GeForce RTX 5070 SUPER
Graphics Card GeForce RTX 5070 SUPER — Specifications, Tests & Benchmarks
GeForce RTX 5070 SUPER
Nvidia GeForce 50
•
GB205
•
2026
•
18 GB
•
GDDR7
•
192 bit
•
275W
🏆 Top #1125
Performance
Synthetic Score
29,803
Efficiency
enthusiast
The GeForce RTX 5070 SUPER graphics card from NVIDIA is built on the Blackwell 2.0 architecture and was officially introduced on 01.01.2026. At the core of the card is a chip created with 4N FinFET lithography. The die houses 31.1 billion transistorsistors transistors, and its area is 263 mm². The model's computational arsenal includes 6400 universal processors. Also available are 50 specialized cores for ray rendering and 200 cores for accelerating artificial intelligence. The GDDR7 video memory capacity is 18 GB. Thanks to the 192 bit-bit bus, bandwidth reaches 672.00 GB/s, eliminating delays when working with demanding textures. In standard mode, the core operates at 2325 MHz, but auto-overclocking technology allows it to reach 2512 MHz under load. The TDP level for this model is 275 W, so for a build, it\'s advisable to choose quality cooling and a power supply unit of at least 600 W.
General Information
| Release Date | 2026 |
| Production Status | Active |
| Usage Class | Hardcore Gaming |
| Market Segment | Desktop |
| Series | Nvidia GeForce 50 |
| GPU Variant | GB205-400-A1 |
GPU
| Architecture | Blackwell 2.0 |
| Codename | GB205 |
| Process Node | 5 nm (TSMC) |
| Process Type | 4N FinFET |
| Transistor Count | 31,100,000.0 million |
| Die Size | 263 mm² |
| Transistor Density | 118,300,000 million / mm² |
Rendering Configuration
| Shading Units (ALU) | 6400 |
| Texture Mapping Units (TMU) | 200 |
| Render Output Units (ROP) | 80 |
| Tensor Cores | 200 |
| RT Cores (Ray Tracing) | 50 |
| Compute Units (SM/CU) | 50 |
Clock Speeds
| Base Clock | 2325 MHz |
| Boost Clock | 2512 MHz |
| Memory Clock (Actual) | 1750 MHz |
| Memory Clock (Effective) | 28000 MHz |
Memory
| Memory Capacity | 18 GB |
| Memory Type | GDDR7 |
| Memory Bus | 192 bit bit |
| Memory Bandwidth | 672.00 GB/s |
Cache Memory
| L1 Cache | 6400 KB |
| L2 Cache | 48 MB |
Board Design & Power
| Interface | PCIe 5.0 x16 |
| Slot Width | 2 |
| Dimensions (L x W x H) | 245 x 40 x 115 mm |
| Power Consumption (TDP) | 275 W |
| Recommended PSU | 600 W |
| Power Connectors | 1x 16-pin |
| Board Number | PG147 SKU 65 |
API Support
| DirectX | 12 Ultimate (12_2) |
| OpenGL | 4.6 |
| OpenCL | |
| Vulkan | 1.4 |
| CUDA | |
| Shader Model |
Theoretical Performance
| Pixel Fill Rate | 201000 GPixel/s |
| Texture Fill Rate | 502400 GTexel/s |
| FP16 Performance (Half) | 32.15 TFLOPS (1:1) |
| FP32 Performance (Float) | 32.15 TFLOPS |
| FP64 Performance (Double) | 502.4 GFLOPS (1:64) |
Synthetic Tests & Benchmarks
| TPU Index Score | 46.48% |
| Performance Tier | enthusiast |
How this graphics card performs in games
Всі тести
Тестів поки немає.
Comments
0Please log in to your account to leave a comment.
Log In or RegisterNo comments yet. Be the first to comment!
Are you sure you want to delete this comment? This action cannot be undone.