Intel has officially introduced its new GPU D-Series, which utilizes the Crescent Island AI accelerator. This chip is the first to be built on the Xe3P architecture, designed specifically for massively parallel computing rather than traditional graphics tasks.
The D-Series products feature an air-cooled PCIe form factor with a TDP of 350W and support for up to 480GB of LPDDR5X memory. Intel reports that the hardware delivers 1.3 peak PFLOPS of FP8 compute performance and 1.5 TB/s of memory bandwidth.
These accelerators are intended for deployment in traditional air-cooled rack servers, supporting configurations of up to eight GPUs. Intel states that this setup can power inference for models with more than one trillion parameters, targeting large language models and latency-sensitive services.
To support these deployments, Intel is leveraging its common Xe software foundation. The company has confirmed that customer sampling of the GPU D-Series is currently underway, with ongoing collaborations involving various OEMs, ODMs, and AI-native companies.
