1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright 2017 The DPDK contributors
10 * **Extended port_id range from uint8_t to uint16_t.**
12 Increased the ``port_id`` range from 8 bits to 16 bits in order to support
13 more than 256 ports in DPDK. All ethdev APIs which have ``port_id`` as
14 parameter have been changed.
16 * **Modified the return type of rte_eth_stats_reset.**
18 Changed return type of ``rte_eth_stats_reset`` from ``void`` to ``int`` so
19 that the caller can determine whether a device supports the operation or not
20 and if the operation was carried out.
22 * **Added a new driver for Marvell Armada 7k/8k devices.**
24 Added the new ``mrvl`` net driver for Marvell Armada 7k/8k devices. See the
25 :doc:`../nics/mvpp2` NIC guide for more details on this new driver.
27 * **Updated mlx4 driver.**
29 Updated the mlx4 driver including the following changes:
31 * Isolated mode (rte_flow) can now be enabled anytime, not only during
32 initial device configuration.
33 * Flow rules now support up to 4096 priority levels usable at will by
35 * Enhanced error message to help debugging invalid/unsupported flow rules.
36 * Flow rules matching all multicast and promiscuous traffic are now allowed.
37 * No more software restrictions on flow rules with the RSS action, their
38 configuration is much more flexible.
39 * Significantly reduced memory footprint for Rx and Tx queue objects.
40 * While supported, UDP RSS is temporarily disabled due to a remaining issue
41 with its support in the Linux kernel.
42 * The new RSS implementation does not automatically spread traffic according
43 to the inner packet of VXLAN frames anymore, only the outer one (like
45 * Partial (Tx only) support for secondary processes was broken and had to be
47 * Refactored driver to get rid of dependency on the components provided by
48 Mellanox OFED and instead rely on the current and public rdma-core
49 package and Linux version from now on.
50 * Removed compile-time limitation on number of device instances the PMD
53 * **Updated mlx5 driver.**
55 Updated the mlx5 driver including the following changes:
57 * Enabled the PMD to run on top of upstream Linux kernel and rdma-core
58 libs, removing the dependency on specific Mellanox OFED libraries.
59 * Improved PMD latency performance.
60 * Improved PMD memory footprint.
61 * Added support for vectorized Rx/Tx burst for ARMv8.
62 * Added support for secondary process.
63 * Added support for flow counters.
64 * Added support for Rx hardware timestamp offload.
65 * Added support for device removal event.
67 * **Added SoftNIC PMD.**
69 Added a new SoftNIC PMD. This virtual device provides applications with
70 software fallback support for traffic management.
72 * **Added support for NXP DPAA Devices.**
74 Added support for NXP's DPAA devices - LS104x series. This includes:
77 * DPAA Mempool driver for supporting offloaded packet memory pool
78 * DPAA PMD for DPAA devices
80 See the :doc:`../nics/dpaa` document for more details of this new driver.
82 * **Updated support for Cavium OCTEONTX Device.**
84 Updated support for Cavium's OCTEONTX device (CN83xx). This includes:
86 * OCTEONTX Mempool driver for supporting offloaded packet memory pool
88 * OCTEONTX Eventdev-Ethdev Rx adapter
90 See the :doc:`../nics/octeontx` document for more details of this new driver.
92 * **Added PF support to the Netronome NFP PMD.**
94 Added PF support to the Netronome NFP PMD. Previously the NFP PMD only
95 supported VFs. PF support is just as a basic DPDK port and has no VF
98 PF support comes with firmware upload support which allows the PMD to
99 independently work from kernel netdev NFP drivers.
101 NFP 4000 devices are also now supported along with previous 6000 devices.
103 * **Updated bnxt PMD.**
105 Major enhancements include:
107 * Support for Flow API
108 * Support for Tx and Rx descriptor status functions
110 * **Added bus agnostic functions to cryptodev for PMD initialization**
112 Added new PMD assist, bus independent, functions
113 ``rte_cryptodev_pmd_parse_input_args()``, ``rte_cryptodev_pmd_create()`` and
114 ``rte_cryptodev_pmd_destroy()`` for drivers to manage creation and
115 destruction of new device instances.
117 * **Updated QAT crypto PMD.**
119 Added several performance enhancements:
121 * Removed atomics from the internal queue pair structure.
122 * Added coalesce writes to HEAD CSR on response processing.
123 * Added coalesce writes to TAIL CSR on request processing.
125 In addition support was added for the AES CCM algorithm.
127 * **Updated the AESNI MB PMD.**
129 The AESNI MB PMD has been updated with additional support for:
131 * The DES CBC algorithm.
132 * The DES DOCSIS BPI algorithm.
134 This change requires version 0.47 of the IPsec Multi-buffer library. For
135 more details see the :doc:`../cryptodevs/aesni_mb` documentation.
137 * **Updated the OpenSSL PMD.**
139 The OpenSSL PMD has been updated with additional support for:
141 * The DES CBC algorithm.
142 * The AES CCM algorithm.
144 * **Added NXP DPAA SEC crypto PMD.**
146 A new ``dpaa_sec`` hardware based crypto PMD for NXP DPAA devices has been
147 added. See the :doc:`../cryptodevs/dpaa_sec` document for more details.
149 * **Added MRVL crypto PMD.**
151 A new crypto PMD has been added, which provides several ciphering and hashing
152 algorithms. All cryptography operations use the MUSDK library crypto API.
153 See the :doc:`../cryptodevs/mvsam` document for more details.
155 * **Add new benchmarking mode to dpdk-test-crypto-perf application.**
157 Added a new "PMD cyclecount" benchmark mode to the ``dpdk-test-crypto-perf``
158 application to display a detailed breakdown of CPU cycles used by hardware
161 * **Added the Security Offload Library.**
163 Added an experimental library - ``rte_security``. This provide security APIs
164 for protocols like IPsec using inline ipsec offload to ethernet devices or
165 full protocol offload with lookaside crypto devices.
167 See the :doc:`../prog_guide/rte_security` section of the DPDK Programmers
168 Guide document for more information.
170 * **Updated the DPAA2_SEC crypto driver to support rte_security.**
172 Updated the ``dpaa2_sec`` crypto PMD to support ``rte_security`` lookaside
173 protocol offload for IPsec.
175 * **Updated the IXGBE ethernet driver to support rte_security.**
177 Updated ixgbe ethernet PMD to support ``rte_security`` inline IPsec offload.
179 * **Updated i40e driver to support GTP-C/GTP-U.**
181 Updated i40e PMD to support GTP-C/GTP-U with GTP-C/GTP-U supporting
182 profiles which can be programmed by dynamic device personalization (DDP)
185 * **Added the i40e ethernet driver to support queue region feature.**
187 This feature enable queue regions configuration for RSS in PF,
188 so that different traffic classes or different packet
189 classification types can be separated into different queues in
190 different queue regions.
192 * **Updated ipsec-secgw application to support rte_security.**
194 Updated the ``ipsec-secgw`` sample application to support ``rte_security``
195 actions for ipsec inline and full protocol offload using lookaside crypto
198 * **Added IOMMU support to libvhost-user**
200 Implemented device IOTLB in the Vhost-user backend, and enabled Virtio's
201 IOMMU feature. The feature is disabled by default, and can be enabled by
202 setting ``RTE_VHOST_USER_IOMMU_SUPPORT`` flag at vhost device registration
205 * **Added the Event Ethernet Adapter Library.**
207 Added the Event Ethernet Adapter library. This library provides APIs for
208 eventdev applications to configure the ethdev for eventdev packet flow.
210 * **Updated DPAA2 Event PMD for the Event Ethernet Adapter.**
212 Added support for the eventdev ethernet adapter for DPAA2.
214 * **Added Membership library (rte_member).**
216 Added a new data structure library called the Membership Library.
218 The Membership Library is an extension and generalization of a traditional
219 filter (for example Bloom Filter) structure that has multiple usages in a
220 wide variety of workloads and applications. In general, the Membership
221 Library is a data structure that provides a "set-summary" and responds to
222 set-membership queries whether a certain member belongs to a set(s).
224 The library provides APIs for DPDK applications to insert a new member,
225 delete an existing member, and query the existence of a member in a given
226 set, or a group of sets. For the case of a group of sets the library will
227 return not only whether the element has been inserted in one of the sets but
228 also which set it belongs to.
230 See the :doc:`../prog_guide/member_lib` documentation in the Programmers
231 Guide, for more information.
233 * **Added the Generic Segmentation Offload Library.**
235 Added the Generic Segmentation Offload (GSO) library to enable
236 applications to split large packets (e.g. MTU is 64KB) into small
237 ones (e.g. MTU is 1500B). Supported packet types are:
240 * VxLAN packets, which must have an outer IPv4 header, and contain
241 an inner TCP/IPv4 packet.
242 * GRE packets, which must contain an outer IPv4 header, and inner
245 The GSO library doesn't check if the input packets have correct
246 checksums, and doesn't update checksums for output packets.
247 Additionally, the GSO library doesn't process IP fragmented packets.
249 * **Added the Flow Classification Library.**
251 Added an experimental Flow Classification library to provide APIs for DPDK
252 applications to classify an input packet by matching it against a set of
253 flow rules. It uses the ``librte_table`` API to manage the flow rules.
259 * **Service core fails to call service callback due to atomic lock**
261 In a specific configuration of multi-thread unsafe services and service
262 cores, a service core previously did not correctly release the atomic lock
263 on the service. This would result in the cores polling the service, but it
264 looked like another thread was executing the service callback. The logic for
265 atomic locking of the services has been fixed and refactored for readability.
271 * **Ethdev device name length increased.**
273 The size of internal device name has been increased to 64 characters
274 to allow for storing longer bus specific names.
276 * **Removed the Ethdev RTE_ETH_DEV_DETACHABLE flag.**
278 Removed the Ethdev ``RTE_ETH_DEV_DETACHABLE`` flag. This flag is not
279 required anymore, with the new hotplug implementation. It has been removed
280 from the ether library. Its semantics are now expressed at the bus and PMD
283 * **Service cores API updated for usability**
285 The service cores API has been changed, removing pointers from the API where
286 possible, and instead using integer IDs to identify each service. This
287 simplifies application code, aids debugging, and provides better
288 encapsulation. A summary of the main changes made is as follows:
290 * Services identified by ID not by ``rte_service_spec`` pointer
291 * Reduced API surface by using ``set`` functions instead of enable/disable
292 * Reworked ``rte_service_register`` to provide the service ID to registrar
293 * Reworked start and stop APIs into ``rte_service_runstate_set``
294 * Added API to set runstate of service implementation to indicate readiness
296 * **The following changes have been made in the mempool library**
298 * Moved ``flags`` datatype from ``int`` to ``unsigned int`` for
300 * Removed ``__rte_unused int flag`` param from ``rte_mempool_generic_put``
301 and ``rte_mempool_generic_get`` API.
302 * Added ``flags`` param in ``rte_mempool_xmem_size`` and
303 ``rte_mempool_xmem_usage``.
304 * ``rte_mem_phy2mch`` was used in Xen dom0 to obtain the physical address;
305 remove this API as Xen dom0 support was removed.
307 * **Added IOVA aliases related to physical address handling.**
309 Some data types, structure members and functions related to physical address
310 handling are deprecated and have new aliases with IOVA wording. For example:
312 * ``phys_addr_t`` can be often replaced by ``rte_iova_t`` of same size.
313 * ``RTE_BAD_PHYS_ADDR`` is often replaced by ``RTE_BAD_IOVA`` of same value.
314 * ``rte_memseg.phys_addr`` is aliased with ``rte_memseg.iova_addr``.
315 * ``rte_mem_virt2phy()`` can often be replaced by ``rte_mem_virt2iova``.
316 * ``rte_malloc_virt2phy`` is aliased with ``rte_malloc_virt2iova``.
317 * ``rte_memzone.phys_addr`` is aliased with ``rte_memzone.iova``.
318 * ``rte_mempool_objhdr.physaddr`` is aliased with
319 ``rte_mempool_objhdr.iova``.
320 * ``rte_mempool_memhdr.phys_addr`` is aliased with
321 ``rte_mempool_memhdr.iova``.
322 * ``rte_mempool_virt2phy()`` can be replaced by ``rte_mempool_virt2iova()``.
323 * ``rte_mempool_populate_phys*()`` are aliased with
324 ``rte_mempool_populate_iova*()``
325 * ``rte_mbuf.buf_physaddr`` is aliased with ``rte_mbuf.buf_iova``.
326 * ``rte_mbuf_data_dma_addr*()`` are aliased with ``rte_mbuf_data_iova*()``.
327 * ``rte_pktmbuf_mtophys*`` are aliased with ``rte_pktmbuf_iova*()``.
329 * **PCI bus API moved outside of the EAL**
331 The PCI bus previously implemented within the EAL has been moved.
332 A first part has been added as an RTE library providing PCI helpers to
333 parse device locations or other such utilities.
334 A second part consisting of the actual bus driver has been moved to its
335 proper subdirectory, without changing its functionalities.
337 As such, several PCI-related functions are not exposed by the EAL anymore:
341 * ``rte_pci_ioport_map``
342 * ``rte_pci_ioport_read``
343 * ``rte_pci_ioport_unmap``
344 * ``rte_pci_ioport_write``
345 * ``rte_pci_map_device``
347 * ``rte_pci_probe_one``
348 * ``rte_pci_read_config``
349 * ``rte_pci_register``
351 * ``rte_pci_unmap_device``
352 * ``rte_pci_unregister``
353 * ``rte_pci_write_config``
355 These functions are made available either as part of ``librte_pci`` or
358 * **Moved vdev bus APIs outside of the EAL**
360 Moved the following APIs from ``librte_eal`` to ``librte_bus_vdev``:
363 * ``rte_vdev_register``
364 * ``rte_vdev_uninit``
365 * ``rte_vdev_unregister``
367 * **Add return value to stats_get dev op API**
369 The ``stats_get`` dev op API return value has been changed to be int.
370 In this way PMDs can return an error value in case of failure at stats
371 getting process time.
373 * **Modified the rte_cryptodev_allocate_driver function.**
375 Modified the ``rte_cryptodev_allocate_driver()`` function in the cryptodev
376 library. An extra parameter ``struct cryptodev_driver *crypto_drv`` has been
379 * **Removed virtual device bus specific functions from librte_cryptodev.**
381 The functions ``rte_cryptodev_vdev_parse_init_params()`` and
382 ``rte_cryptodev_vdev_pmd_init()`` have been removed from librte_cryptodev
383 and have been replaced by non bus specific functions
384 ``rte_cryptodev_pmd_parse_input_args()`` and ``rte_cryptodev_pmd_create()``.
386 The ``rte_cryptodev_create_vdev()`` function was removed to avoid the
387 dependency on vdev in librte_cryptodev; instead, users can call
388 ``rte_vdev_init()`` directly.
390 * **Removed PCI device bus specific functions from librte_cryptodev.**
392 The functions ``rte_cryptodev_pci_generic_probe()`` and
393 ``rte_cryptodev_pci_generic_remove()`` have been removed from librte_cryptodev
394 and have been replaced by non bus specific functions
395 ``rte_cryptodev_pmd_create()`` and ``rte_cryptodev_pmd_destroy()``.
397 * **Removed deprecated functions to manage log level or type.**
399 The functions ``rte_set_log_level()``, ``rte_get_log_level()``,
400 ``rte_set_log_type()`` and ``rte_get_log_type()`` have been removed.
402 They are respectively replaced by ``rte_log_set_global_level()``,
403 ``rte_log_get_global_level()``, ``rte_log_set_level()`` and
404 ``rte_log_get_level()``.
406 * **Removed mbuf flags PKT_RX_VLAN_PKT and PKT_RX_QINQ_PKT.**
408 The ``mbuf`` flags ``PKT_RX_VLAN_PKT`` and ``PKT_RX_QINQ_PKT`` have
409 been removed since their behavior was not properly described.
411 * **Added mbuf flags PKT_RX_VLAN and PKT_RX_QINQ.**
413 Two ``mbuf`` flags have been added to indicate that the VLAN
414 identifier has been saved in the ``mbuf`` structure. For instance:
416 - If VLAN is not stripped and TCI is saved: ``PKT_RX_VLAN``
417 - If VLAN is stripped and TCI is saved: ``PKT_RX_VLAN | PKT_RX_VLAN_STRIPPED``
419 * **Modified the vlan_offload_set_t function prototype in the ethdev library.**
421 Modified the ``vlan_offload_set_t`` function prototype in the ethdev
422 library. The return value has been changed from ``void`` to ``int`` so the
423 caller can determine whether the backing device supports the operation or if
424 the operation was successfully performed.
430 * **Extended port_id range.**
432 The size of the field ``port_id`` in the ``rte_eth_dev_data`` structure
433 has changed, as described in the `New Features` section above.
435 * **New parameter added to rte_eth_dev.**
437 A new parameter ``security_ctx`` has been added to ``rte_eth_dev`` to
438 support security operations like IPsec inline.
440 * **New parameter added to rte_cryptodev.**
442 A new parameter ``security_ctx`` has been added to ``rte_cryptodev`` to
443 support security operations like lookaside crypto.
449 * Xen dom0 in EAL has been removed, as well as the xenvirt PMD and vhost_xen.
451 * The crypto performance unit tests have been removed,
452 replaced by the ``dpdk-test-crypto-perf`` application.
455 Shared Library Versions
456 -----------------------
458 The libraries prepended with a plus sign were incremented in this version.
463 + librte_bitratestats.so.2
464 + librte_bus_dpaa.so.1
465 + librte_bus_fslmc.so.1
466 + librte_bus_pci.so.1
467 + librte_bus_vdev.so.1
470 + librte_cryptodev.so.4
471 librte_distributor.so.1
474 + librte_eventdev.so.3
475 + librte_flow_classify.so.1
483 librte_latencystats.so.1
486 + librte_mempool.so.3
493 + librte_pmd_bnxt.so.2
494 + librte_pmd_bond.so.2
495 + librte_pmd_i40e.so.2
496 + librte_pmd_ixgbe.so.2
498 + librte_pmd_softnic.so.1
499 + librte_pmd_vhost.so.2
505 + librte_security.so.1
514 * Intel(R) platforms with Intel(R) NICs combinations
518 * Intel(R) Atom(TM) CPU C2758 @ 2.40GHz
519 * Intel(R) Xeon(R) CPU D-1540 @ 2.00GHz
520 * Intel(R) Xeon(R) CPU D-1541 @ 2.10GHz
521 * Intel(R) Xeon(R) CPU E5-4667 v3 @ 2.00GHz
522 * Intel(R) Xeon(R) CPU E5-2680 v2 @ 2.80GHz
523 * Intel(R) Xeon(R) CPU E5-2699 v3 @ 2.30GHz
524 * Intel(R) Xeon(R) CPU E5-2695 v4 @ 2.10GHz
525 * Intel(R) Xeon(R) CPU E5-2658 v2 @ 2.40GHz
526 * Intel(R) Xeon(R) CPU E5-2658 v3 @ 2.20GHz
534 * Red Hat Enterprise Linux Server release 7.3
535 * SUSE Enterprise Linux 12
542 * Intel(R) 82599ES 10 Gigabit Ethernet Controller
544 * Firmware version: 0x61bf0001
545 * Device id (pf/vf): 8086:10fb / 8086:10ed
546 * Driver version: 5.2.3 (ixgbe)
548 * Intel(R) Corporation Ethernet Connection X552/X557-AT 10GBASE-T
550 * Firmware version: 0x800003e7
551 * Device id (pf/vf): 8086:15ad / 8086:15a8
552 * Driver version: 4.4.6 (ixgbe)
554 * Intel(R) Ethernet Converged Network Adapter X710-DA4 (4x10G)
556 * Firmware version: 6.01 0x80003205
557 * Device id (pf/vf): 8086:1572 / 8086:154c
558 * Driver version: 2.1.26 (i40e)
560 * Intel(R) Ethernet Converged Network Adapter X710-DA2 (2x10G)
562 * Firmware version: 6.01 0x80003204
563 * Device id (pf/vf): 8086:1572 / 8086:154c
564 * Driver version: 2.1.26 (i40e)
566 * Intel(R) Ethernet Converged Network Adapter XXV710-DA2 (2x25G)
568 * Firmware version: 6.01 0x80003221
569 * Device id (pf/vf): 8086:158b
570 * Driver version: 2.1.26 (i40e)
572 * Intel(R) Ethernet Converged Network Adapter XL710-QDA2 (2X40G)
574 * Firmware version: 6.01 0x8000321c
575 * Device id (pf/vf): 8086:1583 / 8086:154c
576 * Driver version: 2.1.26 (i40e)
578 * Intel(R) Corporation I350 Gigabit Network Connection
580 * Firmware version: 1.63, 0x80000dda
581 * Device id (pf/vf): 8086:1521 / 8086:1520
582 * Driver version: 5.3.0-k (igb)
584 * Intel(R) platforms with Mellanox(R) NICs combinations
588 * Intel(R) Xeon(R) CPU E5-2697A v4 @ 2.60GHz
589 * Intel(R) Xeon(R) CPU E5-2697 v3 @ 2.60GHz
590 * Intel(R) Xeon(R) CPU E5-2680 v2 @ 2.80GHz
591 * Intel(R) Xeon(R) CPU E5-2650 v4 @ 2.20GHz
592 * Intel(R) Xeon(R) CPU E5-2640 @ 2.50GHz
593 * Intel(R) Xeon(R) CPU E5-2620 v4 @ 2.10GHz
597 * Red Hat Enterprise Linux Server release 7.3 (Maipo)
598 * Red Hat Enterprise Linux Server release 7.2 (Maipo)
603 * MLNX_OFED: 4.2-1.0.0.0
607 * Mellanox(R) ConnectX(R)-3 Pro 40G MCX354A-FCC_Ax (2x40G)
609 * Host interface: PCI Express 3.0 x8
610 * Device ID: 15b3:1007
611 * Firmware version: 2.42.5000
613 * Mellanox(R) ConnectX(R)-4 10G MCX4111A-XCAT (1x10G)
615 * Host interface: PCI Express 3.0 x8
616 * Device ID: 15b3:1013
617 * Firmware version: 12.21.1000
619 * Mellanox(R) ConnectX(R)-4 10G MCX4121A-XCAT (2x10G)
621 * Host interface: PCI Express 3.0 x8
622 * Device ID: 15b3:1013
623 * Firmware version: 12.21.1000
625 * Mellanox(R) ConnectX(R)-4 25G MCX4111A-ACAT (1x25G)
627 * Host interface: PCI Express 3.0 x8
628 * Device ID: 15b3:1013
629 * Firmware version: 12.21.1000
631 * Mellanox(R) ConnectX(R)-4 25G MCX4121A-ACAT (2x25G)
633 * Host interface: PCI Express 3.0 x8
634 * Device ID: 15b3:1013
635 * Firmware version: 12.21.1000
637 * Mellanox(R) ConnectX(R)-4 40G MCX4131A-BCAT/MCX413A-BCAT (1x40G)
639 * Host interface: PCI Express 3.0 x8
640 * Device ID: 15b3:1013
641 * Firmware version: 12.21.1000
643 * Mellanox(R) ConnectX(R)-4 40G MCX415A-BCAT (1x40G)
645 * Host interface: PCI Express 3.0 x16
646 * Device ID: 15b3:1013
647 * Firmware version: 12.21.1000
649 * Mellanox(R) ConnectX(R)-4 50G MCX4131A-GCAT/MCX413A-GCAT (1x50G)
651 * Host interface: PCI Express 3.0 x8
652 * Device ID: 15b3:1013
653 * Firmware version: 12.21.1000
655 * Mellanox(R) ConnectX(R)-4 50G MCX414A-BCAT (2x50G)
657 * Host interface: PCI Express 3.0 x8
658 * Device ID: 15b3:1013
659 * Firmware version: 12.21.1000
661 * Mellanox(R) ConnectX(R)-4 50G MCX415A-GCAT/MCX416A-BCAT/MCX416A-GCAT
664 * Host interface: PCI Express 3.0 x16
665 * Device ID: 15b3:1013
666 * Firmware version: 12.21.1000
668 * Mellanox(R) ConnectX(R)-4 50G MCX415A-CCAT (1x100G)
670 * Host interface: PCI Express 3.0 x16
671 * Device ID: 15b3:1013
672 * Firmware version: 12.21.1000
674 * Mellanox(R) ConnectX(R)-4 100G MCX416A-CCAT (2x100G)
676 * Host interface: PCI Express 3.0 x16
677 * Device ID: 15b3:1013
678 * Firmware version: 12.21.1000
680 * Mellanox(R) ConnectX(R)-4 Lx 10G MCX4121A-XCAT (2x10G)
682 * Host interface: PCI Express 3.0 x8
683 * Device ID: 15b3:1015
684 * Firmware version: 14.21.1000
686 * Mellanox(R) ConnectX(R)-4 Lx 25G MCX4121A-ACAT (2x25G)
688 * Host interface: PCI Express 3.0 x8
689 * Device ID: 15b3:1015
690 * Firmware version: 14.21.1000
692 * Mellanox(R) ConnectX(R)-5 100G MCX556A-ECAT (2x100G)
694 * Host interface: PCI Express 3.0 x16
695 * Device ID: 15b3:1017
696 * Firmware version: 16.21.1000
698 * Mellanox(R) ConnectX-5 Ex EN 100G MCX516A-CDAT (2x100G)
700 * Host interface: PCI Express 4.0 x16
701 * Device ID: 15b3:1019
702 * Firmware version: 16.21.1000
704 * ARM platforms with Mellanox(R) NICs combinations
708 * Qualcomm ARM 1.1 2500MHz
714 * MLNX_OFED: 4.2-1.0.0.0
718 * Mellanox(R) ConnectX(R)-4 Lx 25G MCX4121A-ACAT (2x25G)
720 * Host interface: PCI Express 3.0 x8
721 * Device ID: 15b3:1015
722 * Firmware version: 14.21.1000
724 * Mellanox(R) ConnectX(R)-5 100G MCX556A-ECAT (2x100G)
726 * Host interface: PCI Express 3.0 x16
727 * Device ID: 15b3:1017
728 * Firmware version: 16.21.1000