UCIe consortium unveils 3.0 specification with up to 64 GT/s and improved manageability

The UCIe consortium releases its 3.0 specification, offering chiplet connections at up to 64 GT/s and a raft of new capabilities for custom and high-performance chip designs.

The Universal Chiplet Interconnect Express consortium has officially announced the UCIe 3.0 specification, signaling a significant leap forward in open chiplet interconnect standards. This latest release is designed to fulfill the surging industry need for ever-faster, interoperable links within semiconductor packages. Highlighting the update are new data rate supports for 48 GT/s and 64 GT/s, offering a dramatic performance boost for future chiplet-based systems.

Beyond raw speed, the 3.0 specification integrates several architectural refinements targeted at flexibility and efficiency. One of the standouts is runtime recalibration, a feature engineered to support smarter, real-time adjustments that bolster power efficiency in active chiplet connections. The extended sideband reach broadened by this update enhances support for more versatile, complex multi-chip configurations, thus accommodating evolving semiconductor integration strategies.

The consortium has also prioritized manageability in this release, ushering in options such as early firmware download and priority sideband packet capabilities. These additions are designed to drive prompt system responses and robust overall reliability. Notably, the specification´s manageability features are optional, granting designers and manufacturers the autonomy to implement only the tools required by their specific applications. This modular approach promotes adoption across a diverse range of products by balancing advanced capability with design flexibility and silicon efficiency.

72

Impact Score

Saudi Artificial Intelligence startup launches Arabic LLM

Misraj Artificial Intelligence unveiled Kawn, an Arabic large language model, at AWS re:Invent and launched Workforces, a platform for creating and managing Artificial Intelligence agents for enterprises and public institutions.

Introducing Mistral 3: open artificial intelligence models

Mistral 3 is a family of open, multimodal and multilingual Artificial Intelligence models that includes three Ministral edge models and a sparse Mistral Large 3 trained with 41B active and 675B total parameters, released under the Apache 2.0 license.

NVIDIA and Mistral Artificial Intelligence partner to accelerate new family of open models

NVIDIA and Mistral Artificial Intelligence announced a partnership to optimize the Mistral 3 family of open-source multilingual, multimodal models across NVIDIA supercomputing and edge platforms. The collaboration highlights Mistral Large 3, a mixture-of-experts model designed to improve efficiency and accuracy for enterprise artificial intelligence deployments starting Tuesday, Dec. 2.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.