artificial intelligenceAI-driven AI-Workload automation provider for GPUs and AI accelerators has partnered with AMD to enhance AI infrastructure.
The long-term strategic collaboration aims to improve AI’s inference and training workload management and AMD Instinct GPU performance, providing customers with scalable and cost-effective solutions for deploying AI applications.
As AI adoption accelerates, organizations are struggling to address resource allocation, performance bottlenecks and complex GPU management.
By combining RAPT’s intelligent workload platform with AMD Instinct MI300X, MI325X and the upcoming MI350 series GPUs, the collaboration provides scalable, high-performance and cost-effective solutions that enable customers to maximize AI reasoning and training efficiency across square and multi-person constructions.
More effective solutions
Charlie Leeming, CEO of RAPT AI, said in a press conference: “The AI model we see today is so big, and most importantly, dynamic and unpredictable. The older optimization tools are totally inappropriate. We observe these dynamics. We observe these dynamics. Businesses spend a lot of money. Hiring a new set of talent in these areas. It is the reward.
Leeming said Anil Ravindranath, chief technology officer of Rapt AI, saw the solution. This involves deploying monitors to enable observations of infrastructure.
“We feel like we have the right solution at the right time. We stood out from invisibility last fall. We are growing in Fortune 100 companies. Both companies run that code in cloud service providers.”
He said: “We do have strategic partners, but the conversation with AMD is going very well. They are building huge GPUs, AI accelerators. We are known for putting the maximum workload on the GPU on the maximum workload. Now, inference is stepping down. Now, it is in the stage. Now, the AI workload is exploded. Their data scientists need to explode. They need to work, they need to work. They need to work. They need to work. For inefficient solutions, customers are less than 30% efficient.
He said improvements that could be done in three minutes, which could take nine hours. Ravindranath said at a press conference that the RAPT AI platform can run 10 times the model run capability on the same AI computing spending level, saving up to 90% of cost, while zero humans and zero zeros, and no code changes. For productivity, this means no longer waiting for calculations and spending time tweaking the infrastructure.
Other technologies have been around for some time, but have not been cut, Laiming said. Run AI, competitors, overlap in competitive ways. He said his company observes in minutes instead of hours and then optimizes infrastructure. Ravindranath said Run AI is more like a scheduler, but Rapt AI positions itself as an unpredictable result and processes it.
“We run the model and figure it out, that’s a huge benefit for the inference workload. It should run automatically,” Ravindranath said.
Benefits: Reduce costs and better GPU utilization
AMD Instinct GPUs, with its industry-leading memory capabilities, said the two companies
RAPT’s intelligent resource optimization helps ensure maximum GPU utilization for AI workloads and helps reduce total cost of ownership (TCO).
Rapt’s platform simplifies GPU management, eliminating the need for data scientists to spend valuable time trial infrastructure configurations. By optimizing resource allocation for their specific workloads, it enables them to focus on innovation rather than infrastructure. It seamlessly supports a variety of GPU environments (whether in the cloud, whether in the cloud or in the cloud) with one instance, helping to ensure maximum infrastructure flexibility.
The joint solution can intelligently optimize job density and resource allocation on AMD Instinct GPUs, providing better inference performance and scalability for production AI deployments. RAPT’s automatic scaling capability further helps ensure efficient utilization of resources based on demand, thereby reducing latency and maximizing cost efficiency.
RAPT’s platform with AMD Instinct GPUs are available out of the box, helping to ensure immediate performance benefits. The ongoing collaboration between RAPT and AMD will drive further optimization in exciting areas such as GPU planning, memory utilization, etc., help ensure customers are equipped with the AI Ready AI infrastructure of the future.
“At AMD, we are committed to delivering high-performance, scalable AI solutions that empower organizations to unlock the full potential of their AI workload,” said Negin Oliver, vice president of business development at AMD’s Data Center GPU business, in a statement. “Our collaboration with RAPT AI combines cutting-edge capabilities of AMD Instinct GPUs with intelligent workload automation from Rapt, enabling customers to achieve greater efficiency, flexibility and cost savings in their AI infrastructure.”
Source link
