IRON MOUNTAIN DATA CENTERS
AI ’ S APPETITE IN ACTION
IMDC provides the infrastructure for many High-Power Compute ( HPC ) configurations running generative AI , and has developed specialist facilities that meet their needs . High-density power , modular architecture , high-bandwidth training – input – and inference – output – connectivity and advanced cooling are all critical factors for customers .
Healthcare One of IMDC ’ s healthcare customers has developed a supercomputer for AI-driven imaging apps . While the total consumption is not massive and training cycles are much shorter than for LLMs , the data growth curve of this supercomputer since it was first built in 2018 has been steep . It began with just 10,000- 50,000 images and achieved 85 % accuracy . Now it uses up to half a billion images with accuracy of 95 % and runs 50,000 deep learning training experiments per month . Despite the compactness and efficiency of the GPUs , a few racks in the data centre have become a full module of 60 racks with
26 petabytes of processing power storing close to two billion datasets . In less than two years a petaflop of processing power will be needed .
Research The Computational Research Accelerator department at Arizona State University was running out of network ports , space and power for ‘ Agave ’, its supercomputer , so they built a new supercomputer called ‘ Sol ’ in 2022 in one of IMDC ’ s Phoenix data centres . Sol is a Dellbuilt system spanning 178 nodes . It uses AMD Epyc 7713 CPUs , consisting of around 18,000 cores , with the bulk of the nodes carrying 512GB of memory and five large-memory nodes equipped with 2TB . It has 56 GPU nodes with quadruple Nvidia A100 , 80GB , GPUs each and four nodes with triple Nvidia A30 , 24GB , GPUs . The system is networked with Nvidia ’ s 200GB / s HDR InfiniBand and supported by four Petabytes of Dell BeeGFS scratch storage . The R & D potential of Sol is extremely exciting , and a steep physical growth curve is anticipated .