Form Factor
8× NVIDIA Rubin SXM GPUs

Inference Performance (NVFP4)
400 PFLOPS

Training Performance (NVFP4)
280 PFLOPS

FP8 / FP6 Training
140 PFLOPS

INT8 Tensor Core Performance
2 PFLOPS

FP16 / BF16 Tensor Core Performance
32 PFLOPS

TF32 Tensor Core Performance
16 PFLOPS

FP32 Performance
1040 TFLOPS

FP64 / FP64 Tensor Core Performance
264 TFLOPS

Matrix Compute (SGEMM / DGEMM)
3200 TFLOPS | 1600 TFLOPS

Total GPU Memory
2.3 TB

Interconnect Technology
Sixth-Generation NVIDIA NVLink

NVLink Switch
NVLink 6 Switch

GPU-to-GPU Bandwidth
3.6 TB/s

Total NVLink Switch Bandwidth
28.8 TB/s

Networking Bandwidth
1.6 TB/s

Supported GPU Configurations
NVIDIA Rubin, NVIDIA Blackwell, NVIDIA Blackwell Ultra SXM

Primary Workloads
AI Training, AI Inference, Generative AI, Scientific Simulation, HPC Applications

Deployment Environment
AI Factories, Hyperscale Data Centers, Supercomputing Clusters