Download Download NVIDIA Mellanox MCX515A-CCAT SmartNIC Firmware 16.29.1016 for OS Independent Free

The firmware added during manufacture is used to run user programs on the device and can be considered as software that allows the hardware to run. Embedded firmware is used to control the functions of various hardware devices and systems, just as the operating system (OS) of a computer controls the functions of software applications. The firmware can be written to read-only memory (ROM), erasable programmable read-only memory (EPROM), or flash memory.

Important notes:

-Security reinforcement enhancement: This version contains important reliability improvements and security reinforcement enhancements. Mellanox recommends upgrading your device firmware to this version to improve the device’s firmware security and reliability.
-When upgrading or changing the configuration on a multi-host adapter card, for the changes to take effect, PCIe restarts must be sent from both hosts (servers) at the same time.
-SR-IOV-Virtual function (VF) per port-The maximum virtual function (VF) per port is 127.
-For functions that require a large amount of PCIe resources, it is recommended to enable the BIOS setting of “Decoding above 4G”. Such functions include: SR-IOV with a large number of VFs, PCIe emulation switches, and large BAR requests.

Changes and new features:

-Added an option that allows applications to build their own QoS tree on the NIC hierarchy by connecting QP to responder/requester queue groups.
-Added the “InfiniBand” property set as a network device function Redfish object.
-Added hardware support for flow measurement to take advantage of advanced steering operations (ASO). Compared with FW Flow Metering, HW Flow Meter has higher scale, higher accuracy and better performance.
-Added trust level for VF. Once the VF is trusted, it will acquire a set of trusted capabilities.
-Added 2 new Mini CQE formats: Responder Mini CQE with Flow Tag Layout and Responder Mini CQE with l3_l4_info Layout
-UCX can now enable AR by exposing the out-of-order bitmask of each SL using the “ooo_per_sl” field in the HCA_VPORT context. You can also query by running the QUERY_HCA_VPORT_CONTEXT command.
-Added support for switching to DP hash stream group.
-Added a new counter for each vPort to count the number of packets that reached the Ethernet RQ but could not fit into the WQE due to the large size. In addition, we have added an option to control whether such packets will cause “CQE with Error” or “CQE_MOCK”.
-PCIe Rx modified to prevent the adapter card from disappearing from the system.
-ignore_flow_level is now enabled by TRUST LEVEL access to the registry.
-Added support for cq_overrun counter. The counter indicates the number of times the CQ entered an error state due to an overflow when the device tried to send the CQE to the complete CQ buffer.
—— [Beta] Enable the function that allows the virtual function to send pause frame packets.
-Enable 10/25GbE and 3rd party auto-sensing when using 10/25GbE optical cable.
-Hardware steering dump output for debugging and troubleshooting.

Bug fix:

-Fixed a rare race condition that caused incorrect writing of firmware images under certain power outages. This causes the firmware to be recognized as corrupt and prevents the system from recognizing the adapter card due to the lack of a valid Flash image.
-Change the default value of DCQCN’s NP parameter min_time_between_cnps on all devices to 4 to support greater cluster scalability.
-Fixed the problem that VXLAN packets with svlan/cvlan tags could not be matched.
-Fixed the problem that caused the eth_wqe_too_small counter to calculate the ODP page error.
-When NVCONFIG enables MKEY_BY_NAME and a large number of VFs are configured, VM restart (VF/PF FLR) will take longer than when MKEY_BY_NAME is disabled.
-Fixed the problem of poor performance after enabling the RoCE accelerator function.
-In rare cases, fatal errors related to PCI transport layer errors may be reported during FLR.
-Fixed the calculation of the Chassis Manager for multi-host and Socket-Direct adapter cards to allow the Chassis Manager BMC to run NC-SI commands. Now, no matter how many BMCs there are, the Chassis Manager is regarded as the BMC with index 0.
-Fixed an issue that caused the device to enter a dead IRISC because one of the firmware semaphores could not be released when the speed change or port status change was triggered.

About the network adapter driver:

After connecting, the operating system usually installs a universal driver to help the computer recognize the newly connected device.

However, if you want to take advantage of all the available functions of the network adapter, you must apply the appropriate software. This task also allows the computer to correctly identify all device characteristics, such as manufacturer, chipset, technology, etc.

Updating the adapter’s driver and utility version may improve overall performance and stability, increase transmission speed, fix different compatibility issues and some network-related errors, and bring various other changes.

To install this version, simply obtain the package, unzip it if necessary, run the installer, and follow the instructions displayed on the screen. When you are done, don’t forget to perform a system restart and reconnect the network adapter to ensure that all changes take effect correctly.

Without further ado, if you plan to apply this version, please click the download button and install the package. In addition, please visit our website as often as possible so as not to miss any new versions.