Business Goals
ITZKXY enterprise networking and AI infrastructure support
AIandhigh-performance computing
Product Information: data centersnetwork solutions
Product Information: AI,,,
Product Information: 2026 4
Product Information
This product category is a NVIDIA switch designed for AI clusters, HPC. It supports InfiniBand connectivity, NDR, low latency.
NVIDIA Quantum-2, 1U 64 non-blocking400Gb/s InfiniBand, Switchescan deliver51.2Tb/s aggregate bidirectional throughputThis product is a networking product designed for AI clusters.
and
Product Information
AI:, and W H, meet
GPU: latency GPU, 50%
H:, and
Product Information: data centersnetworking equipment
Product Information
NDR is a networking product designed for AI clusters, enterprise networks. It supports InfiniBand connectivity, NDR. Key specifications include 400G.
products
1. leading
| Product Information | Product Information |
|---|---|
| Product Information | 64 NDR 400Gb/snon-blocking ports (32 OSFP) |
| throughput | 51.2 Tb/s |
| Product Information | 665/(BPPS) |
| latency | latency |
| dimensions | 1U |
2. in-network computing
adaptive routing:,, W
RDMAsupports: Remote Direct Memory Access, CPUandoperating system, reduceslatencyandCPU
3.
QM97 provides management, deployment:
(QM9700)：
externally managed (QM9790)：
NVIDIA Unified Fabric Manager (UFM)
data centers
provides and
4. H
Product Information: 6+1
Product Information: and
backward compatible: InfiniBand products
Model
| Product Information | QM9700 | QM9790 | QM9701 |
|---|---|---|---|
| Management | Product Information | externally managed (UFM) | (DGX) |
| Product Information | 64 NDR 400G | 64 NDR 400G | 64 NDR 400G |
| power supplies | AC 1+1 | AC 1+1 | DC 48V |
| Product Information | P2C/C2P | P2C/C2P | C2P |
| Product Information | Product Information | Product Information | NVIDIA DGX POD |
| SKU | MQM9700-NS2F/R | MQM9790-NS2F/R | MQM9701-NS2R |
supports topologies
NVIDIA Quantum-2Switchessupports, deploymentprovidesflexibility:
Fat Tree ()Product Information
SlimFly: and,
DragonFly+: deployment
Torus: and HPC
port-splitting technology, 400G 2 200G, supports 128 200Gb/s, provides H anddeploymentflexibility。
Product Information
: AI
Product Information: GPU, W H
Product Information: usesQM9700builtSpine-Leaf
supports GPU
adaptive routingensures
Product Information
: cloud data centers
Product InformationProduct Information
Product Information: QM9790 + NVIDIA UFM
UFMprovides and
and
supportsDragonFly+ H, reducesTCO
: DGX POD/AI
Product Information: NVIDIA DGX,
Product Information: QM9701 DGX Switches
DGX D
DC data centers
andDGX H100/B200
Product Information
customers
GPU
, GPU 90%
, AI
reducesTCO
1UH,
H power supplies (80 Gold+)
UFM
Product Information
supportsNDR 400G, HDR/EDR
deployment
SHARPin-network computing AI
WeChat
Profile