GeForce RTX 4090 vs RTX 3090
VS
Nvidia GeForce RTX 4090
AD102
Nvidia GeForce RTX 3090
GA102
We compared two discrete desktop gaming GPUs: the GeForce RTX 4090 with 128 pipelines and 16384 shaders against the 2 years older RTX RTX 3090 that utilizes 82 pipelines and 10496 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 93%) in 3DMark Steel Nomad Lite
- 2.3x higher maximum theoretical performance (82.6 vs 35.6 TFLOPS)
- Shows 48% higher average frame rate in modern games at QHD resolution – 177 vs 120 FPS
- Manufactured using a more efficient 5 nm process technology
- Features 184 more tensor cores for effective ML and AI workloads
- Supports Nvidia DLSS 3 technology
- Achieves 49% more points in the GeekBench 6 Compute test (316K vs 211K)
- Has 56% more shading units (16384 vs 10496)
Gaming Performance
Frame rate comparison across popular AAA titles at different resolutionsGames
| Forza Horizon 5 | 275 199 193 155 |
186 128 117 91 |
| The Witcher 3 | 292 268 191 101 |
205 170 128 66 |
| Counter-Strike 2 | 339 238 188 90 |
275 207 155 83 |
| Far Cry 6 | 203 179 170 104 |
156 145 124 77 |
| Hogwarts Legacy | 200 163 132 74 |
128 108 86 47 |
| CoD: Modern Warfare III | 266 263 219 151 |
171 159 123 80 |
| Ghost of Tsushima | 188 152 137 91 |
120 97 85 48 |
| Cyberpunk 2077 | 226 208 143 66 |
148 132 90 43 |
| Shadow of the Tomb Raider | 302 284 222 118 |
237 226 168 91 |
| 1080p High | 255 | 181 |
| 1080p Ultra | 217 | 152 |
| 1440p Ultra | 177 | 120 |
| 4K Ultra | 106 | 70 |
| Margin of Error | Low | Low |
Benchmarks
Graphics cards’ performance in recent benchmarking apps3D Mark
Steel Nomad Lite Score
| Time Spy | 36299 | 19896 |
| Solar Bay | 187525 | 97395 |
| Port Royal | 26127 | 13633 |
| Fire Strike | 72432 | 47512 |
| Wild Life Extreme | 85218 | 43669 |
| Night Raid | 195813 | 139766 |
GeekBench 6 OpenCL
GB6 Compute Score
| Background Blur | 318.9 img/sec | 320.8 img/sec |
| Face Detection | 222.8 img/sec | 207.2 img/sec |
| Horizon Detection | 14.2 Gpixels/sec | 10.1 Gpixels/sec |
| Edge Detection | 21 Gpixels/sec | 14.4 Gpixels/sec |
| Gaussian Blur | 23.9 Gpixels/sec | 11.5 Gpixels/sec |
| Feature Matching | 2.38 Gpixels/sec | 1.48 Gpixels/sec |
| Stereo Matching | 1950 Gpixels/sec | 984.2 Gpixels/sec |
| Particle Physics | 48090.4 FPS | 27764.9 FPS |
| API | OpenCL | OpenCL |
Cinebench 2024 GPU
Cinebench 2024 GPU
Passmark Graphics
G3D Mark Score
| G2D Mark | 1303 | 1069 |
| DirectX 11 | 323 FPS | 218 FPS |
| DirectX 12 | 151 FPS | 109 FPS |
| GPU Compute | 25991 Ops/s | 15028 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 3090
No benchmark results yet
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 | 10665 |
| Image Segmentation (HP) | 36657 | 28963 |
| Image Super Resolution (Q) | 48114 | 22697 |
| Face Detection (HP) | 78111 | 55232 |
| Pose Estimation (Q) | 289639 | 128155 |
| Text Classification (SP) | 4074 | 3152 |
| Machine Translation (HP) | 6349 | 5088 |
| Object Detection (SP) | 21186 | 13891 |
| Depth Estimation (Q) | 75599 | 37984 |
| Style Transfer (SP) | 601428 | 272000 |
| Framework | ONNX | ONNX |
| Backend | DirectML | DirectML |
Specifications
Technical specifications of GeForce RTX 4090 and RTX 3090General
| Vendor | Nvidia | Nvidia |
| Build | Discrete | Discrete |
| Released | September 20, 2022 | September 24, 2020 |
| Launch price (MSRP) | $1599 | $1499 |
| Case | Desktop | Desktop |
| Purpose | Gaming | Gaming |
| Segment | High-end | High-end |
| Architecture | Ada Lovelace | Ampere |
| GPU Codename | AD102 | GA102 |
| Rival Equivalent | - | - Radeon RX 6900 XT |
| Successor | - GeForce RTX 5090 | - GeForce RTX 5090 |
| Recommended CPU | - Intel Core i9 14900K or above | - Intel Core i9 12900K or above |
Desktop GPU rating (3rd and 20th place)
Graphics Processing Unit
| Base Clock | 2235 MHz | 1395 MHz |
| Boost Clock | 2520 MHz | 1695 MHz |
| Shading Units | 16384 | 10496 |
| Texture Mapping Units (TMUs) | 512 | 328 |
| Render Output Units (ROPs) | 176 | 112 |
| Compute Units (Pipelines) | 128 | 82 |
| Tensor Cores | 512 | 328 |
| Ray-tracing Cores | 128 | 82 |
| L1 Cache | 128KB per cluster | 128KB per cluster |
| L2 Cache | 72MB shared | 6MB shared |
| Instructions Per Cycle | 2 IPC | 2 IPC |
Raw Performance
| Pixel Fill Rate | 444 GPixel/s | 190 GPixel/s |
| Texture Fill Rate | 1290 GTexel/s | 556 GTexel/s |
FLOPS (FP32)
Physical
| Interface | PCIe 4.0 x16 | PCIe 4.0 x16 |
| TGP | 450 W | 350 W |
| Manufacturing | TSMC | Samsung |
| Fabrication Process | 5 nm | 8 nm |
| Die Size | 609 mm² | 628 mm² |
| Transistor Count | 76.3 billion | 28 billion |
| Transistor Density | 125.29 MTr/mm² | 44.59 MTr/mm² |
Memory
| Memory Type | GDDR6X | GDDR6X |
| Memory Size | 24 GB | 24 GB |
| Memory Clock | 2625 MHz | 1219 MHz |
| Effective Memory Speed | 21000 Mbps | 19500 Mbps |
| Bus | 384-bit | 384-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 | 8.6 |
| Ray Tracing | Yes | Yes |
| DLSS | DLSS 3 | DLSS 2 |
| DisplayPort | 1.4a | 1.4a |
Cast your vote
Total votes: 58