Last updated：2021-06-15 11:07:10
A GEPC instance can provide eight NVIDIA Tesla V100 GPU accelerators with 512 GB memory, which means 40960 CUDA cores, 5120 Tensor cores, and a single-precision floating-point computing capability of up to 112 TFLOPS. Powered by the up-to-date NVIDIA Volta architecture, a GEPC instance provides excellent performance for deep learning and high-performance computing.
A GEPC instance can support a lossless 100-Gbps RDMA network cluster that consists of up to 256 nodes, satisfying user requirements for high-performance computing, high-bandwidth, and low-latency clusters.
GEPC allows you to use up-to-date GPU accelerator hardware at a low server investment risk and a low server operation cost.
VPCs of users are isolated from each other, and Kingsoft Cloud security products are used.