The hardware support of SP4C3 is more focused on AI calculation. Combined with the system architecture, we give the minimum parameters of hardware equipment that can be used to test the network in the early stage.
In the internal testing phase, we will test data with different AI servers and publish corresponding data. To provide hardware equipment suggestions for computing power providers joining SP4C3 network in the future, and better help SP4C3 can be put into application.
AI Server
*Minimum Requirement:
Computing Power 10.0 TFLOPs MAX FP32 Compute
CPU Core 8 Cores @2.0GHz
Disk Storage 16 TB
Support Framework Paddlepaddle, Tensorflow, Pytorch, Caffe.
Network Bandwidth 50Mbps Upstream&Downstream
Hardware Video Encoding 50 FPS@1080p H.265
Hardware Video Decoding 500 FPS@1080p H.265
Graphics Memory 12GB
SP4C3 also welcomes more hardware manufacturers to provide us with different AI servers and related hardware to build SP4C3 network together.
Topic: [ANN] Sp4c3: Hardware Equipment Recommendations (Read 92 times)
Jump to: