NVIDIA GH200 96GB HBM3

Hopper SXM 2023 4nm Spec confidence: Official
FP8 (dense)
1,979
TFLOPS
FP32
67
TFLOPS
VRAM
96 GB
HBM3
Bandwidth
4.0 TB/s
memory
TDP
1.0 kW
67.0 TFLOPS/kW

Performance

Peak theoretical throughput by precision type

INT8
8-bit integer
1,979 TOPS
3,958 with sparsity
FP8
8-bit floating point
1,979 TFLOPS
3,958 with sparsity
FP16
16-bit floating point
990 TFLOPS
1,979 with sparsity
BF16
Brain Float 16
990 TFLOPS
1,979 with sparsity
TF32
TensorFloat-32
494 TFLOPS
989 with sparsity
FP32
32-bit floating point
67 TFLOPS
FP64
64-bit floating point
34 TFLOPS
Dense Added with sparsity
1.979 TFLOPS/W FP8 at 1000 W
0.990 TFLOPS/W FP16 at 1000 W
0.067 TFLOPS/W FP32 at 1000 W

The GH200 96GB HBM3 in the GPU landscape

Peak FP16 TFLOPS (dense) against TDP, single-GPU parts tracked by Flopper

06001,2001,8002,4003,0000 W250 W500 W750 W1000 W1250 W1500 WInstinct MI355XGH200 96GB HBM3

Higher and further left is better: more half-precision throughput for less power.

Specifications

Architecture

Hopper

Form Factor

SXM

Launch Year

2023

Process Node

4nm

Memory

96 GB HBM3

Bandwidth

4,000 GB/s

TDP

1.0 kW

Max Power

~1.1 kW

Interconnect

900 GB/s NVLink-C2C

Spec Confidence

Official

Full Specifications

Memory
VRAM 96 GB
Memory Type HBM3
Bandwidth 4.0 TB/s
Interconnect & I/O
GPU-to-GPU NVLink-C2C
Interconnect Bandwidth 900 GB/s
Power & Thermal
TDP 1.0 kW
Enterprise Features
Sparsity Yes
General
Form Factor SXM
Architecture Hopper
Launch Year 2023

Datasheet & Resources

Data Provenance

Every figure traced to a source

Primary Source

Publisher
NVIDIA
Published
2024-03-01

Data Quality

Spec confidence
Official
Clock basis
Boost
Precisions tracked
7
Normalization
All values in TFLOPS

Rent this GPU

$6.50 / GPU-hr

Lowest of 1 live listing across 1 provider

See all rental prices →

Price history and every provider on the pricing page. Some links are affiliate links.

Similar GPUs

NVIDIA H100 SXM5 80GB

SXM · 2022
FP32: 67 TFLOPS
Compare vs H100

NVIDIA H200 SXM 141GB

SXM · 2024
FP32: 67 TFLOPS
Compare vs H200

NVIDIA GH200 144GB HBM3e

SXM · 2024
FP32: 67 TFLOPS
Compare vs GH200

NVIDIA H100 NVL 94GB

PCIe · 2023
FP32: 60 TFLOPS
Compare vs H100

NVIDIA H200 NVL 141GB

PCIe · 2024
FP32: 60 TFLOPS
Compare vs H200

Systems Using This GPU

Pre-configured systems featuring the NVIDIA GH200

SystemGPU CountPeak PerformanceTotal Power
HPE Cray Supercomputing EX254n
Cray Supercomputing EX · blade ·
8x GH200 -- -- View System
QCT QuantaGrid S74G-2U
QuantaGrid · rack ·
1x GH200 -- -- View System

Frequently Asked Questions

How many TFLOPS does the NVIDIA GH200 have?

The NVIDIA GH200 delivers 67 TFLOPS for FP32 operations, 990 TFLOPS for FP16, and 1979 TFLOPS for FP8 precision.

What is the power consumption of the NVIDIA GH200?

The NVIDIA GH200 has a TDP (Thermal Design Power) rating of 1000 watts.

How much memory does the NVIDIA GH200 have?

The NVIDIA GH200 is equipped with 96 GB of VRAM with 4000 GB/s memory bandwidth.

What architecture is the NVIDIA GH200 based on?

The NVIDIA GH200 is based on the Hopper architecture, launched in 2023.

Stay Updated on GPU Releases

Get notified when new GPUs are added or specifications are updated.

Loading verification...

No spam, unsubscribe anytime.

Back to GPUs