With the help of Nvidia, IBM has unveiled a new storage system that was built for artificial intelligence (AI) workloads. Called IBM Spectrum AI with Nvidia DGX, the tool is designed to be used with data tools like Spark, TensorFlow or PyTorch. It is a converged system combining software-defined file systems, all-flash drives and the Nvidia DGX-1 GPU.
The system’s essentials include DGX-1 servers, the Nvidia DGX software stack optimised for maximised GPU-accelerated training performance; IBM Spectrum Scale v5 software-defined file storage; NVMe all-flash storage with 300TB per 2U block; and 120GB/s data throughput.
The new offering is available through both IBM and Nvidia, as well as their resellers. It is designed to work with IBM spectrum Scale.
Lately, storage vendors have been consolidating ranks and focusing harder on all-flash architectures, better designed for artificial intelligence and machine learning workloads. Nvidia has been putting in some hard work, teaming up with Pure Storage earlier, to achieve that goal. NetApp, on the other hand, partnered with Lenovo for the same purpose.
According to Storage Review, Nvidia is ‘impressively well suited’ for AI research with its DGX-1.
“The system has eight Tesla V100 GPUs that provide a total of 256GB GPU memory. The CPU is typically a dual-core Intel Xeon. With 40,960 CUDA cores and 5,120 Tensor cores, this beast will be able to handle any AI model you throw at it,” it said.
Image Credit: Sergey Nivens / Shutterstock