Intel renames Crescent Island as GPU D-Series and details datacenter specs
Intel has started sampling the new line, giving buyers a clearer view of its performance, power and server fit.
Intel has formally branded its Crescent Island AI accelerator line as the Intel GPU D-Series and disclosed key specs that matter for datacenter buyers: 1.3 PFLOPS of FP8 peak performance and 1.5 TB/s of memory bandwidth. The chips use Intel’s Xe3P-based design, ship in an air-cooled PCIe form factor, draw 350W, and support up to 480GB of LPDDR5X memory. Intel says the D-Series is aimed at traditional rack servers with as many as eight GPUs, targeting inference workloads for models above one trillion parameters as well as long-context and concurrent inference. The company says customer sampling is already underway, but it has not given a wider launch date.
Why it matters
The rebrand turns a previously named accelerator into a more concrete product family with published limits on performance, bandwidth, power and memory. For datacenter buyers, that makes it easier to judge whether the cards fit air-cooled rack servers that scale to eight GPUs. Sampling already underway also means the line is moving beyond announcement stage, even though Intel has not given a wider launch date.
Keep or strike?
Does this story matter, or is it hype? Mark it before you see what everyone else did.
Sources
- Tom's Hardware