NVIDIA later shipped a GDDR6 version of this card alongside the original GDDR6X one, at the same 504 GB/s. The compare page lists the memory as "12 GB GDDR6 / 12 GB GDDR6X" for that reason. This row carries the GDDR6X specification the architecture whitepaper documents.

NVIDIA GeForce RTX 4070 12GB

Ada Lovelace PCIe 2023 4nm Spec confidence: Official
FP8 (dense)
117
TFLOPS
FP32
29
TFLOPS
VRAM
12 GB
GDDR6X
Bandwidth
504 GB/s
memory
TDP
200 W
145.5 TFLOPS/kW

Overview

The GeForce RTX 4070 is the 46 SM AD104 part that anchored the middle of the RTX 40 range from April 2023: 5,888 CUDA cores, 184 fourth-generation Tensor cores, and 12 GB of GDDR6X at 21 Gbps on a 192-bit bus for 504 GB/s, inside a 200 W board that runs from two 8-pin connectors. NVIDIA prints its entire precision ladder in the RTX Blackwell architecture whitepaper as a comparison column, so the 116.6 TFLOPS of dense FP8 and 233.2 TOPS of dense INT8 below are vendor figures rather than derived ones. Its 36 MB of L2 cache is six times what an equivalent Ampere part carried, which is why it performs better than its bandwidth alone suggests. For AI use it is a capable, quiet, widely available card whose 12 GB is the thing to plan around.

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
29TFLOPS Structured sparsity is a tensor-core feature; this vector precision has no sparse form
TF32
TensorFloat-32
29TFLOPS 58TFLOPS
BF16
Brain Float 16
58TFLOPS 117TFLOPS
FP16
16-bit floating point
58TFLOPS 117TFLOPS
FP8
8-bit floating point
117TFLOPS 233TFLOPS
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
233TOPS 466TOPS

The GeForce RTX 4070 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 4070 12GB

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

Specifications

Architecture

Ada Lovelace

Form Factor

PCIe

Launch Year

2023

Process Node

4nm

Memory

12 GB GDDR6X

Bandwidth

504 GB/s

TDP

200 W

Max Power

200 W

Transistors

35.8 billion

CUDA Cores

5,888

Spec Confidence

Official

Full Specifications

Compute Engine
CUDA Cores 5,888
Tensor Cores 184 (4th Gen)
Streaming Multiprocessors 46
Base Clock 1.92 GHz
Boost Clock 2.48 GHz
Chip Design
Transistors 35.8 billion
Die Size 295 mm²
Process Node 4nm
Memory
VRAM 12 GB
Memory Type GDDR6X
Bandwidth 504 GB/s
Interface Width 192-bit
Memory Clock 21 GT/s
Interconnect & I/O
PCIe 4.0
Cache
L2 Cache 36 MB
Power & Thermal
TDP 200 W
Max Board Power 200 W
Cooling Active
Enterprise Features
Sparsity Yes
Physical & Media
Card Length 244 mm
Width 2-slot
General
Form Factor PCIe
Architecture Ada Lovelace
Process Node 4nm
Launch Year 2023

Datasheet & Resources

Data Provenance

Every figure traced to a source

Primary Source

Publisher
NVIDIA
Published
2025-05-12

Data Quality

Spec confidence
Official
Clock basis
Boost
Core precisions with figures
7 of 9
Normalization
All values in TFLOPS

Rent this GPU

$0.07 / 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 L40S 48GB

PCIe · 2023
FP32: 92 TFLOPS
Compare vs L40S

NVIDIA L40 48GB

PCIe · 2023
FP32: 91 TFLOPS
Compare vs L40

Frequently Asked Questions

How many TFLOPS does the NVIDIA GeForce RTX 4070 have?

The NVIDIA GeForce RTX 4070 delivers 29 TFLOPS FP32, 58 TFLOPS FP16 and 117 TFLOPS FP8 at peak.

What is the power consumption of the NVIDIA GeForce RTX 4070?

The NVIDIA GeForce RTX 4070 has a TDP (Thermal Design Power) rating of 200 watts.

How much memory does the NVIDIA GeForce RTX 4070 have?

The NVIDIA GeForce RTX 4070 is equipped with 12 GB of VRAM with 504 GB/s of memory bandwidth.

What architecture is the NVIDIA GeForce RTX 4070 based on?

The NVIDIA GeForce RTX 4070 is based on the Ada Lovelace 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