Supermicro to Accelerate AI Storage with AMD and WEKA
Supermicro is launching a full stack optimised storage solution for artificial intelligence and machine learning (AI/ML) data pipelines.
Utilising its extensive AMD and WEKA software, the new solution aims to maximise AI time-to-value, allowing large amounts of raw data to be collected for AI training purposes. This can then be loaded into an organisation’s workflow pipeline.
The goal is to accelerate data pipelines that businesses can leverage for high-performance AI training and inference. Supermicro’s solution is designed to reduce implementation risks, enable organisations to train models faster and quickly use the resulting data for AI inference.
Gaining greater enterprise insights from AI
With AI having been defined by AMD itself as the next era of computing, with the benefits having been seen in multiple key industries such as healthcare and technology. Arguably, its full potential has not yet been realised, as businesses are still in the early stages of developing and harnessing AI models.
Training all types of AI requires billions of parameters and huge amounts of compute capability.
These solutions from Supermicro are designed to ultimately help businesses gain greater data insights from AI. Its multi-rack scale solution from Supermicro is designed to reduce implementation risks, enable organisations to train models faster and quickly use the resulting data for AI inference.
The new storage solution consists of Supermicro Petascale storage servers and InfiniBand and Ethernet Switches, in addition to WEKA’s data platform and Quantum ActiveScale object storage. Using the latest Supermicro systems with WEKA software aims to allow users to see significant increases in the performance of AI applications.
In this vein, the new storage solution for AI training enables customers to maximise their usage of Supermicro’s GPU servers and increase their AI performance as a result.
“With 20 PB per rack of high-performance flash storage driving four application-optimised NVIDIA HGX H100 8-GPU based air-cooled servers or eight NVIDIA HGX H100 8-GPU based liquid-cooled servers, customers can accelerate their AI and ML applications running at rack scale,” says Charles Liang, President and CEO of Supermicro.
Aiding faster enterprise AI expansion
Supermicro’s solution also includes an extensive portfolio of Intel and AMD based storage servers, designed to be available worldwide for customers with ML, generative AI (Gen AI) and other complex compute workloads.
The company is also utilising software from both WEKA and AMD to better train AI. Supermicro is aware that processing unstructured data must be available to the GPU servers with low low latencies and high bandwidth to keep the GPUs productive. During a time of unprecedented high demand for AI, this solution aims to make AI deployment faster and more efficient.
“The high performance and large flash capacity of Supermicro’s All-Flash Petascale Storage Servers perfectly complement WEKA’s AI-native data platform software,” says Jonathan Martin, President at WEKA. “Together, they provide the unparalleled speed, scale, and simplicity demanded by today’s enterprise AI customers.”
AMD’s AI chips in particular have surged in popularity over the past year, with the company’s shares skyrocketing in the midst of immense business demand. The record-high share prices suggest that increasing numbers of investors are only continuing to harness AI.
Multiple high-profile technology companies are planning on buying AMD chips, including Tesla CEO Elon Musk, who has been reported stating that he plans to purchase them as part of a larger investment into computing hardware to better harness AI.
Dell Technologies is also expanding its Gen AI solutions portfolio with AMD, aiming to advance workloads and offer businesses greater choice when harnessing the technology.
******
Make sure you check out the latest edition of Data Centre Magazine and also sign up to our global conference series - Tech & AI LIVE 2024
******
Data Centre Magazine is a BizClik brand
