Description
NVIDIA GB200 NVL72 System
The NVIDIA GB200 NVL72 is a perfect solution for large-scale AI applications as it can handle even trillion-parameter datasets with ease. Based on the new Blackwell architecture paired with Arm Grace CPU, the final superchip is capable of groundbreaking performance while keeping the TCO and power consumption up to 25x lower compared to previous generation systems.
Substantial performance gain with GB200 NVL72
LLM Inference
LLM Training
Energy Efficiency
Data Processing
*NVIDIA: LLM inference and energy efficiency: TTL = 50 milliseconds (ms) real time, FTL = 5s, 32,768 input/1,024 output, NVIDIA HGX™ H100 scaled over InfiniBand (IB) vs. GB200 NVL72, training 1.8T MOE 4096x HGX H100 scaled over IB vs. 456x GB200 NVL72 scaled over IB. Cluster size: 32,768
A database join and aggregation workload with Snappy / Deflate compression derived from TPC-H Q4 query. Custom query implementations for x86, H100 single GPU and single GPU from GB200 NLV72 vs. Intel Xeon 8480+
Projected performance subject to change.
NVIDIA GB200 NVL72 Configurations
GB200 NVL72 | GB200 Superchip | |
---|---|---|
Configuration | 36x Grace CPU, 72x B200 GPU | 1x Grace CPU, 2x B200 GPU |
FP4 Tensor Core* | 1,440 PFLOPS | 40 PFLOPS |
FP8 / FP6 Tensor Core* | 720 PFLOPS | 20 PFLOPS |
INT8 Tensor Core* | 720 POPS | 20 POPS |
FP16 / BF16 Tensor Core* | 360 PFLOPS | 10 PFLOPS |
TF32 Tensor Core* | 180 PFLOPS | 5 PFLOPS |
FP64 Tensor Core | 3,240 TFLOPS | 90 TFLOPS |
GPU Memory | Up to 13.5 TB HBM3e, 576 TBps | Up to 384 GB HBM3e, 16 TBps |
NVLink Bandwidth | 130 TBps | 3.6 TBps |
CPU Cores | 2,952 Arm Neoverse V2 Cores | 72 Arm Neoverse V2 Cores |
CPU Memory | Up to 17 TB LPDDR5X, Up to 18.4 TBps | Up to 480 GB LPDDR5X, Up to 18.4 TB/s |
Product info | Datasheet |
High speed infiniband
GB200 NVL72 supports advanced networking options at speeds up to 800 gigabits per second (Gb/s). For the highest AI performance, GB200 supports the latest NVIDIA Quantum-X800 InfiniBand and Spectrum™-X800 Ethernet platforms. GB200 NVL72 also includes NVIDIA BlueField-3 data processing units (DPUs) to enable cloud networking, composable storage, zero-trust security and GPU compute elasticity in hyperscale AI clouds.
Reviews
There are no reviews yet.