NVIDIA HGX B100 8-GPU

baseboard
8× B100
NVSwitch + NVLink
2024
FP64
0.2
PFLOPS
FP32
0.5
PFLOPS
Power
5.6
kW Total
Memory
1536
GB Total

We will point you at suppliers who have it. Free, and no signup.

FP64
0.24
PFLOPS
FP32
0.48
PFLOPS
TF32
7.00
PFLOPS
FP16
14.00
PFLOPS

System Details

GPU Configuration

GPU Count: 8 GPUs
Architecture: Blackwell
Interconnect: NVSwitch + NVLink

System Specifications

Form Factor: baseboard
Total Power: 5.6 kW
Total Memory: 1.5 TB
Memory Bandwidth: 64 TB/s

Precision Performance Breakdown

PrecisionSystem PerformancePer GPUEfficiency
0.240 PFLOPS 30.0 TFLOPS
0.480 PFLOPS 60.0 TFLOPS
7.000 PFLOPS 875.0 TFLOPS
14.000 PFLOPS 1750.0 TFLOPS
14.000 PFLOPS 1750.0 TFLOPS
28.000 PFLOPS 3500.0 TFLOPS
56.000 PFLOPS 7000.0 TFLOPS
28.000 PFLOPS 3500.0 TFLOPS

Where a vendor states a basis, the figure shown is the dense one, and any figure it publishes is listed separately. Vendors commonly headline the sparse number instead: for NVIDIA's 2:4 structured sparsity that is exactly twice the dense figure, though other vendors' sparse modes do not all follow that ratio.

Powered by NVIDIA B100

This system utilizes 8 × NVIDIA B100 SXM 192GB GPUs, each delivering exceptional performance for AI and HPC workloads.

Per GPU TDP

700W

Per GPU Memory

192 GB

Process Node

4nm

Architecture

Blackwell

Documentation & Resources

Flopper Spec Sheet

NVIDIA HGX B100 8-GPU specifications, generated from our database. Printable.

Vendor product page

NVIDIA HGX B100 8-GPU documentation

View ↗

workstation-datasheet-dgx-spark-gtc25-spring-nvidia-us-3716899-web.pdf

NVIDIA • 2025-10-16

View Document ↗

NVIDIA DGX Documentation

System guides, deployment resources

Browse ↗

Typical Use Cases

Large Language Model Training
Distributed Deep Learning
Multi-GPU Inference
HPC Simulations
Scientific Computing

The NVIDIA HGX B100 8-GPU runs 8× B100 GPUs over NVSwitch + NVLink, delivering 28.0 PFLOPS FP8 dense.