Intel Arc Pro B70 vs GeForce RTX 5090
VS
Intel Arc Pro B70
BMG-G31
Nvidia GeForce RTX 5090
GB202-300
We compared two discrete desktop graphics cards: the Intel Arc Pro B70 with 32 pipelines and 4096 shaders against the 1 year and 2 months older GeForce RTX 5090 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
NanoReview Final Score
Overall video card score
Scores marked with a red asterisk (**) represent initial estimates from early testing.
Value for money (Beta)
Key differences
Reasons to consider the GeForce RTX 5090
- Performs significantly better (up to 3.5x) in 3DMark Steel Nomad Lite
- 4.6x higher maximum theoretical performance (104.8 vs 22.9 TFLOPS)
- Shows 2.5x higher average frame rate in modern games at QHD resolution – 207 vs 82 FPS
- Manufactured using a more efficient 4 nm process technology
Gaming Performance
Frame rate comparison across popular AAA titles at different resolutionsGames
| Forza Horizon 5 | 144 97 84 62 |
336 247 233 187 |
| The Witcher 3 | 182 155 116 62 |
364 327 237 123 |
| Counter-Strike 2 | 181 128 96 45 |
363 265 202 99 |
| Far Cry 6 | 125 114 93 55 |
237 202 190 121 |
| Hogwarts Legacy | 89 71 56 32 |
249 199 166 97 |
| CoD: Modern Warfare III | 108 110 80 51 |
334 322 260 184 |
| Ghost of Tsushima | 83 71 53 29 |
221 185 175 113 |
| Cyberpunk 2077 | 109 97 63 29 |
261 233 160 77 |
| Shadow of the Tomb Raider | 133 123 94 50 |
329 302 238 128 |
| 1080p High | 128 | 299 |
| 1080p Ultra | 107 | 254 |
| 1440p Ultra | 82 | 207 |
| 4K Ultra | 46 | 125 |
| Margin of Error | High | Low |
Benchmarks
Graphics cards’ performance in recent benchmarking apps3D Mark
Steel Nomad Lite Score
| Time Spy | 18329 | 46873 |
| Solar Bay | 76054 | 236689 |
| Port Royal | 10682 | 38202 |
| Fire Strike | 46210 | 87569 |
| Wild Life Extreme | 37743 | 108751 |
| Night Raid | 183003 | 206648 |
GeekBench 6 GPU Compute
GB6 Compute Score
| Background Blur | 413.1 img/sec | 333 img/sec |
| Face Detection | 226.4 img/sec | 298.5 img/sec |
| Horizon Detection | 5.05 Gpixels/sec | 21.7 Gpixels/sec |
| Edge Detection | 9.66 Gpixels/sec | 33.7 Gpixels/sec |
| Gaussian Blur | 10.6 Gpixels/sec | 34.5 Gpixels/sec |
| Feature Matching | 1.67 Gpixels/sec | 2.53 Gpixels/sec |
| Stereo Matching | 712.1 Gpixels/sec | 2850 Gpixels/sec |
| Particle Physics | 20998.9 FPS | 59146.9 FPS |
| API | Vulkan | OpenCL |
Passmark Graphics
G3D Mark Score
| G2D Mark | 689 | 1410 |
| DirectX 11 | 142 FPS | 340 FPS |
| DirectX 12 | 96 FPS | 179 FPS |
| GPU Compute | 9800 Ops/s | 24383 Ops/s |
Blender
Blender GPU
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) | - | 18948 |
| Image Segmentation (HP) | - | 55109 |
| Image Super Resolution (Q) | - | 58478 |
| Face Detection (HP) | - | 114685 |
| Pose Estimation (Q) | - | 344729 |
| Text Classification (SP) | - | 4464 |
| Machine Translation (HP) | - | 8184 |
| Object Detection (SP) | - | 27802 |
| Depth Estimation (Q) | - | 85572 |
| Style Transfer (SP) | - | 687080 |
| Framework | - | ONNX |
| Backend | - | DirectML |
Specifications
Technical specifications of Intel Arc Pro B70 and GeForce RTX 5090General
| Vendor | Intel | Nvidia |
| Build | Discrete | Discrete |
| Released | March 25, 2026 | January 7, 2025 |
| Launch price (MSRP) | $949 | $1999 |
| Case | Desktop | Desktop |
| Purpose | Professional | Gaming |
| Segment | High-end | High-end |
| Architecture | Battlemage Pro Series | Blackwell 2.0 |
| GPU Codename | BMG-G31 | GB202-300 |
| Recommended CPU | - | - Intel Core Ultra 9 285K or above |
Desktop GPU rating (28th and 2nd place)
Graphics Processing Unit
| Base Clock | 2280 MHz | 2017 MHz |
| Boost Clock | 2800 MHz | 2407 MHz |
| Shading Units | 4096 | 21760 |
| Texture Mapping Units (TMUs) | 256 | 680 |
| Render Output Units (ROPs) | 128 | 176 |
| Compute Units (Pipelines) | 32 | 170 |
| Tensor Cores | 256 | 680 |
| Ray-tracing Cores | 32 | 170 |
| L1 Cache | 256KB per cluster | 128KB per cluster |
| L2 Cache | 16MB shared | 96MB shared |
| Instructions Per Cycle | 2 IPC | 2 IPC |
Raw Performance
| Pixel Fill Rate | 358 GPixel/s | 424 GPixel/s |
| Texture Fill Rate | 717 GTexel/s | 1637 GTexel/s |
FLOPS (FP32)
Physical
| Interface | PCIe 5.0 x16 | PCIe 5.0 x16 |
| TGP | 230 W | 575 W |
| Manufacturing | TSMC | TSMC |
| Fabrication Process | 5 nm | 4 nm |
| Die Size | - | 750 mm² |
| Transistor Count | - | 92.2 billion |
| Transistor Density | - | 122.93 MTr/mm² |
Memory
| Memory Type | GDDR6 | GDDR7 |
| Memory Size | 32 GB | 32 GB |
| Memory Clock | 2375 MHz | 1750 MHz |
| Effective Memory Speed | 19000 Mbps | 28000 Mbps |
| Bus | 256-bit | 512-bit |
| ECC | No | No |
Memory Bandwidth
API
| DirectX | 12.2 | 12.2 |
| Vulkan | 1.3 | 1.4 |
| OpenGL | 4.6 | 4.6 |
| OpenCL | 3.0 | 3.0 |
| CUDA | No | 12.0 |
| Ray Tracing | Yes | Yes |
| DLSS | No | DLSS 4 |
| DisplayPort | 2.1a | 2.1b |
Cast your vote
Total votes: 16