AMD Radeon RX Vega 3 vs GeForce RTX 5090
VS
AMD Radeon RX Vega 3
Picasso
Nvidia GeForce RTX 5090
GB202-300
We performed a head-to-head comparison of the Radeon RX Vega 3 with 3 pipelines and 192 shaders against the 6 years and 1 month newer GeForce RTX 5090 32 GB that utilizes 170 pipelines and 21760 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
The "Energy Efficiency" metric has less impact on the DeviceFacts Score for desktop GPU.
Key differences
Reasons to consider the GeForce RTX 5090
- Performs significantly better (up to 135.3x) in 3DMark Steel Nomad Lite
- 209.6x higher maximum theoretical performance (104.8 vs 0.5 TFLOPS)
- Shows 41.4x higher average frame rate in modern games at QHD resolution – 207 vs 5 FPS
- Manufactured using a more efficient 4 nm process technology
- Ray-tracing support with 170 dedicated RT cores
Gaming Performance
Frame rate comparison across popular AAA titles at different resolutionsGames
| Forza Horizon 5 | 11 - - - |
336 247 233 187 |
| The Witcher 3 | - | 364 327 237 123 |
| Counter-Strike 2 | 18 12 - - |
363 265 202 99 |
| Far Cry 6 | 9 - - - |
237 202 190 121 |
| Hogwarts Legacy | - | 249 199 166 97 |
| CoD: Modern Warfare III | 12 9 9 - |
334 322 260 184 |
| Ghost of Tsushima | - - - - |
221 185 175 113 |
| Cyberpunk 2077 | - | 261 233 160 77 |
| Shadow of the Tomb Raider | - | 329 302 238 128 |
| 1080p High | 9 | 299 |
| 1080p Ultra | - | 254 |
| 1440p Ultra | - | 207 |
| 4K Ultra | - | 125 |
| Margin of Error | Low | Low |
Benchmarks
Graphics cards’ performance in recent benchmarking apps3D Mark
Steel Nomad Lite Score
| Time Spy | 412 | 46873 |
| Solar Bay | - | 236689 |
| Port Royal | - | 38202 |
| Fire Strike | 1448 | 87569 |
| Wild Life Extreme | 867 | 108751 |
| Night Raid | 6135 | 206648 |
GeekBench 6 OpenCL
GB6 Compute Score
| Background Blur | 10.1 img/sec | 333 img/sec |
| Face Detection | 4.97 img/sec | 298.5 img/sec |
| Horizon Detection | 0.16 Gpixels/sec | 21.7 Gpixels/sec |
| Edge Detection | 0.33 Gpixels/sec | 33.7 Gpixels/sec |
| Gaussian Blur | 0.22 Gpixels/sec | 34.5 Gpixels/sec |
| Feature Matching | 0.04 Gpixels/sec | 2.53 Gpixels/sec |
| Stereo Matching | 14.1 Gpixels/sec | 2850 Gpixels/sec |
| Particle Physics | 500.2 FPS | 59146.9 FPS |
| API | OpenCL | OpenCL |
Passmark Graphics
G3D Mark Score
| G2D Mark | 302 | 1410 |
| DirectX 11 | 7 FPS | 340 FPS |
| DirectX 12 | 4 FPS | 179 FPS |
| GPU Compute | 441 Ops/s | 24383 Ops/s |
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) | 476 | 18948 |
| Image Segmentation (HP) | 498 | 55109 |
| Image Super Resolution (Q) | 982 | 58478 |
| Face Detection (HP) | 1535 | 114685 |
| Pose Estimation (Q) | 2770 | 344729 |
| Text Classification (SP) | 430 | 4464 |
| Machine Translation (HP) | 706 | 8184 |
| Object Detection (SP) | 532 | 27802 |
| Depth Estimation (Q) | 1792 | 85572 |
| Style Transfer (SP) | 5517 | 687080 |
| Framework | ONNX | ONNX |
| Backend | DirectML | DirectML |
Specifications
Technical specifications of Radeon RX Vega 3 and GeForce RTX 5090General
| Vendor | Amd | Nvidia |
| Build | Integrated | Discrete |
| Released | January 1, 2019 | January 7, 2025 |
| Launch price (MSRP) | - | $1999 |
| Case | Laptop | Desktop |
| Purpose | Generic | Gaming |
| Segment | Entry-level | High-end |
| Architecture | GCN 5.0 | Blackwell 2.0 |
| GPU Codename | Picasso | GB202-300 |
| Recommended CPU | - AMD Ryzen 3 3200U or above | - Intel Core Ultra 9 285K or above |
| Used in CPUs | - AMD Ryzen 3 3250U - AMD Ryzen 3 3200U |
- |
Laptop GPU ranking (#144th place)
Desktop GPU rating (#2nd place)
Graphics Processing Unit
| Base Clock | 300 MHz | 2017 MHz |
| Boost Clock | 1000-1200 MHz (configurable) | 2407 MHz |
| Shading Units | 192 | 21760 |
| Texture Mapping Units (TMUs) | 12 | 680 |
| Render Output Units (ROPs) | 4 | 176 |
| Compute Units (Pipelines) | 3 | 170 |
| Tensor Cores | No | 680 |
| Ray-tracing Cores | No | 170 |
| L1 Cache | - | 128KB per cluster |
| L2 Cache | - | 96MB shared |
| Instructions Per Cycle | 2 IPC | 2 IPC |
Raw Performance
| Pixel Fill Rate | 4-5 GPixel/s | 424 GPixel/s |
| Texture Fill Rate | 12-14 GTexel/s | 1637 GTexel/s |
FLOPS (FP32)
Physical
| Interface | IGP | PCIe 5.0 x16 |
| TGP | 15 W | 575 W |
| Manufacturing | GlobalFoundries | TSMC |
| Fabrication Process | 14 nm | 4 nm |
| Die Size | 210 mm² | 750 mm² |
| Transistor Count | 4.94 billion | 92.2 billion |
| Transistor Density | 23.52 MTr/mm² | 122.93 MTr/mm² |
Memory
| Memory Type | System Shared | GDDR7 |
| Memory Size | System RAM Dependent | 32 GB |
| Memory Clock | System RAM Dependent | 1750 MHz |
| Effective Memory Speed | System RAM Dependent | 28000 Mbps |
| Bus | - | 512-bit |
| ECC | No | No |
Memory Bandwidth
API
| DirectX | 12 | 12.2 |
| Vulkan | 1.2.131 | 1.4 |
| OpenGL | 4.6 | 4.6 |
| OpenCL | 2.0 | 3.0 |
| CUDA | - | 12.0 |
| Ray Tracing | No | Yes |
| DLSS | No | DLSS 4 |
| DisplayPort | - | 2.1b |
Cast your vote
Total votes: 11
