“With the flexibility to deploy a staggering 100 billion parameter mannequin on a single-width, energy-efficient 150W card, it units a brand new commonplace within the business.”~ Andrew Feldman.
Qualcomm has launched the Qualcomm Cloud AI 100 Extremely, the most recent addition to its lineup of cloud synthetic intelligence (AI) inference playing cards, particularly designed to deal with generative AI and enormous language fashions (LLMs).
This new card is predicted to have as much as 4 occasions the efficiency of its predecessor utilizing Qualcomm’s superior AI cores. It effectively helps fashions with 100 billion parameters on a single 150-watt card, and fashions with 175 billion parameters with two playing cards. Bigger fashions could be managed by combining a number of Qualcomm Cloud AI 100 Extremely playing cards utilizing the Qualcomm AI Stack and Cloud AI SDK.
As a programmable AI accelerator, the Qualcomm Cloud AI 100 Extremely adapts to current AI developments and knowledge codecs. It seamlessly integrates with the Qualcomm AI Stack, facilitating simple mannequin porting and optimization.
Hewlett Packard Enterprise (HPE) plans to include the Qualcomm Cloud AI 100 Extremely in its HPE ProLiant DL380a Gen 11 server, which is designed for generative AI workloads, together with pure language processing (NLP).
“To unlock worth from generative AI, enterprises would require an AI-native structure that’s purpose-built to assist any a part of their journey, together with inferencing. In collaboration with Qualcomm, we stay up for providing our clients a compute resolution that’s optimized for inference and offers the efficiency and energy effectivity essential to deploy and speed up AI inference at-scale.” stated Justin Hotard, govt vice chairman and common supervisor of high-performance computing, AI and labs at HPE.
Andrew Feldman, chief govt officer of Cerebras, praised the Qualcomm Cloud AI 100 Extremely for its capabilities, power effectivity, and efficiency, significantly together with Cerebras’ coaching expertise.
In numerous purposes, the Qualcomm Cloud AI 100 Extremely delivers two to 5 occasions the efficiency per whole value of possession (TCO) greenback over rivals in generative AI, LLMs, NLP, and pc imaginative and prescient workloads. It offers a stability of price-performance, power-efficiency, scalability, and safety, making it a great resolution for organizations aiming to leverage AI for transformative outcomes whereas supporting sustainability targets.
