Intel has officially released its new GPU D-Series chips, formerly identified by the codename Crescent Island.
Utilizing the Xe3P architecture, these chips are purpose-built for AI workloads, offering a theoretical peak performance of 1.3 PFLOPS in FP8 calculations.
Technical highlights include 1.5 TB/s memory bandwidth, support for up to 480GB of LPDDR5X memory, and a 350W TDP, provided in a PCIe form factor.
Although their name includes the term commonly found in graphics cards, these components are not intended for the average home computer. They are specifically designed to handle the complex and parallel computing tasks required by artificial intelligence applications.
According to official data, the new chips offer the following features:
- Speed: Maximum performance of 1,3 PFLOPS for specific types of calculations.
- Memory: Support for up to 480GB of LPDDR5X memory, with a data transfer rate of 1,5 TB/s.
- Consumption: The power required for their operation is set at 350W.
The products will be available in PCIe form factor and will be based on air cooling systems.
Top comments (0)