Tesla S2050 1U GPU high-performance computing server provides the world's highest computing density and can be used to deploy clusters and data centers in enterprise departments.
The Tesla S2050 1U is a 1U high‑density GPU server designed for high‑performance and scientific computing. It integrates four Tesla C2050/C2070 GPUs, supports ECC memory, and provides both single‑ and double‑precision CUDA acceleration.
Key specifications include 12 GB GDDR5 GPU memory (4 × 3 GB), typical power draw of 900 W (max 1200 W), physical dimensions of 1.71 in × 17.425 in × 28.5 in, PCIe x16 Gen2 interface, and peak performances of 4.13 TFLOPS (single‑precision) and 2.06 TFLOPS (double‑precision). The system supports CUDA, OpenCL, Fortran, DirectCompute and NVIDIA Parallel Nsight tools. This manual is written in Chinese.
| GPU Memory | 12 GB GDDR5 (4 × 3 GB) |
|---|---|
| Typical Power Consumption | 900 W |
| Maximum Power Consumption | 1200 W |
| Physical Dimensions | 1.71 in × 17.425 in × 28.5 in (H × W × D) |
| Single‑Precision Peak Performance | 4.13 TFLOPS |
| Double‑Precision Peak Performance | 2.06 TFLOPS |
| Memory Bandwidth | 148 GB/s |
| PCIe Interface | PCIe x16 Gen2 |
Typical power consumption is 900 W, with a maximum of 1200 W.
It delivers up to 2.06 TFLOPS of double‑precision performance.