Web Hosting

Tachyum Licenses Its TPU Core IP: Elevating Trained AI to the Data Center

Tachyum Stand

Tachyum enhances the unique value proposition of its Tachyum Prodigy by divesting the intellectual property of its Tachyum TPU (Tachyum Processing Unit) as a licensable core. This would allow developers to take full advantage of artificial intelligence (AI) which was taught in data centers during the development of IoT and edge devices.

The Tachyum Prodigy is a “universal processor,” meaning it combines many types of AI on a single chip, including general-purpose processors, high-performance computing (HPC), AI, learning deep AI, explainable AI, bio-AI and other types of AI. AI.

Tachyum is attempting to extend its proprietary Tachyum AI data type outside the confines of the data center by making its globally recognized and trademarked intellectual property available to third-party developers. The move is a direct response to the explosive expansion of the AI ​​chipset market for edge inference.

TPU inference and Generative AI/ML The IP architecture has a number of key features, some of which are transactional and accurate architectural simulators, tool and compiler support, and hardware-licensed intellectual property, which includes RTL in Verilog, UVM Testbench, and synthesis restrictions.

Tachyum now has 4 billion people working on AI training, with an additional 2 billion working under the proprietary Tachyum AI (TAI) data type, which will be revealed later this year.

AI/ML, HPC and Cloud

“Inference and generative AI are coming to almost every consumer product,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “We believe TPU licensing is a key avenue for Tachyum to proliferate our cutting-edge AI in this market for models trained on Tachyum’s Prodigy Universal Processor chip. Inference and generative AI are coming to almost every consumer product.

“This would represent a significant commercial advantage not only for Tachyum but also for all suppliers who respect this brand and ensure that they correctly authorize its use as part of their products. Indeed, Tachyum is the sole owner of the TPU brand within the AI ​​industry.

Data center servers powered by Tachyum Prodigy would be able to move “effortlessly and dynamically” between IT domains (such as AI/ML, HPC and cloud) on a single architecture since Prodigy is a Universal processor that provides utility for all workloads. .

Tachyum Prodigy significantly reduces CAPEX and OPEX while delivering unmatched levels of performance, power and economics for data centers. It does this by removing the need for expensive hardware dedicated specifically to AI and significantly increasing server utilization. To deliver “up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x the performance of the highest-performing GPUs for high-performance computing (HPC), and 6x the performance of for artificial intelligence applications. “, Tachyum prodigy integrates 192 custom-designed, high-performance 64-bit compute cores.

Source link

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button