LeaderGPU is granting clients needing infrastructure for AI workloads entry to Nvidia GeForce RTX 6000 Ada graphics playing cards, among the best GPUs on the market, on a rental foundation from the top of November 2023.
The server configuration on supply consists of eight RTX 6000 Ada GPUs with eight models of 48GB GDDR6 RAM, and every one consists of 18,176 CUDA cores. This additionally boasts as much as 960GB/s of reminiscence bandwidth.
CUDA cores, constructed by Nvidia, are much like CPU cores, and are designed to course of a number of calculations on the similar time – and also you’d want a whole bunch of hundreds to successfully undertake probably the most intensive generative AI workloads, with the extra CUDA cores you’ve got entry to, the higher.
Unleashing AI by way of the cloud
These configurations, which will be rented for €0.25 (or 27¢) per minute, additionally embrace two Intel Xeon Gold 6248R CPUs, fitted alongside 384GB RAM, in addition to 2TB NVMe storage capability. They will run on Ubuntu in addition to Home windows Server.
Starting November 30, you possibly can hire this configuration for €5,682.22 ($6,060) monthly, €1,420.55 ($1,515) per week, or €284.11 ($300) per day.
For these searching for to get an hourly price, that is equal to simply €15 ($16), which means that renting as much as 1,000,000 CUDA cores would set you again simply €105 ($110) per hour.
The {hardware} you’d have to run AI workloads is extremely costly to buy, preserve, and run, to not point out being extremely energy-intensive. In consequence, the prospect of leasing the IT infrastructure required – notably the GPUs – is turning into tempting for extra organizations.
That is the place entities like LeaderGPU are available, with the cloud turning into a way more viable route for companies to run their AI workloads.
Japanese firm Third Wave additionally final month introduced a GPU Cloudwhich supplies clients entry to fleets of the perfect GPUs at comparatively inexpensive costs.
