Overview
The NVIDIA® Messaging Accelerator (VMA) library accelerates latency-sensitive and throughput-demanding TCP and UDP socket-based applications by offloading traffic from the user-space directly to the network interface card (NIC) or Host Channel Adapter (HCA), without going through the kernel and the standard IP stack (kernel-bypass).
VMA leverages the following benefits:
-
Implements the legacy POSIX socket interface
-
Increases:
-
Throughput
-
Packets per Second (PPS)
-
Requests per Second (RPS)
-
-
Reduces:
-
Network latency
-
The magnitude of network latency spikes
-
Context switches and interrupts
-
Network congestion
-
Data copying and moving in unicast and multicast applications
-
-
Improves CPU utilization
-
Compatible with Ethernet
VMA can work on top of DOCA-Host driver stack.
System Requirements
The following table presents the currently certified combinations of stacks and platforms, and supported CPU architectures for the current VMA version.
|
Specification |
Value |
|
Network Adapter Cards |
NVIDIA ConnectX-7 (4x25G) |
|
NVIDIA ConnectX-7 (200G) Crypto-disabled |
|
|
NVIDIA ConnectX-5 / NVIDIA ConnectX-5 Ex |
|
|
Firmware |
ConnectX-5 16.35.3502 |
|
ConnectX-7 28.47.1026 |
|
|
Driver Stack |
NOTE: Starting this version, the host driver is part of the NVIDIA DOCA package. For further information, please see NVIDIA MLNX_OFED to DOCA-OFED Transition Guide. |
|
Supported Operating Systems and Kernels
|
|
|
CPU Architecture |
|
|
Minimum memory requirements |
1 GB of free memory for installation 800 MB per process running with VMA |
|
Minimum disk space requirements |
1 GB |
Intended Audience
-
Market data professionals
-
Messaging specialists
-
Software engineers and architects
-
Systems administrators tasked with installing/uninstalling/maintaining VMA
-
ISV partners who want to test/integrate their traffic-consuming/producing applications with VMA
Document Revision History
For the list of changes made to this document, refer to User Manual Revision History.
Last updated: