AMD Radeon HD 8950 OEM vs NVIDIA Tesla K8
What is the difference between AMD Radeon HD 8950 OEM and NVIDIA Tesla K8. Find out which graphics card has better performance.
Graphics Processor (GPU)
| Tahiti | GPU Name | GK104 |
| GCN 1.0 | Architecture | Kepler |
| TSMC | Foundry | TSMC |
| 28 nm | Process Size | 28 nm |
| 4,313 million | Transistors | 3,540 million |
| 352 mm² | Die Size | 294 mm² |
Graphics Card
| Jan 8th, 2013 | Release Date | Sep 16th, 2014 |
| Sea Islands (HD 8900) | Family | Tesla (Kxx) |
| End-of-life | Production | End-of-life |
| PCIe 3.0 x16 | Bus Interface | PCIe 2.0 x16 |
Memory
| 3 GB | Memory Size | 8 GB |
| GDDR5 | Memory Type | GDDR5 |
| 384 bit | Memory Bus | 256 bit |
| 240.0 GB/s | Bandwidth | 160.0 GB/s |
Performance
| 29.60 GPixel/s | Pixel fillrate | 25.95 GPixel/s |
| 103.6 GTexel/s | Texture fillrate | 103.8 GTexel/s |
| 3.315 TFLOPS | FP32 (float) performance | 2.491 TFLOPS |
| 828.8 GFLOPS (1:4) | FP64 (double) performance | 103.8 GFLOPS (1:24) |
Clock Speeds
| 850 MHz | Base Clock | 693 MHz |
| 925 MHz | Boost Clock | 811 MHz |
| 1250 MHz 5 Gbps effective | Memory Clock | 1250 MHz 5 Gbps effective |
Render Config
| 1792 | Shading Units | 1536 |
| 112 | TMUs | 128 |
| 32 | ROPs | 32 |
| 16 KB (per CU) | L1 Cache | 16 KB (per SMX) |
| 768 KB | L2 Cache | 512 KB |
Board Design
| Dual-slot | Slot Width | Single-slot |
| 267 mm 10.5 inches | Length | 241 mm 9.5 inches |
| 200 W | Thermal design power (TDP) | 100 W |
| 550 W | Suggested PSU | 300 W |
| 1x DVI 1x HDMI 2x mini-DisplayPort | Display Connectors | No outputs |
| 2x 6-pin | Power Connectors | 1x 6-pin |
| C386-42 | Board Number | P2004 |
API support
| 12 (11_1) | DirectX | 12 (11_0) |
| 4.6 | OpenGL | 4.6 |
| 1.2 | OpenCL | 3.0 |
| 1.2 | Vulkan | 1.1 |
| 5.1 | Shader Model | 5.1 |