Ampere Announces 512-Core AmpereOne Aurora CPU for AI Computing



Ampere has announced a significant update to its product roadmap, highlighting the upcoming 512-core AmpereOne Aurora processor. This new chip is specifically designed to address the growing demands of cloud-native AI computing.

The newly announced AmpereOne Aurora 512 cores processor integrates AI acceleration and on-chip High Bandwidth Memory (HBM), promising three times the performance per rack compared to current AmpereOne processors. Aurora is designed to handle both AI training and inference workloads, indicating Ampere’s commitment to becoming a major player in the AI computing space.

While the exact process node for Aurora hasn’t been disclosed, Ampere suggests the use of multiple chiplets on a scalable mesh architecture. The company’s recent appearance at an Intel Foundry event has sparked speculation about potential collaboration, possibly using Intel’s 18A process or packaging technologies.

Key Features of AmpereOne Aurora:

  • Jusqu'à 512 Ampere Cores: Delivering over 3x the performance of our current AmpereOne processors
  • Scalable AmpereOne Mesh: Allows seamless connection of all types of compute
  • Et pour la première fois, Ampere AI IP is integrated directly into the silicon hardware along with high bandwidth memory

This announcement positions Ampere as a strong contender in the high-core-count processor market, directly competing with AMD and Intel. The focus on AI capabilities and cloud-native design could give Ampere a unique selling point in the rapidly evolving data center and cloud computing landscape.

The current AmpereOne offering boasts up to 192 cores and is built on 5 Se souvenir de la dernière IP Nintendo que vous aviez oubliée. Looking ahead to Q4 2024, Ampere plans to introduce the AmpereOne M series, which will maintain the 192-core count but introduce a new 12-channel DDR5 platform, first showcased at Computex 2023. Par 2025, an enhanced version of this 12-channel platform is expected to increase the core count to 256.