# Intel Shares First Benchmarks of Its Next-Gen Crescent Island “D270” GPU, Claiming a 5.9x Throughput Over RTX PRO 6000 As Xe3P Goes Big On FP64

> **Open Intelligence Dossier** · First detected: 2026-11-16 20:00 UTC · Category: Technology

## Executive Summary
Intel unveils first benchmarks for its Crescent Island D270 GPU, claiming 5.9x throughput over NVIDIA RTX PRO 6000 in financial modeling at half the power.

## Intelligence Brief
Intel has released the first performance data for its next-generation Intel GPU D270, formerly code-named Crescent Island, positioning the accelerator as a unified solution for high-performance computing and AI inference. Built on the Xe 3P architecture, the D270 features 32 Xe cores and 256 Xe Matrix Extensions (XMX) engines capable of 64 FP64 fused multiply-add operations per core, delivering native double-precision compute alongside lower-precision inference throughput. The Intel-branded card packs 160 GB of LPDDR5X memory with 1.5 TB/s bandwidth in a 350-watt, air-cooled PCIe form factor, while partner designs can scale to 480 GB.


According to Intel, real-world application benchmarks on pre-production silicon show the D270 delivered up to 5.9× higher application throughput than the NVIDIA RTX PRO 6000 Blackwell Server Edition in options-pricing financial modeling, while operating at 350W compared to the NVIDIA GPU&#039;s 600W. For seismic exploration workloads using FP32 compute, Intel says the combination of high memory capacity and cache-aware stencil optimizations keeps large wavefield arrays resident on the card, with a live demonstration running at Supercomputing 2026 in Chicago.


Intel also disclosed AI inference metrics: the D270 achieves up to 29% lower latency than the RTX PRO 6000 on Drex 2.0 and 1.12x throughput on Qwen 3 30B – A3B, both at 350W. The company is demonstrating a disaggregated heterogeneous inference architecture at SC26, pairing the D270 for compute-intensive prefill stages with SambaNova SN50 Reconfigurable Dataflow Units for memory-bound decode stages, orchestrated by Intel Xeon 6 processors.


On software, Intel emphasizes an open SYCL-based stack designed to protect existing code investments. The company says porting the Veridien seismic workload to D270 required minimal effort, with 90% of kernels performing well without Intel-specific optimizations and complex kernels reaching up to 70% of theoretical peak performance on the first unoptimized port. The seismic imaging mini-application was ported in less than a month with numerical accuracy and performance validated. Intel states the architecture is turnkey for deploying agentic AI pipelines without custom orchestration software.

## Multi-Source Evidence Table
| Source Outlet | Headline | Verification URL |
|---|---|---|
| igor´sLAB | Intel Xeon and GPU D-Series: HPC and AI in a Unified System | [Source Link](https://news.google.com/rss/articles/CBMihgFBVV95cUxOaVNpNlZwaG5fSGpmekVDR0w2dlRySnFvSDNPU2NPbXh0YkdDWjlqTHd5LWM0YnhWWXdKNTJmLXVNTnNrMnJhR08teE44SjZSV0dLbnp6MWNoa0JfanR4eUxSbDBQWHpFa0gwME1SVjNOLVV6LURxZTVWd2JmSndROEpRcXBuQQ?oc=5) |
| TechPowerUp | First Photos of Intel's 256-Core Xeon 7 "Diamond Rapids" Package Surface | [Source Link](https://news.google.com/rss/articles/CBMipwFBVV95cUxQT20yN1JHLUJzT0g0MzNhUV93U2Vad2lpdTZDX0hrZE0wSE5SVGNXVzlLWG0zTjAyWTFKSVk0WXlYMWd2RlA3NUdwSmRCNTRQWDVzbUNtcU5CNzBtTzJoNk9iNVZCbTZkMWlRcXpGTmE2YXVmOENISW9TVEplYlFzU1MwbzRuMDhySzdXRlVNWlZ4SkVYU3haN3Q4Vy1VWmhMYldFS24xONIBrAFBVV95cUxQc0VrQ3VHUmJyMmFKWWYyMmQyNjJyTy1CRV9yUGpUeVpIRTRRT0g0T012YkFkRmo0bHpvWUJxbEdFSy1MdElRbU5qVW9zLWt1UUxLRGlhMEprSGpRb3BQUkpKZ3pKTG5SM2xtcFVzQjlFa0dFWXVyNGJ4X1cyX25aV2NNWmY4Zm8wTnk5QjFySUhoZ3Znb1B6WUdpanFEd0xrdkRRNjl5YTNJbzF) |
| Phoronix | Intel Talks Up GPU D270 Performance & Open-Source Software Support | [Source Link](https://news.google.com/rss/articles/CBMibEFVX3lxTE1zNlZ0TU9rMkw2SG05TGk5UWR2Z0NUOVlLR3BmX2lGYnlpMWNKd29Ram9YRVpMYnhGWE8yREtOR3hyWUdVSklrNUpCMmUwZkhJRl9FN1g0SVdscDEyRHN2Q0RjbE9GaXRkQU9WSA?oc=5) |
| Intel | The case for Intel GPU D270 in high-performance computing | [Source Link](https://news.google.com/rss/articles/CBMivgFBVV95cUxOQ2hDZkU4Q0VHYWF4SmVmZC1qUk40d2hTNFQwLWJhclpJR05GMm42MGpmeUNrX0J6M0taT2hnYTJkZmczNkdoUURWYUs5TXNKZnk5d3ZKbWdPLU1fVXpJUHpGMk04Zmx6TldDWldXaGN3Uk5hRVRGSVNmcFNXYU5TV3Q2dmJpVkg4Q3pMVGpiOHA3QzE3c1FlMUxOSnlQdUVIOEZHY0JkVlNnUzZsRENsSWE5WmJQZEtuV2JWaEhR?oc=5) |
| Wccftech | Intel Shares First Benchmarks of Its Next-Gen Crescent Island “D270” GPU, Claiming a 5.9x Throughput Over RTX PRO 6000 As Xe3P Goes Big On FP64 | [Source Link](https://news.google.com/rss/articles/CBMiqgFBVV95cUxQeDZ5aWpnVFo2dFVFYk55MUFkXzBsY0RuZXdqSG1QN1laS2dhRzBiWmZJWHcyR1R0UzBlSTJJbkc5T2FyZjZXYm93SnlQLWFPTFNndHFYRXg5UVprZXRvMzItN1FTTnZqMTdQTEZoY1h3YkZzUlF5ZGE0X0o4aTRxVVRxaDI0b3dhZmgyOGxvRTd4UDNzQUdoWC1MMFFucEpUUVdsSExkc2FqZ9IBrwFBVV95cUxQdnNVNmdXMUlsbWlSRnFRWEZ0VGVSbV8xU0Z2ZERVZ0RUVG1qZDhqLTVYVWxjTk15dnFSTzZ2cEdNUGRiUks1Tlc2TGFiN0MxQndyZUpYR0V0MlF5UkYxMXdtdC00TXh1YWUzY2F0UkUyZFhMOC1tYXpaQzYzOFEtQzF1S1c1N09ydzE4RXJweGMzbFRUMmg4ek1nMWZCOG9KU0VrMjIxVHR) |

---
*Canonical Source: https://pulse.byoaiapps.com/trend/2026-11-16/intel-shares-first-benchmarks-of-its-next-gen-crescent-island-d270-gpu-claiming*
