In order to simplify the complexity of artificial intelligence development on the cloud GPU platform, NCHC, therefore, launched the Taiwan GPU Cloud (TWGC), cloud GPU software service, and continuously updated the version with different hosts to meet the needs of developers.
For research developers and data experts, there are two major problems if you want to master deep learning. The first is to stack the indispensable software components in a single system, including in-depth learning architecture, functional library, operating system and drivers, and second, how to obtain the latest GPU computing software and functional library for neural network training.
TWGC will be able to solve these two problems by constructing the critical software elements in NVIDIA DGX-1 artificial intelligence supercomputer by using container technology. For research developers, this software stack package will not only be easier to obtain and use, but will also be continuously updated and optimized for achieving maximum performance.
To address hardware compatibility issues, TWGC currently provides research developers with V100 DGX-1 to perform operations that will provide newer, faster GPU computing resources in the future that can be learned in depth from the cloud, and research developers can omit software installation and hardware build time as well.
TWGC will make it easier for research developers to conduct in-depth learning training, experimentation and deployment to accelerate and simplify the development of in-deep learning. Research developers will be able to design more sophisticated neural networks more easily, process more data, and speed up the iteration and launch of products. At present, the latest version of TWGC is 0.5a. The main new features of this version are:
For detailed information, please visit: https://twgc.nchc.org.tw/announce_01.aspx Please review the detailed description of TWGC Version 0.5a Release Notes.