GMO GPU Cloud, a high-performance GPU cloud service provided by GMO Internet, Inc. a GMO Internet Group company, has integrated NVIDIA Multi-Instance GPU (MIG) technology.
With the provision of this function, customers of the “Dedicated Plan” of “GMO GPU Cloud” can select an instance size appropriate for the scale and nature of the job they want to run, and divide one NVIDIA GPU into up to seven independent GPUs at no additional charge. This is expected to lead to the parallel execution of different workloads, optimization of resource usage efficiency, and improved cost performance.
“GMO GPU Cloud” is equipped with NVIDIA H200 GPUs, and is ranked 37th in the world and 6th in Japan in the “TOP500” supercomputer performance ranking. It is also ranked 1st among domestic commercial cloud services, proving its high processing performance.
Also Read: NHN Techorus launches new service “Forwardy” based on the concept of a rental server using AWS
On the other hand, the required environments are becoming more diverse, such as cases where not all jobs necessarily require advanced computing resources depending on the scale and characteristics of the workload, or cases where a high turnover rate is required by running multiple jobs in parallel.
Therefore, we have decided to integrate and start offering NVIDIA MIG as a function that takes advantage of the advantages of the job scheduler to enable customers to use computing resources more flexibly and efficiently.
SOURCE: PRTimes
