NVIDIA GeForce RTX 3070 8GB

Ampere PCIe 2020 8nm

Overview

20.3

FP32 TFLOPS

8 GB

Memory GDDR6

448 GB/s

Memory Bandwidth

220 W

TDP

Performance Metrics

Precision	Peak (dense)	2:4 Sparse	Bits	Type
FP64	0.3	N/A	64	Standard
FP32	20.3	N/A	32	Standard
TF32	20.3	40.6	32	Standard
BF16	40.6	81.3	16	Standard
FP16	40.6	81.3	16	Standard
FP8	—	not published	—	Standard
FP6	—	not published	—	Standard
FP4	—	not published	—	Standard
INT8	162.6	325.2	8	Standard
INT4	325.2	650.4	4	Standard

Memory Specifications

Capacity: 8 GB
Type: GDDR6
Bandwidth: 448 GB/s
Interface Width: 256-bit

Power & Efficiency

TDP: 220 W
Max power (Flopper estimate): ~253 W est.
FP32 dense TFLOPS per W of TDP: 0.09 TFLOPS/W
FP16 dense TFLOPS per W of TDP: 0.18 TFLOPS/W

Flopper estimate: 220 W TDP x 1.15. The vendor publishes no maximum board power for this part. Per W of TDP figures divide dense peak by accelerator TDP. Board power only: excludes host CPUs, networking, cooling and facility overhead.

Hardware Details

Vendor: NVIDIA

Architecture: Ampere

Process Node: 8nm

Form Factor: PCIe

Launch Year: 2020

Variant: 8GB

Compute Engine

CUDA Cores: 5,888

Tensor Cores: 184 (3rd Gen)

SMs: 46

Base Clock: 1.50 GHz

Boost Clock: 1.73 GHz

Chip Design

Transistors: 17.4 billion

Die Size: 393 mm²

Enterprise Features

Sparsity: Yes