Intel's Crescent Island AI accelerator officially named D270: 480GB memory, 1.3 PFLOPS FP8
scaling01 · x · 2026-10-13
Intel has given its AI accelerator, announced in October 2025 under the codename Crescent Island, an official product name: GPU D-Series. The D270 card features 480GB of LPDDR5X memory with 1.5TB/s bandwidth, 1.3 PFLOPS of FP8 compute, and a 350W TDP. After detailing specs at Computex 2026 and Hot-Chips, Intel further concretized the product at the OCP conference. The huge-memory, modest-bandwidth design points to large-context inference workloads rather than training.
More from Infra
- NVIDIA RTX Spark launches Oct 16 with 128GB unified memory and 1 petaflop of local AI compute — nordicinst · 2026-10-13
- A 1GW data center now costs over $30B to build, with chips and servers eating $17B+ — asusarla · 2026-10-13
- Yoniq Compute repo ships inference recipes validated on 1x-8x RTX Pro 6000 with SGLang and vLLM — TheZachMueller · 2026-10-13
- RL post-training compute accounting: one weight update per episode blurs the train/inference line — davidmanheim · 2026-10-13
- Ex-Apple engineer eyeing Surface with NVIDIA RTX Spark: 128GB + CUDA in a backpack — LinusEkenstam · 2026-10-13
- Local 125B MoE on RTX 5090 finishes Three.js horse challenge in 23 minutes — Beneficial-Ad-8127 · 2026-10-13