NVIDIA roadmap for 24.07

NVIDIA roadmap for 24.07

[Maayan Kashani]

[-- Attachment #1: Type: text/plain, Size: 3663 bytes --]

Please find below NVIDIA roadmap for 24.07 release:



A. rte_flow new APIs

=================

1.      Matching VXLAN-GBP header​ in template API

Support matching VXLAN-GBP extension header format which contains the G bit and group ID inside the VXLAN header.​

2.      Add support for modifying the VXLAN Header’s reserved field (last 8-bit field)



B. Net/mlx5 PMD updates

=====================

1.       Add NVGRE​ header matching support in template API

Support matching NVGRE headers, a comparison to the non-template API​

2.       Support FLAG action​ in template API

Supported via reusing the MARK (flow_tag) action​.

3.       User input validation for template API

Validate the user’s input in both PMD and HWS layer.​

Compiled with debug flag​, no performance impact, Improve debuggability with no performance impact.

Used through a compilation-based debug flag.

4.       Support Geneve-OPT matching with default flex profile

Support matching on Geneve option using default parser mechanism.

With default profile it can also meet the needs with one DW matching.​

Improve user’s experience w/o changing FW.​

5.       Optimize FDB domain tables​

Reduce the duplication of Tx / Rx tables in the FDB domain to optimize the performance

and reduce memory usage to allow room for more offloads (increase total scale).

This optimization can be used starting from ConnectX-6-Dx and above.

6.       E-Switch manager vport ID support in metadata​

Provide a unify solution in the match of E-Switch manager vport ID to match on the traffic from E-Switch manager.

In multi-port E-Switch case, the VFs belong to different PFs port will have same misc.source_vport value,

but the metadata value will now be different, this will enable to distinguish between the VF’s related to a different ports.

7.       Hairpin statistics query​

Support per queue hairpin related counters, Rx miss, Tx error and queue drop, etc.​

This will provide more visibility for debug, telemetry, and monitoring purposes to the users​.

8.   non-template API added support for HWS technology (HWS mode) -  (Alpha)

Strating from ConnectX-8 and BlueField-4 DPU,  SWS mode with non-template API will not be supported.

This feature will allow the non-template API to be supported with HWS mode for interim application translation

and with functional and performance limitations (NVIDIA suggest using the template API to avoid possible limitations).

9.      Crypto key allocation optimization​

Optimize mkey allocation to speed up the initialization time in the cases when many queues and descriptors are required.

10.  Add debug capabilities to PMD

There is a need to retrieve debug information from the setup with limited access in real time due to the production environment.

This feature dumps the RQ/SQ/CQ Contexts from DPDK PMD to debug Tx/Rx issues.



C. Test apps updates  

================= 

1.      Support the changes in rte_flow listed above in testpmd. 

2.      Support hairpin with 4 physical ports in testpmd.

For Connectx -7 Nic's that can support up to 4 physical ports, adjust the hairpin logic to allow use of all ports.

Current testpmd only allows consecutive pair for hairpin. This feature is intended to allow use of any pair.



D. Others

================= 

     *   Documentation of action optional ordering for template API

list optional actions order and provide to the user, also suggest optimizations for the application offloading pipelines build​.


[-- Attachment #2: Type: text/html, Size: 43886 bytes --]