xref: /dpdk/doc/guides/rel_notes/release_18_11.rst (revision 3aaa46977de3066d862296429378f92a1fa2961d)
1..  SPDX-License-Identifier: BSD-3-Clause
2    Copyright 2018 The DPDK contributors
3
4DPDK Release 18.11
5==================
6
7New Features
8------------
9
10* **Added support for using externally allocated memory in DPDK.**
11
12  DPDK has added support for creating new ``rte_malloc`` heaps referencing
13  memory that was created outside of DPDK's own page allocator, and using that
14  memory natively with any other DPDK library or data structure.
15
16* **Added check for ensuring allocated memory is addressable by devices.**
17
18  Some devices can have addressing limitations so a new function,
19  ``rte_mem_check_dma_mask()``, has been added for checking that allocated
20  memory is not out of the device range. Since memory can now be allocated
21  dynamically after initialization, a DMA mask is stored and any new allocated
22  memory will be checked against it and rejected if it is out of range. If
23  more than one device has addressing limitations, the DMA mask is the more
24  restrictive one.
25
26* **Updated the C11 memory model version of the ring library.**
27
28  Added changes to decrease latency for architectures using the C11 memory
29  model version of the ring library.
30
31  On Cavium ThunderX2 platform, the changes decreased latency by 27-29%
32  and 3-15% for MPMC and SPSC cases respectively (with 2 lcores). The
33  real improvements may vary with the number of contending lcores and
34  the size of the ring.
35
36* **Added hot-unplug handle mechanism.**
37
38  Added ``rte_dev_hotplug_handle_enable()`` and
39  ``rte_dev_hotplug_handle_disable()`` for enabling or disabling the hotplug
40  handle mechanism.
41
42* **Added support for device multi-process hotplug.**
43
44  Added support for hotplug and hot-unplug in a multiprocessing scenario. Any
45  ethdev devices created in the primary process will be regarded as shared and
46  will be available for all DPDK processes. Synchronization between processes
47  will be done using DPDK IPC.
48
49* **Added new Flow API actions to rewrite fields in packet headers.**
50
51  Added new Flow API actions to:
52
53  * Modify source and destination IP addresses in the outermost IPv4/IPv6
54    headers.
55  * Modify source and destination port numbers in the outermost TCP/UDP
56    headers.
57
58* **Added new Flow API action to swap MAC addresses in Ethernet header.**
59
60  Added new Flow API action to swap the source and destination MAC
61  addresses in the outermost Ethernet header.
62
63* **Add support to offload more flow match and actions for CXGBE PMD.**
64
65  The Flow API support has been enhanced for the CXGBE Poll Mode Driver to
66  offload:
67
68  * Match items: destination MAC address.
69  * Action items: push/pop/rewrite vlan header,
70    rewrite IP addresses in outermost IPv4/IPv6 header,
71    rewrite port numbers in outermost TCP/UDP header,
72    swap MAC addresses in outermost Ethernet header.
73
74* **Added a devarg to use the latest supported vector path in i40e.**
75
76  A new devarg ``use-latest-supported-vec`` was introduced to allow users to
77  choose the latest vector path that the platform supported. For example, users
78  can use AVX2 vector path on BDW/HSW to get better performance.
79
80* **Added support for SR-IOV in netvsc PMD.**
81
82  The ``netvsc`` poll mode driver now supports the Accelerated Networking
83  SR-IOV option in Hyper-V and Azure. This is an alternative to the previous
84  vdev_netvsc, tap, and failsafe drivers combination.
85
86* **Added a new net driver for Marvell Armada 3k device.**
87
88  Added the new ``mvneta`` net driver for Marvell Armada 3k device. See the
89  :doc:`../nics/mvneta` NIC guide for more details on this new driver.
90
91* **Added NXP ENETC PMD.**
92
93  Added the new enetc driver for the NXP enetc platform. See the
94  :doc:`../nics/enetc` NIC driver guide for more details on this new driver.
95
96* **Added Ethernet poll mode driver for Aquantia aQtion family of 10G devices.**
97
98  Added the new ``atlantic`` ethernet poll mode driver for Aquantia XGBE devices.
99  See the :doc:`../nics/atlantic` NIC driver guide for more details on this
100  driver.
101
102* **Updated mlx5 driver.**
103
104  Updated the mlx5 driver including the following changes:
105
106  * Improved security of PMD to prevent the NIC from getting stuck when
107    the application misbehaves.
108  * Reworked flow engine to supported e-switch flow rules (transfer attribute).
109  * Added support for header re-write(L2-L4), VXLAN encap/decap, count, match
110    on TCP flags and multiple flow groups with e-switch flow rules.
111  * Added support for match on metadata, VXLAN and MPLS encap/decap with flow
112    rules.
113  * Added support for ``RTE_ETH_DEV_CLOSE_REMOVE`` flag to provide better
114    support for representors.
115  * Added support for meson build.
116  * Fixed build issue with PPC.
117  * Added support for BlueField VF.
118  * Added support for externally allocated static memory for DMA.
119
120* **Updated Solarflare network PMD.**
121
122  Updated the sfc_efx driver including the following changes:
123
124  * Added support for Rx scatter in EF10 datapath implementation.
125  * Added support for Rx descriptor status API in EF10 datapath implementation.
126  * Added support for TSO in EF10 datapath implementation.
127  * Added support for Tx descriptor status API in EF10 (ef10 and ef10_simple)
128    datapaths implementation.
129
130* **Updated the enic driver.**
131
132  * Added AVX2-based vectorized Rx handler.
133  * Added VLAN and checksum offloads to the simple Tx handler.
134  * Added the "count" flow action.
135  * Enabled the virtual address IOVA mode.
136
137* **Updated the failsafe driver.**
138
139  Updated the failsafe driver including the following changes:
140
141  * Added support for Rx and Tx queues start and stop.
142  * Added support for Rx and Tx queues deferred start.
143  * Added support for runtime Rx and Tx queues setup.
144  * Added support multicast MAC address set.
145
146* **Added a devarg to use a PCAP interface physical MAC address.**
147
148  A new devarg ``phy_mac`` was introduced to allow users to use the physical
149  MAC address of the selected PCAP interface.
150
151* **Added TAP Rx/Tx queues sharing with a secondary process.**
152
153  Added support to allow a secondary process to attach a TAP device created
154  in the primary process, probe the queues, and process Rx/Tx in a secondary
155  process.
156
157* **Added classification and metering support to SoftNIC PMD.**
158
159  Added support for flow classification (rte_flow API), and metering and
160  policing (rte_mtr API) to the SoftNIC PMD.
161
162* **Added Crypto support to the Softnic PMD.**
163
164  The Softnic is now capable of processing symmetric crypto workloads such
165  as cipher, cipher-authentication chaining, and AEAD encryption and
166  decryption. This is achieved by calling DPDK Cryptodev APIs.
167
168* **Added cryptodev port to port library.**
169
170  Cryptodev port is a shim layer in the port library that interacts with DPDK
171  Cryptodev PMDs including burst enqueuing and dequeuing crypto operations.
172
173* **Added symmetric cryptographic actions to the pipeline library.**
174
175  In the pipeline library support was added for symmetric crypto action
176  parsing and an action handler was implemented. The action allows automatic
177  preparation of the crypto operation with the rules specified such as
178  algorithm, key, and IV, etc. for the cryptodev port to process.
179
180* **Updated the AESNI MB PMD.**
181
182  The AESNI MB PMD has been updated with additional support for the AES-GCM
183  algorithm.
184
185* **Added NXP CAAM JR PMD.**
186
187  Added the new caam job ring driver for NXP platforms. See the
188  :doc:`../cryptodevs/caam_jr` guide for more details on this new driver.
189
190* **Added support for GEN3 devices to Intel QAT driver.**
191
192  Added support for the third generation of Intel QuickAssist devices.
193
194* **Updated the QAT PMD.**
195
196  The QAT PMD was updated with additional support for:
197
198  * The AES-CMAC algorithm.
199
200* **Added support for Dynamic Huffman Encoding to Intel QAT comp PMD.**
201
202  The Intel QuickAssist (QAT) compression PMD has been updated with support
203  for Dynamic Huffman Encoding for the Deflate algorithm.
204
205* **Added Event Ethernet Tx Adapter.**
206
207  Added event ethernet Tx adapter library that provides configuration and
208  data path APIs for the ethernet transmit stage of an event driven packet
209  processing application. These APIs abstract the implementation of the
210  transmit stage and allow the application to use eventdev PMD support or
211  a common implementation.
212
213* **Added Distributed Software Eventdev PMD.**
214
215  Added the new Distributed Software Event Device (DSW), which is a
216  pure-software eventdev driver distributing the work of scheduling
217  among all eventdev ports and the lcores using them. DSW, compared to
218  the SW eventdev PMD, sacrifices load balancing performance to
219  gain better event scheduling throughput and scalability.
220
221* **Added extendable bucket feature to hash library (rte_hash).**
222
223  This new "extendable bucket" feature provides 100% insertion guarantee to
224  the capacity specified by the user by extending hash table with extra
225  buckets when needed to accommodate the unlikely event of intensive hash
226  collisions. In addition, the internal hashing algorithm was changed to use
227  partial-key hashing to improve memory efficiency and lookup performance.
228
229* **Added lock free reader/writer concurrency to hash library (rte_hash).**
230
231  Lock free reader/writer concurrency prevents the readers from getting
232  blocked due to a preempted writer thread. This allows the hash library
233  to be used in scenarios where the writer thread runs on the control plane.
234
235* **Added Traffic Pattern Aware Power Control Library.**
236
237  Added an experimental library that extends the Power Library and provides
238  empty_poll APIs. This feature measures how many times empty_polls are
239  executed per core and uses the number of empty polls as a hint for system
240  power management.
241
242  See the :doc:`../prog_guide/power_man` section of the DPDK Programmers
243  Guide document for more information.
244
245* **Added JSON power policy interface for containers.**
246
247  Extended the Power Library and vm_power_manager sample app to allow power
248  policies to be submitted via a FIFO using JSON formatted strings. Previously
249  limited to Virtual Machines, this feature extends power policy functionality
250  to containers and host applications that need to have their cores frequency
251  controlled based on the rules contained in the policy.
252
253* **Added Telemetry API.**
254
255  Added a new telemetry API which allows applications to transparently expose
256  their telemetry in JSON via a UNIX socket. The JSON can be consumed by any
257  Service Assurance agent, such as CollectD.
258
259* **Updated KNI kernel module, rte_kni library, and KNI sample application.**
260
261  Updated the KNI kernel module with a new kernel module parameter,
262  ``carrier=[on|off]`` to allow the user to control the default carrier
263  state of the KNI kernel network interfaces. The default carrier state
264  is now set to ``off``, so the interfaces cannot be used until the
265  carrier state is set to ``on`` via ``rte_kni_update_link`` or
266  by writing ``1`` to ``/sys/devices/virtual/net/<iface>/carrier``.
267  In previous versions the default carrier state was left undefined.
268  See `Kernel NIC Interface <https://doc.dpdk.org/guides-18.11/sample_app_ug/kernel_nic_interface.html>`_ for more information.
269
270  Also added the new API function ``rte_kni_update_link()`` to allow the user
271  to set the carrier state of the KNI kernel network interface.
272
273  Also added a new command line flag ``-m`` to the KNI sample application to
274  monitor and automatically reflect the physical NIC carrier state to the
275  KNI kernel network interface with the new ``rte_kni_update_link()`` API.
276  See `Kernel NIC Interface <https://doc.dpdk.org/guides-18.11/sample_app_ug/kernel_nic_interface.html>`_ for more information.
277
278* **Added ability to switch queue deferred start flag on testpmd app.**
279
280  Added a console command to testpmd app, giving ability to switch
281  ``rx_deferred_start`` or ``tx_deferred_start`` flag of the specified queue of
282  the specified port. The port must be stopped before the command call in order
283  to reconfigure queues.
284
285* **Add a new sample application for vDPA.**
286
287  The vdpa sample application creates vhost-user sockets by using the
288  vDPA backend. vDPA stands for vhost Data Path Acceleration which utilizes
289  virtio ring compatible devices to serve virtio driver directly to enable
290  datapath acceleration. As vDPA driver can help to set up vhost datapath,
291  this application doesn't need to launch dedicated worker threads for vhost
292  enqueue/dequeue operations.
293
294* **Added cryptodev FIPS validation example application.**
295
296  Added an example application to parse and perform symmetric cryptography
297  computation to the NIST Cryptographic Algorithm Validation Program (CAVP)
298  test vectors.
299
300* **Allow unit test binary to take parameters from the environment.**
301
302  The unit test "test", or "dpdk-test", binary is often called from scripts,
303  which can make passing additional parameters, such as a coremask,
304  difficult. Support has been added to the application to allow it to take
305  additional command-line parameter values from the ``DPDK_TEST_PARAMS``
306  environment variable to make this application easier to use.
307
308
309API Changes
310-----------
311
312* eal: ``rte_memseg_list`` structure now has an additional flag indicating
313  whether the memseg list is externally allocated. This will have implications
314  for any users of memseg-walk-related functions, as they will now have to skip
315  externally allocated segments in most cases if the intent is to only iterate
316  over internal DPDK memory.
317
318  In addition the ``socket_id`` parameter across the entire DPDK has gained
319  additional meaning, as some socket ID's will now be representing externally
320  allocated memory. No changes will be required for existing code as backwards
321  compatibility will be kept, and those who do not use this feature will not
322  see these extra socket ID's. Any new API's must not check socket ID
323  parameters themselves, and must instead leave it to the memory subsystem to
324  decide whether socket ID is a valid one.
325
326* eal: The following devargs functions, which were deprecated in 18.05,
327  were removed in 18.11:
328  ``rte_eal_parse_devargs_str()``, ``rte_eal_devargs_add()``,
329  ``rte_eal_devargs_type_count()``, and ``rte_eal_devargs_dump()``.
330
331* eal: The parameters of the function ``rte_devargs_remove()`` have changed
332  from bus and device names to ``struct rte_devargs``.
333
334* eal: The deprecated functions attach/detach were removed in 18.11.
335  ``rte_eal_dev_attach`` can be replaced by
336  ``rte_dev_probe`` or ``rte_eal_hotplug_add``.
337  ``rte_eal_dev_detach`` can be replaced by
338  ``rte_dev_remove`` or ``rte_eal_hotplug_remove``.
339
340* eal: The scope of ``rte_eal_hotplug_add()``/``rte_dev_probe()``
341  and ``rte_eal_hotplug_remove()``/``rte_dev_remove()`` has been extended.
342  In the multi-process model, they will guarantee that the device is
343  attached or detached on all processes.
344
345* mbuf: The ``__rte_mbuf_raw_free()`` and ``__rte_pktmbuf_prefree_seg()``
346  functions were deprecated since 17.05 and are replaced by
347  ``rte_mbuf_raw_free()`` and ``rte_pktmbuf_prefree_seg()``.
348
349* ethdev: The deprecated functions attach/detach were removed in 18.11.
350  ``rte_eth_dev_attach()`` can be replaced by ``RTE_ETH_FOREACH_MATCHING_DEV``
351  and ``rte_dev_probe()`` or ``rte_eal_hotplug_add()``.
352  ``rte_eth_dev_detach()`` can be replaced by
353  ``rte_dev_remove()`` or ``rte_eal_hotplug_remove()``.
354
355* ethdev: A call to ``rte_eth_dev_release_port()`` has been added in
356  ``rte_eth_dev_close()``. As a consequence, a closed port is freed
357  and seen as invalid because of its state ``RTE_ETH_DEV_UNUSED``.
358  This new behavior is enabled per driver for a migration period.
359
360* A new device flag, ``RTE_ETH_DEV_NOLIVE_MAC_ADDR``, changes the order of
361  actions inside ``rte_eth_dev_start()`` regarding MAC set. Some NICs do not
362  support MAC changes once the port has started and with this new device
363  flag the MAC can be properly configured in any case. This is particularly
364  important for bonding.
365
366* The default behavior of CRC strip offload has changed in this
367  release. Without any specific Rx offload flag, default behavior by a PMD is
368  now to strip CRC. ``DEV_RX_OFFLOAD_CRC_STRIP`` offload flag has been removed.
369  To request keeping CRC, application should set ``DEV_RX_OFFLOAD_KEEP_CRC``
370  Rx offload.
371
372* eventdev: The type of the second parameter to
373  ``rte_event_eth_rx_adapter_caps_get()`` has been changed from uint8_t to
374  uint16_t.
375
376* kni: By default, interface carrier status is ``off`` which means there won't
377  be any traffic. It can be set to ``on`` via ``rte_kni_update_link()`` API or
378  via ``sysfs`` interface: ``echo 1 > /sys/class/net/vEth0/carrier``.
379
380  Note interface should be ``up`` to be able to read/write sysfs interface.
381  When KNI sample application is used, ``-m`` parameter can be used to
382  automatically update the carrier status for the interface.
383
384* kni: When ethtool support is enabled (``CONFIG_RTE_KNI_KMOD_ETHTOOL=y``)
385  ethtool commands ``ETHTOOL_GSET & ETHTOOL_SSET`` are no longer supported for
386  kernels that have ``ETHTOOL_GLINKSETTINGS & ETHTOOL_SLINKSETTINGS`` support.
387  This means ``ethtool "-a|--show-pause", "-s|--change"`` won't work, and
388  ``ethtool <iface>`` output will have less information.
389
390
391ABI Changes
392-----------
393
394* eal: added ``legacy_mem`` and ``single_file_segments`` values to
395  ``rte_config`` structure on account of improving DPDK usability when
396  using either ``--legacy-mem`` or ``--single-file-segments`` flags.
397
398* eal: EAL library ABI version was changed due to previously announced work on
399  supporting external memory in DPDK:
400
401  - Structure ``rte_memseg_list`` now has a new field indicating length
402    of memory addressed by the segment list
403  - Structure ``rte_memseg_list`` now has a new flag indicating whether
404    the memseg list refers to external memory
405  - Structure ``rte_malloc_heap`` now has a new field indicating socket
406    ID the malloc heap belongs to
407  - Structure ``rte_mem_config`` has had its ``malloc_heaps`` array
408    resized from ``RTE_MAX_NUMA_NODES`` to ``RTE_MAX_HEAPS`` value
409  - Structure ``rte_malloc_heap`` now has a ``heap_name`` member
410  - Structure ``rte_eal_memconfig`` has been extended to contain next
411    socket ID for externally allocated segments
412
413* eal: Added ``dma_maskbits`` to ``rte_mem_config`` for keeping the most
414  restrictive DMA mask based on the devices addressing limitations.
415
416* eal: The structure ``rte_device`` has a new field to reference a
417  ``rte_bus``.  It thus changes the size of the ``struct rte_device`` and the
418  inherited device structures of all buses.
419
420
421Shared Library Versions
422-----------------------
423
424The libraries prepended with a plus sign were incremented in this version.
425
426.. code-block:: diff
427
428     librte_acl.so.2
429     librte_bbdev.so.1
430     librte_bitratestats.so.2
431     librte_bpf.so.1
432   + librte_bus_dpaa.so.2
433   + librte_bus_fslmc.so.2
434   + librte_bus_ifpga.so.2
435   + librte_bus_pci.so.2
436   + librte_bus_vdev.so.2
437   + librte_bus_vmbus.so.2
438     librte_cfgfile.so.2
439     librte_cmdline.so.2
440     librte_compressdev.so.1
441     librte_cryptodev.so.5
442     librte_distributor.so.1
443   + librte_eal.so.9
444     librte_efd.so.1
445   + librte_ethdev.so.11
446   + librte_eventdev.so.6
447     librte_flow_classify.so.1
448     librte_gro.so.1
449     librte_gso.so.1
450     librte_hash.so.2
451     librte_ip_frag.so.1
452     librte_jobstats.so.1
453     librte_kni.so.2
454     librte_kvargs.so.1
455     librte_latencystats.so.1
456     librte_lpm.so.2
457     librte_mbuf.so.4
458     librte_member.so.1
459     librte_mempool.so.5
460     librte_meter.so.2
461     librte_metrics.so.1
462     librte_net.so.1
463     librte_pci.so.1
464     librte_pdump.so.2
465     librte_pipeline.so.3
466     librte_pmd_bnxt.so.2
467     librte_pmd_bond.so.2
468     librte_pmd_i40e.so.2
469     librte_pmd_ixgbe.so.2
470     librte_pmd_dpaa2_qdma.so.1
471     librte_pmd_ring.so.2
472     librte_pmd_softnic.so.1
473     librte_pmd_vhost.so.2
474     librte_port.so.3
475     librte_power.so.1
476     librte_rawdev.so.1
477     librte_reorder.so.1
478     librte_ring.so.2
479     librte_sched.so.1
480     librte_security.so.1
481     librte_table.so.3
482     librte_timer.so.1
483     librte_vhost.so.4
484
485
486Known Issues
487------------
488
489* When using SR-IOV (VF) support with netvsc PMD and the Mellanox mlx5
490  bifurcated driver the Linux netvsc device must be brought up before the
491  netvsc device is unbound and passed to the DPDK.
492
493* IBM Power8 is not supported in this release of DPDK. IBM Power9 is
494  supported.
495
496* ``AVX-512`` support has been disabled for ``GCC`` builds [1] because of a
497  crash [2]. This can affect ``native`` machine type build targets on the
498  platforms that support ``AVX512F`` like ``Intel Skylake`` processors, and
499  can cause a possible performance drop. The immediate workaround is to use
500  ``clang`` compiler on these platforms. The issue has been identified as a
501  GCC defect and reported to the GCC community [3]. Further actions will be
502  taken based on the GCC defect result.
503
504  - [1]: Commit 8d07c82b239f ("mk: disable gcc AVX512F support")
505  - [2]: https://bugs.dpdk.org/show_bug.cgi?id=97
506  - [3]: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88096
507
508
509Tested Platforms
510----------------
511
512* Intel(R) platforms with Intel(R) NICs combinations
513
514   * CPU
515
516     * Intel(R) Atom(TM) CPU C3758 @ 2.20GHz
517     * Intel(R) Xeon(R) CPU D-1541 @ 2.10GHz
518     * Intel(R) Xeon(R) CPU E5-2680 v2 @ 2.80GHz
519     * Intel(R) Xeon(R) CPU E5-2699 v3 @ 2.30GHz
520     * Intel(R) Xeon(R) CPU E5-2699 v4 @ 2.20GHz
521     * Intel(R) Xeon(R) Platinum 8180 CPU @ 2.50GHz
522
523   * OS:
524
525     * CentOS 7.5
526     * Fedora 25
527     * Fedora 28
528     * FreeBSD 11.2
529     * Red Hat Enterprise Linux Server release 7.5
530     * Open SUSE 15
531     * Wind River Linux 8
532     * Ubuntu 14.04
533     * Ubuntu 16.04
534     * Ubuntu 16.10
535     * Ubuntu 17.10
536     * Ubuntu 18.04
537
538   * NICs:
539
540     * Intel(R) 82599ES 10 Gigabit Ethernet Controller
541
542       * Firmware version: 0x61bf0001
543       * Device id (pf/vf): 8086:10fb / 8086:10ed
544       * Driver version: 5.2.3 (ixgbe)
545
546     * Intel(R) Corporation Ethernet Connection X552/X557-AT 10GBASE-T
547
548       * Firmware version: 0x800003e7
549       * Device id (pf/vf): 8086:15ad / 8086:15a8
550       * Driver version: 4.4.6 (ixgbe)
551
552     * Intel(R) Ethernet Converged Network Adapter X710-DA4 (4x10G)
553
554       * Firmware version: 6.01 0x80003221
555       * Device id (pf/vf): 8086:1572 / 8086:154c
556       * Driver version: 2.4.6 (i40e)
557
558     * Intel(R) Corporation Ethernet Connection X722 for 10GbE SFP+ (4x10G)
559
560       * Firmware version: 3.33 0x80000fd5 0.0.0
561       * Device id (pf/vf): 8086:37d0 / 8086:37cd
562       * Driver version: 2.4.6 (i40e)
563
564     * Intel(R) Ethernet Converged Network Adapter XXV710-DA2 (2x25G)
565
566       * Firmware version: 6.01 0x80003221
567       * Device id (pf/vf): 8086:158b / 8086:154c
568       * Driver version: 2.4.6 (i40e)
569
570     * Intel(R) Ethernet Converged Network Adapter XL710-QDA2 (2X40G)
571
572       * Firmware version: 6.01 0x8000321c
573       * Device id (pf/vf): 8086:1583 / 8086:154c
574       * Driver version: 2.4.6 (i40e)
575
576     * Intel(R) Corporation I350 Gigabit Network Connection
577
578       * Firmware version: 1.63, 0x80000dda
579       * Device id (pf/vf): 8086:1521 / 8086:1520
580       * Driver version: 5.4.0-k (igb)
581
582* Intel(R) platforms with Mellanox(R) NICs combinations
583
584   * CPU:
585
586     * Intel(R) Xeon(R) Gold 6154 CPU @ 3.00GHz
587     * Intel(R) Xeon(R) CPU E5-2697A v4 @ 2.60GHz
588     * Intel(R) Xeon(R) CPU E5-2697 v3 @ 2.60GHz
589     * Intel(R) Xeon(R) CPU E5-2680 v2 @ 2.80GHz
590     * Intel(R) Xeon(R) CPU E5-2650 v4 @ 2.20GHz
591     * Intel(R) Xeon(R) CPU E5-2640 @ 2.50GHz
592     * Intel(R) Xeon(R) CPU E5-2620 v4 @ 2.10GHz
593
594   * OS:
595
596     * Red Hat Enterprise Linux Server release 7.6 (Maipo)
597     * Red Hat Enterprise Linux Server release 7.5 (Maipo)
598     * Red Hat Enterprise Linux Server release 7.4 (Maipo)
599     * Red Hat Enterprise Linux Server release 7.3 (Maipo)
600     * Red Hat Enterprise Linux Server release 7.2 (Maipo)
601     * Ubuntu 18.10
602     * Ubuntu 18.04
603     * Ubuntu 17.10
604     * Ubuntu 16.04
605     * SUSE Linux Enterprise Server 15
606
607   * MLNX_OFED: 4.4-2.0.1.0
608   * MLNX_OFED: 4.5-0.3.1.0
609
610   * NICs:
611
612     * Mellanox(R) ConnectX(R)-3 Pro 40G MCX354A-FCC_Ax (2x40G)
613
614       * Host interface: PCI Express 3.0 x8
615       * Device ID: 15b3:1007
616       * Firmware version: 2.42.5000
617
618     * Mellanox(R) ConnectX(R)-4 10G MCX4111A-XCAT (1x10G)
619
620       * Host interface: PCI Express 3.0 x8
621       * Device ID: 15b3:1013
622       * Firmware version: 12.23.8022 and above
623
624     * Mellanox(R) ConnectX(R)-4 10G MCX4121A-XCAT (2x10G)
625
626       * Host interface: PCI Express 3.0 x8
627       * Device ID: 15b3:1013
628       * Firmware version: 12.23.8022 and above
629
630     * Mellanox(R) ConnectX(R)-4 25G MCX4111A-ACAT (1x25G)
631
632       * Host interface: PCI Express 3.0 x8
633       * Device ID: 15b3:1013
634       * Firmware version: 12.23.8022 and above
635
636     * Mellanox(R) ConnectX(R)-4 25G MCX4121A-ACAT (2x25G)
637
638       * Host interface: PCI Express 3.0 x8
639       * Device ID: 15b3:1013
640       * Firmware version: 12.23.8022 and above
641
642     * Mellanox(R) ConnectX(R)-4 40G MCX4131A-BCAT/MCX413A-BCAT (1x40G)
643
644       * Host interface: PCI Express 3.0 x8
645       * Device ID: 15b3:1013
646       * Firmware version: 12.23.8022 and above
647
648     * Mellanox(R) ConnectX(R)-4 40G MCX415A-BCAT (1x40G)
649
650       * Host interface: PCI Express 3.0 x16
651       * Device ID: 15b3:1013
652       * Firmware version: 12.23.8022 and above
653
654     * Mellanox(R) ConnectX(R)-4 50G MCX4131A-GCAT/MCX413A-GCAT (1x50G)
655
656       * Host interface: PCI Express 3.0 x8
657       * Device ID: 15b3:1013
658       * Firmware version: 12.23.8022 and above
659
660     * Mellanox(R) ConnectX(R)-4 50G MCX414A-BCAT (2x50G)
661
662       * Host interface: PCI Express 3.0 x8
663       * Device ID: 15b3:1013
664       * Firmware version: 12.23.8022 and above
665
666     * Mellanox(R) ConnectX(R)-4 50G MCX415A-GCAT/MCX416A-BCAT/MCX416A-GCAT (2x50G)
667
668       * Host interface: PCI Express 3.0 x16
669       * Device ID: 15b3:1013
670       * Firmware version: 12.23.8022 and above
671       * Firmware version: 12.23.8022 and above
672
673     * Mellanox(R) ConnectX(R)-4 50G MCX415A-CCAT (1x100G)
674
675       * Host interface: PCI Express 3.0 x16
676       * Device ID: 15b3:1013
677       * Firmware version: 12.23.8022 and above
678
679     * Mellanox(R) ConnectX(R)-4 100G MCX416A-CCAT (2x100G)
680
681       * Host interface: PCI Express 3.0 x16
682       * Device ID: 15b3:1013
683       * Firmware version: 12.23.8022 and above
684
685     * Mellanox(R) ConnectX(R)-4 Lx 10G MCX4121A-XCAT (2x10G)
686
687       * Host interface: PCI Express 3.0 x8
688       * Device ID: 15b3:1015
689       * Firmware version: 14.23.8022 and above
690
691     * Mellanox(R) ConnectX(R)-4 Lx 25G MCX4121A-ACAT (2x25G)
692
693       * Host interface: PCI Express 3.0 x8
694       * Device ID: 15b3:1015
695       * Firmware version: 14.23.8022 and above
696
697     * Mellanox(R) ConnectX(R)-5 100G MCX556A-ECAT (2x100G)
698
699       * Host interface: PCI Express 3.0 x16
700       * Device ID: 15b3:1017
701       * Firmware version: 16.23.8022 and above
702
703     * Mellanox(R) ConnectX(R)-5 Ex EN 100G MCX516A-CDAT (2x100G)
704
705       * Host interface: PCI Express 4.0 x16
706       * Device ID: 15b3:1019
707       * Firmware version: 16.23.8022 and above
708
709* ARM platforms with Mellanox(R) NICs combinations
710
711   * CPU:
712
713     * Qualcomm ARM 1.1 2500MHz
714
715   * OS:
716
717     * Red Hat Enterprise Linux Server release 7.5 (Maipo)
718
719   * NICs:
720
721     * Mellanox(R) ConnectX(R)-4 Lx 25G MCX4121A-ACAT (2x25G)
722
723       * Host interface: PCI Express 3.0 x8
724       * Device ID: 15b3:1015
725       * Firmware version: 14.24.0220
726
727     * Mellanox(R) ConnectX(R)-5 100G MCX556A-ECAT (2x100G)
728
729       * Host interface: PCI Express 3.0 x16
730       * Device ID: 15b3:1017
731       * Firmware version: 16.24.0220
732
733* Mellanox(R) BlueField SmartNIC
734
735   * Mellanox(R) BlueField SmartNIC MT416842 (2x25G)
736
737       * Host interface: PCI Express 3.0 x16
738       * Device ID: 15b3:a2d2
739       * Firmware version: 18.24.0246
740
741   * SoC ARM cores running OS:
742
743     * CentOS Linux release 7.4.1708 (AltArch)
744     * MLNX_OFED 4.4-2.5.3.0
745
746  * DPDK application running on ARM cores inside SmartNIC
747
748* ARM SoC combinations from NXP (with integrated NICs)
749
750   * SoC:
751
752     * NXP/Freescale QorIQ LS1046A with ARM Cortex A72
753     * NXP/Freescale QorIQ LS2088A with ARM Cortex A72
754
755   * OS:
756
757     * Ubuntu 18.04.1 LTS with NXP QorIQ LSDK 1809 support packages
758     * Ubuntu 16.04.3 LTS with NXP QorIQ LSDK 1803 support packages
759