 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
685 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
686 cores from different cores (e.g core18 and core19).
688 4. Bind these two ports to igb_uio.
690 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
691 will be required, and each queue pair will need a dedicated CPU core for receiving/transmitting packets.
693 6. The DPDK sample application ``l3fwd`` will be used for performance testing, with using two ports for bi-directional forwarding.
694 Compile the ``l3fwd sample`` with the default lpm mode.
696 7. The command line of running l3fwd would be something like the following::
698 ./l3fwd -l 18-21 -n 4 -w 82:00.0 -w 85:00.0 \
699 -- -p 0x3 --config '(0,0,18),(0,1,19),(1,0,20),(1,1,21)'
701 This means that the application uses core 18 for port 0, queue pair 0 forwarding, core 19 for port 0, queue pair 1 forwarding,
702 core 20 for port 1, queue pair 0 forwarding, and core 21 for port 1, queue pair 1 forwarding.
704 8. Configure the traffic at a traffic generator.
706 * Start creating a stream on packet generator.
708 * Set the Ethernet II type to 0x0800.
710 Tx bytes affected by the link status change
711 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
713 For firmware versions prior to 6.01 for X710 series and 3.33 for X722 series, the tx_bytes statistics data is affected by
714 the link down event. Each time the link status changes to down, the tx_bytes decreases 110 bytes.