Feature Description
While this solves network scalability issues, it hides the TCP packet from
the hardware offloading engines, placing higher loads on the host CPU.
NVIDIA DPU effectively addresses this by providing advanced NVGRE
and VXLAN hardware offloading engines that encapsulate and de-
capsulate the overlay protocol.
RDMA and
RDMA over
Converged
InfiniBand/V
PI (RoCE)
NVIDIA DPU, utilizing IBTA RDMA (Remote Data Memory Access) and
RoCE (RDMA over Converged InfiniBand/VPI) technology, delivers low-
latency and high-performance over InfiniBand/VPI networks. Leveraging
data center bridging (DCB) capabilities as well as advanced congestion
control hardware mechanisms, RoCE provides efficient low-latency
RDMA services over Layer 2 and Layer 3 networks.
NVIDIA
PeerDirect™
PeerDirect communication provides high-efficiency RDMA access by
eliminating unnecessary internal data copies between components on
the PCIe bus (for example, from GPU to CPU), significantly reducing
application run time. NVIDIA DPU's advanced acceleration technology
enables higher cluster efficiency and scalability to tens of thousands of
nodes.
Quality of
Service (QoS)
Support for port-based Quality of Service enabling various application
requirements for latency and SLA.
Storage
Acceleration
A consolidated compute and storage network achieves significant
cost-performance advantages over multi-fabric networks. Standard
block and file access protocols can leverage RDMA for high-
performance storage access: NVMe over Fabric offloads for the
target machine
BlueField-2 DPU may operate as a co-processor offloading specific
storage tasks from the host, isolating part of the storage media
from the host, or enabling abstraction of software-defined storage
logic using the NVIDIA BlueField-2 Arm cores. On the storage
initiator side, NVIDIA BlueField-2 DPU can prove an efficient
solution for hyper-converged systems to enable the host CPU to
focus on computing while all the storage interface is handled
through the Arm cores.
NVMe-oF Non-volatile Memory Express (NVMe) over Fabrics is a protocol for
communicating block storage IO requests over RDMA to transfer data
between a host computer and a target solid-state storage device or
system over a network. NVIDIA BlueField-2 DPU may operate as a co-