NVIDIA Mellanox MCX653106A-EFAT SmartNIC Firmware 20.27.6008

Manufacturer:

Description

DOWNLOAD NOW

Firmware which is added at the time of manufacturing, is used to run user programs on the device and can be thought of as the software that allows hardware to run. Embedded firmware is used to control the functions of various hardware devices and systems, much like a computer’s operating system (OS) controls the function of software applications. Firmware may be written into read-only memory (ROM), erasable programmable read-only memory (EPROM) or flash memory.

Important Notes:

- Security Hardening Enhancements: This release contains important reliability improvements and security hardening enhancements. Mellanox recommends upgrading your devices firmware to this release to improve the devices’ firmware security and reliability.
- When upgrading or changing configuration on multi-host adapter cards, for the changes to take effect, PCIe restart must be simultaneously send from both hosts (servers).
- SR-IOV - Virtual Functions (VF) per Port - The maximum Virtual Functions (VF) per port is 127.
- It is recommended to enable the “above 4G decoding” BIOS setting for features that require large amount of PCIe resources. Such features are: SR-IOV with numerous VFs, PCIe Emulated Switch, and Large BAR Requests.

Changes and New Features:

- Enabled KP4RS FEC on Active Fiber cables (OPN: MFS1S00-V0xxE).
- Disabled PLR on Active Fiber cables.
- Added support FDR protocol.
- SHARP SAT is at GA level.
- Enabled updating End-to-End (E2E) credit packets instantly.

Bug Fixes:

- CWDM4 AOM cable is currently not supported.
- Links between ConnectX-6 adapter cards and Spectrum-2 SN3700 switch systems do not go up when using Cisco BiDi module at 100GbE link speed.
- Occasionally, the link is not raised when working with split 2 X 100 mode with Optical cables.
- Fixed an issue that prevented the load of the correct PCIe Tx parameters when the speed was changed after the PCIe link was disabled.
- Fixed an issue that prevented the desched_threshold field from working properly.
- Fixed an issue that caused the "roce_adp_retrans" counter to present the values of the "local_ack_timeout_err" counter.
- If Relaxed Ordering is disabled by running the "setpci" command, it will not be functional even after re-enabling it by running the "setpci" command again.
- Updated the RoCE Tx CNP's BECN value to be 1.
- Aligned the User Memory page to 2, e.g., the user should use 8k aligned (like 0, 8k, 16k etc., instead of 0, 4k, 12k etc.).
- NRZ and PAM4 Limitations in 100GbE adapter cards.
- The default configuration is NRZ speeds in Auto-Neg mode
- PAM4 speeds should be set manually (using the MFT tool – mlxlink)
- PAM4 speeds should be configured as single speeds: 50G_1x or 100G_2x
- 200G_4x optical cables (including 200G_4x splitter cable) will raise link only after manual configuration of PAM4 speeds (50G_1x or 100G_2x)
- NRZ and PAM4 Limitations in 200GbE adapter cards.
- The default configuration is 200G_x4 in Force mode
- NRZ speeds should be set manually. Configuration can include multiple NRZ speeds (using the MFT tool – mlxlink)
- PAM4 speeds should be configured as single speeds: 50G_1x or 100G_2x or 200G_4x
- In Ethernet mode, at 25/50/100GbE speeds, only RS-FEC in Force mode is supported. Other user configurations are overridden.
- Fixed an issue that caused the adapter card to get stuck in "polling" state after resetting/power cycling the card.
- Fixed a PCIe PLL varactor calibration logic to address potential silicon variations.
- Fixed an issue that caused the ZTR counters query to always return 0.
- Fixed an issue that prevented the adapter card from going into the bypass mode when the BMC disabled the hardware arbitration.
- Fixed an issue that resulted in error report messages being discarded due to the “error report" filter not functioning properly.
- Fixed redundant page consumption in the init_hca.
- Degradation in throughput might be experienced when using HDR100 links with cables lengths of 10m-30m.
- The sw_reset option is not supported when ATS is enabled.
- Fixed ATS functionality issues.
- Due to the string DB not being updated after Live-Patch, the tracer cannot function after Live-Patch.
- Fixed a rare issue that caused other active functions to receive a malformed CQE during driver (PF or VF) unload or FLR flows.
- Fixed an issue that prevented PCI link from being established when the firmware was corrupted.
- Fixed an issue that triggered a FW assert and resulted in a wrong deallocation of a resource when Packet Pacing was enabled, and a QP was being destroyed.
- There is a configuration limitation when using a ConnectX-6 VPI card and one port is set in EN mode and the other port in IB mode.
- Fixed an issue that caused the firmware to hang when an FLR occurred at the same time as the teardown. As a result, the teardown flow took a lock, and never released it because it was being aborted by an FLR.
- Fixed a performance issue that occured when CAPI was enabled in the NVConfig.
- When working with Multi-GVMI and SR-IOV, and with a high number of Virtual Functions and sub-functions, the driver start may fail for the VFs/sub-functions.
- Fixed an issue that slowed the firmware flows when executing many destroy XRQ commands on an XRQ that supported DC transport service.
- Upon temperature changes, HDR link can potentially go down due to a temperature change higher than 6 degrees Celsius.
- Fixed a rare issue that resulted in firmware getting stuck with the below message in dmesg during driver restart or driver reboot, and under stress of QP timeouts (packet drops/network congestion): mlx5_core 0000:a1:00.0: Firmware over 120000 MS in pre-initializing state, aborting.
- The bandwidth on MFS1S00-H050E cables is 99G/s and on MFS1S00-H100E cables is 67Gb/s when connecting at HDR speed to an HDR switch.
- On rare occasions, when firmware coalesce Host stuck events occur, a async event might be delayed to be reported, and not be triggered until the next time the PCIe hangs on one of the hosts.
- Renamed the GMP Mellanox Vendor Specific External Capability mask enum from IsDiagnosticCountersSupported to IsDiagnosticDataSupported.
- HDR link up time when using optical and copper cables may take 6 minutes or more (up to 20 minutes).
- A port may hang while Link-Maintenance runs on it and the second port’s link is toggled.
- Fixed an issue that prevented the firmware from resetting the port speed to its default speed HDR due to a TEARDOWN_HCA with open resources event that led to a bad state in firmware, where attributes were not reset to their default between drivers.
- Link flapping and packet loss during High/Low temperature changes.
- When using a copper splitter cable up to 2m length in HDR100 mode, traffic may drop.
- On rare occasions, HDR link may not raise properly due to temperature changes.
- In a rare scenario when the driver is executing the "2err" command and the QP is in SQ drain state, the firmware might post event of broken WQ instead of sending error CQEs on all the WQEs.
- Fixed an issue that prevented users with non-port owner privilege from using the read DCBX access registry key REGID_DCBX_APP/REGID_DCBX_PARAM.
- When connected to a Switch-IB or Switch-IB 2 based switch system using HDR cables, link may become stuck during polling.
- Fixed an issue that resulted in out-of-order packet transmission due to an internal bypass race.
- Fixed a rare issue that caused the QP to falsely transition into the error state as a result of handling duplicate read/atomic request followed by memory key invalidation.

About Network Adapter Drivers:

When connected, the operating system usually installs a generic driver that helps the computer to recognize the newly attached device.

However, proper software must be applied if you want to make use of all features that the network adapter has available. This task also allows computers to properly recognize all device characteristics such as manufacturer, chipset, technology, and others.

Updating the adapter's drivers and utilities version might improve overall performance and stability, increase transfer speeds, fix different compatibility problems and several network-related errors, as well as bring various other changes.

To install this release, simply get the package, extract it if necessary, run the setup, and follow the instructions displayed on-screen. When done, don't forget to perform a system restart and reconnect the network adapter to make sure that all changes take effect properly.

Without further ado, if you intend to apply this version, click the download button and install the package. Moreover, check with our website as often as possible so that you don't miss a single new release.

Mellanox ConnectX-6 VPI NIC Firmware Mellanox ConnectX-6 VPI SmartNIC Firmware Mellanox MCX653106A-EFAT Firmware MCX653106A-EFAT Firmware Mellanox NVIDIA

  CATEGORY:
Network Card
  COMPATIBLE WITH:
OS Independent
  file size:
2.6 MB
  filename:
fw-ConnectX6-rel-20_27_6008-MCX653106A-EFA_Ax-UEFI-14.20.25-FlexBoot-3.5.903.bin.zip