This is the second GA102 bin sold as an RTX 3080. It is a different card from the 10 GB launch model: 70 SMs against 68, a 384-bit bus against 320-bit, 12 GB against 10 GB and 350 W against 320 W. NVIDIA never issued a press release for it, so its launch year comes from the Game Ready driver 511.23 release notes of January 2022, which add support for "the NVIDIA GeForce RTX 3080 (12GB) GPU". Memory bandwidth is not published for this SKU and is deliberately left blank rather than derived from an assumed data rate.

NVIDIA GeForce RTX 3080 12GB

Ampere PCIe 2022 8nm Spec confidence: Vendor claimed
FP32
31
TFLOPS
VRAM
12 GB
GDDR6X
TDP
350 W
87.5 TFLOPS/kW
Download datasheet

Overview

The 12 GB GeForce RTX 3080 is the larger of the two GA102 bins NVIDIA sold under the RTX 3080 name, and the more useful of the pair for local AI work: 12 GB of GDDR6X on a full 384-bit bus rather than 10 GB on 320-bit, and 70 streaming multiprocessors rather than 68. The extra two gigabytes are what decide whether a 7B model fits in half precision without offloading, which is why second-hand 12 GB cards still command a premium over the 10 GB version. Tensor throughput follows the GeForce GA10x pattern: TF32 runs at the plain FP32 shader rate rather than the doubled rate professional Ampere cards get, while INT8 and INT4 are left at full speed, so this is a strong quantised-inference card and an ordinary TF32 training one. There is no ECC and no NVLink on this bin.

Performance

Peak theoretical throughput by precision type

PrecisionDense2:4 Sparse
FP64
64-bit floating point
0.5TFLOPS Structured sparsity is a tensor-core feature; this vector precision has no sparse form
FP32
32-bit floating point
31TFLOPS Structured sparsity is a tensor-core feature; this vector precision has no sparse form
TF32
TensorFloat-32
31TFLOPS 61TFLOPS
BF16
Brain Float 16
61TFLOPS 123TFLOPS
FP16
16-bit floating point
61TFLOPS 123TFLOPS
FP8
No verified data available The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision
FP6
No verified data available The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision
FP4
No verified data available The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision
INT8
8-bit integer
245TOPS 490TOPS
INT4
4-bit integer
490TOPS 981TOPS

The GeForce RTX 3080 12GB 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 MI355XGeForce RTX 3080 12GB

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

Specifications

Architecture

Ampere

Form Factor

PCIe

Launch Year

2022

Process Node

8nm

Memory

12 GB GDDR6X

Bandwidth

No verified data available

TDP

350 W

Max Power

350 W

Transistors

28.3 billion

CUDA Cores

8,960

Spec Confidence

Vendor claimed

Full Specifications

Compute Engine
CUDA Cores 8,960
Tensor Cores 280 (3rd Gen)
Streaming Multiprocessors 70
Base Clock 1.26 GHz
Boost Clock 1.71 GHz
Chip Design
Transistors 28.3 billion
Die Size 628 mm²
Process Node 8nm
Memory
VRAM 12 GB
Memory Type GDDR6X
Bandwidth No verified data available
Interface Width 384-bit
Interconnect & I/O
PCIe 4.0 x16
Power & Thermal
TDP 350 W
Max Board Power 350 W
Cooling Active
Enterprise Features
ECC Memory No
Sparsity Yes
Compute APIs CUDA, DirectX 12, Vulkan, OpenGL
General
Form Factor PCIe
Architecture Ampere
Process Node 8nm
Launch Year 2022

Datasheet & Resources

Data Provenance

Every figure traced to a source

Primary Source

Publisher
NVIDIA
Published
2021-06-03

Data Quality

Spec confidence
Vendor claimed
Clock basis
Boost
Core precisions with figures
6 of 9
Normalization
All values in TFLOPS

Compare cloud providers offering on-demand GPU instances for AI training, inference, and HPC workloads.

Browse GPU Cloud Providers

Similar GPUs

NVIDIA RTX A6000 48GB

PCIe · 2020
FP32: 39 TFLOPS
Compare vs RTX A6000

NVIDIA A40 48GB

PCIe · 2021
FP32: 37 TFLOPS
Compare vs A40

NVIDIA A10G 24GB

PCIe · 2022
FP32: 35 TFLOPS
Compare vs A10G

Frequently Asked Questions

How many TFLOPS does the NVIDIA GeForce RTX 3080 have?

The NVIDIA GeForce RTX 3080 delivers 31 TFLOPS FP32 and 61 TFLOPS FP16 at peak. Flopper does not currently have a verified FP8 throughput figure for it.

What is the power consumption of the NVIDIA GeForce RTX 3080?

The NVIDIA GeForce RTX 3080 has a TDP (Thermal Design Power) rating of 350 watts.

How much memory does the NVIDIA GeForce RTX 3080 have?

The NVIDIA GeForce RTX 3080 is equipped with 12 GB of VRAM. Flopper does not currently have a verified memory bandwidth figure for it.

What architecture is the NVIDIA GeForce RTX 3080 based on?

The NVIDIA GeForce RTX 3080 is based on the Ampere architecture, launched in 2022.

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