GeForce RTX 4070
53B
Memory12 GB
Memory TypeGDDR6X
Boost Clock2475 MHz
TGP200 W
ArchitectureAda Lovelace
Launch Price$599
Benchmarks
Gaming54
Render55
TFLOPS (FP32)29.1
Geekbench OpenCL167,522
3DMarkTime Spy17,799
Steel Nomad Lite Score18,089
GB6 Compute Score167,522
Cinebench 2024 GPU18,051
G3D Mark Score26,856
Blender GPU5387.27
GB6 ML Single Precision27,166
GB6 ML Half Precision40,963
GB6 ML Quantized20,332
FLOPS (FP32)29.1 TFLOPS
Memory Bandwidth504 GB/s
Full specifications
3D Mark
Time Spy
17799
Solar Bay
82861
Port Royal
11156
Fire Strike
44067
Wild Life Extreme
35472
Night Raid
153385
GeekBench 6 OpenCL
Background Blur
233 img/sec
Face Detection
141.7 img/sec
Horizon Detection
6.62 Gpixels/sec
Edge Detection
9.19 Gpixels/sec
Gaussian Blur
9.2 Gpixels/sec
Feature Matching
1.78 Gpixels/sec
Stereo Matching
845.4 Gpixels/sec
Particle Physics
25043.9 FPS
API
OpenCL
Passmark Graphics
G2D Mark
1165
DirectX 11
243 FPS
DirectX 12
103 FPS
GPU Compute
14561 Ops/s
GeekBench 6 ML
Image Classification (SP)
10246
Image Segmentation (HP)
26153
Image Super Resolution (Q)
32054
Face Detection (HP)
51337
Pose Estimation (Q)
124617
Text Classification (SP)
3703
Machine Translation (HP)
5484
Object Detection (SP)
13134
Depth Estimation (Q)
50548
Style Transfer (SP)
263477
Framework
ONNX
Backend
DirectML
General
Vendor
Nvidia
Build
Discrete
Released
April 13, 2023
Launch price (MSRP)
$599
Case
Desktop
Purpose
Gaming
Segment
Mid-range
Architecture
Ada Lovelace
GPU Codename
AD104
AMD Equivalent
- Radeon RX 7900 XT
Successor
- GeForce RTX 5070
Recommended CPU
- Intel Core i7 14700K or above
Graphics Processing Unit
Base Clock
1920 MHz
Boost Clock
2475 MHz
Shading Units
5888
Texture Mapping Units (TMUs)
184
Render Output Units (ROPs)
64
Compute Units (Pipelines)
46
Tensor Cores
184
Ray-tracing Cores
46
L1 Cache
128KB per cluster
L2 Cache
36MB shared
Instructions Per Cycle
2 IPC
Raw Performance
Pixel Fill Rate
158 GPixel/s
Texture Fill Rate
455 GTexel/s
Physical
Interface
PCIe 4.0 x16
TGP
200 W
Manufacturing
TSMC
Fabrication Process
5 nm
Die Size
294 mm²
Transistor Count
35 billion
Transistor Density
119.05 MTr/mm²
Memory
Memory Type
GDDR6X
Memory Size
12 GB
Memory Clock
1313 MHz
Effective Memory Speed
21000 Mbps
Bus
192-bit
ECC
No
API
DirectX
12
Vulkan
1.3
OpenGL
4.6
OpenCL
3.0
CUDA
8.9
Ray Tracing
Yes
DLSS
DLSS 3
DisplayPort
1.4a
Specifications compiled from NanoReview. Last updated 2026-09-29.
More graphics cards
View all →97
Nvidia
Nvidia RTX Pro 6000 Blackwell
- Memory96 GB
- Memory TypeGDDR7
- Boost Clock2617 MHz
- TGP600 W
Desktop GPU
96
Nvidia
GeForce RTX 5090
- Memory32 GB
- Memory TypeGDDR7
- Boost Clock2407 MHz
- TGP575 W
Desktop GPU
89
Nvidia
GeForce RTX 4090
- Memory24 GB
- Memory TypeGDDR6X
- Boost Clock2520 MHz
- TGP450 W
Desktop GPU
84
Nvidia
GeForce RTX 5080
- Memory16 GB
- Memory TypeGDDR7
- Boost Clock2617 MHz
- TGP360 W
Desktop GPU