GeForce RTX 4090 vs RTX 4070
VS
Nvidia GeForce RTX 4090
AD102
Nvidia GeForce RTX 4070
AD104
We compared two discrete desktop gaming GPUs: the GeForce RTX 4090 24 GB with 128 pipelines and 16384 shaders against the 7 months newer RTX RTX 4070 12 GB that utilizes 46 pipelines and 5888 shaders. Here you will find complete details about specs, efficiency, performance tests, and more.
Review
Gaming
Performance in DirectX, OpenCL, and Vulkan games
Workstation
Perf. in 3D modeling, video editing and rendering apps
AI/ML
Capabilities for machine learning and AI-related tasks
Energy Efficiency
Power consumption efficiency in different scenarios
DeviceFacts Final Score
Overall video card score
Value for money (Beta)
Key differences
Reasons to consider the GeForce RTX 4090
- Performs significantly better (up to 2.3x) in 3DMark Steel Nomad Lite
- 2.8x higher maximum theoretical performance (82.6 vs 29.1 TFLOPS)
- Shows 67% higher average frame rate in modern games at QHD resolution – 177 vs 106 FPS
- Includes 12 GB more video memory
- Features 328 more tensor cores for effective ML and AI workloads
- Has 2x higher memory bandwidth: 1010 vs 504 GB/s
- Achieves 89% more points in the GeekBench 6 Compute test (316K vs 167K)
- Has 2.8x more shading units (16384 vs 5888)
Gaming Performance
Frame rate comparison across popular AAA titles at different resolutionsGames
| Forza Horizon 5 | 275 199 193 155 |
172 115 102 79 |
| The Witcher 3 | 292 268 191 101 |
194 159 114 61 |
| Counter-Strike 2 | 339 238 188 90 |
272 205 149 81 |
| Far Cry 6 | 203 179 170 104 |
149 133 115 68 |
| Hogwarts Legacy | 200 163 132 74 |
117 97 77 39 |
| CoD: Modern Warfare III | 266 263 219 151 |
149 145 107 69 |
| Ghost of Tsushima | 188 152 137 91 |
106 88 72 43 |
| Cyberpunk 2077 | 226 208 143 66 |
129 119 75 37 |
| Shadow of the Tomb Raider | 302 284 222 118 |
211 190 145 80 |
| 1080p High | 255 | 167 |
| 1080p Ultra | 217 | 139 |
| 1440p Ultra | 177 | 106 |
| 4K Ultra | 106 | 62 |
| Margin of Error | Low | Low |
Benchmarks
Graphics cards’ performance in recent benchmarking apps3D Mark
Steel Nomad Lite Score
| Time Spy | 36299 | 17799 |
| Solar Bay | 187525 | 82861 |
| Port Royal | 26127 | 11156 |
| Fire Strike | 72432 | 44067 |
| Wild Life Extreme | 85218 | 35472 |
| Night Raid | 195813 | 153385 |
GeekBench 6 OpenCL
GB6 Compute Score
| Background Blur | 318.9 img/sec | 233 img/sec |
| Face Detection | 222.8 img/sec | 141.7 img/sec |
| Horizon Detection | 14.2 Gpixels/sec | 6.62 Gpixels/sec |
| Edge Detection | 21 Gpixels/sec | 9.19 Gpixels/sec |
| Gaussian Blur | 23.9 Gpixels/sec | 9.2 Gpixels/sec |
| Feature Matching | 2.38 Gpixels/sec | 1.78 Gpixels/sec |
| Stereo Matching | 1950 Gpixels/sec | 845.4 Gpixels/sec |
| Particle Physics | 48090.4 FPS | 25043.9 FPS |
| API | OpenCL | OpenCL |
Cinebench 2024 GPU
Cinebench 2024 GPU
Passmark Graphics
G3D Mark Score
| G2D Mark | 1303 | 1165 |
| DirectX 11 | 323 FPS | 243 FPS |
| DirectX 12 | 151 FPS | 103 FPS |
| GPU Compute | 25991 Ops/s | 14561 Ops/s |
Blender
Blender GPU
Recent User Tests
GeForce RTX 4090
| Date | Benchmark | Result |
|---|---|---|
| 📘 2026-10-06 (Devansh) | Geekbench 6 OpenCL | 394674 |
| 📘 2026-10-06 (Devansh) | Geekbench 7 OpenCL | 338129 |
| 📘 2026-04-03 (afroman420IU) | Cinebench 2024 | 32578 |
GeForce RTX 4070
| Date | Benchmark | Result |
|---|---|---|
| 📘 2025-12-14 (kashitsu) | Cinebench 2024 | 18051 |
Artificial Intelligence Tests
Performance in machine learning and artificial intelligence tasksGeekBench 6 ML
GB6 ML Single Precision
GB6 ML Half Precision
GB6 ML Quantized
| Image Classification (SP) | 15600 | 10246 |
| Image Segmentation (HP) | 36657 | 26153 |
| Image Super Resolution (Q) | 48114 | 32054 |
| Face Detection (HP) | 78111 | 51337 |
| Pose Estimation (Q) | 289639 | 124617 |
| Text Classification (SP) | 4074 | 3703 |
| Machine Translation (HP) | 6349 | 5484 |
| Object Detection (SP) | 21186 | 13134 |
| Depth Estimation (Q) | 75599 | 50548 |
| Style Transfer (SP) | 601428 | 263477 |
| Framework | ONNX | ONNX |
| Backend | DirectML | DirectML |
Specifications
Technical specifications of GeForce RTX 4090 and RTX 4070General
| Vendor | Nvidia | Nvidia |
| Build | Discrete | Discrete |
| Released | September 20, 2022 | April 13, 2023 |
| Launch price (MSRP) | $1599 | $599 |
| Case | Desktop | Desktop |
| Purpose | Gaming | Gaming |
| Segment | High-end | Mid-range |
| Architecture | Ada Lovelace | Ada Lovelace |
| GPU Codename | AD102 | AD104 |
| Rival Equivalent | - | - Radeon RX 7900 XT |
| Successor | - GeForce RTX 5090 | - GeForce RTX 5070 |
| Recommended CPU | - Intel Core i9 14900K or above | - Intel Core i7 14700K or above |
Desktop GPU rating (3rd and 30th place)
Graphics Processing Unit
| Base Clock | 2235 MHz | 1920 MHz |
| Boost Clock | 2520 MHz | 2475 MHz |
| Shading Units | 16384 | 5888 |
| Texture Mapping Units (TMUs) | 512 | 184 |
| Render Output Units (ROPs) | 176 | 64 |
| Compute Units (Pipelines) | 128 | 46 |
| Tensor Cores | 512 | 184 |
| Ray-tracing Cores | 128 | 46 |
| L1 Cache | 128KB per cluster | 128KB per cluster |
| L2 Cache | 72MB shared | 36MB shared |
| Instructions Per Cycle | 2 IPC | 2 IPC |
Raw Performance
| Pixel Fill Rate | 444 GPixel/s | 158 GPixel/s |
| Texture Fill Rate | 1290 GTexel/s | 455 GTexel/s |
FLOPS (FP32)
Physical
| Interface | PCIe 4.0 x16 | PCIe 4.0 x16 |
| TGP | 450 W | 200 W |
| Manufacturing | TSMC | TSMC |
| Fabrication Process | 5 nm | 5 nm |
| Die Size | 609 mm² | 294 mm² |
| Transistor Count | 76.3 billion | 35 billion |
| Transistor Density | 125.29 MTr/mm² | 119.05 MTr/mm² |
Memory
| Memory Type | GDDR6X | GDDR6X |
| Memory Size | 24 GB | 12 GB |
| Memory Clock | 2625 MHz | 1313 MHz |
| Effective Memory Speed | 21000 Mbps | 21000 Mbps |
| Bus | 384-bit | 192-bit |
| ECC | No | No |
Memory Bandwidth
API
| DirectX | 12 | 12 |
| Vulkan | 1.3 | 1.3 |
| OpenGL | 4.6 | 4.6 |
| OpenCL | 3.0 | 3.0 |
| CUDA | 8.9 | 8.9 |
| Ray Tracing | Yes | Yes |
| DLSS | DLSS 3 | DLSS 3 |
| DisplayPort | 1.4a | 1.4a |
Cast your vote
Total votes: 18