#mi450 — Public Fediverse posts

Attention Decode on AMD MI450 GPUs: A Gluon Kernel Optimization Guide

https://rocm.blogs.amd.com/software-tools-optimization/gluon-attention-decode-mi450/README.html

Comments: https://news.ycombinator.com/item?id=49077646

#HackerNews #AttentionDecode #AMD #MI450 #GluonKernel #Optimization #GPUOptimization

#AMD announced a strategic partnership with #Anthropic, investing up to $5 billion in the AI company. As part of the deal, Anthropic will deploy 2 gigawatts of AMD Instinct #MI450 Series GPUs in AMD #Helios solutions, with the first gigawatt deployed in the first half of next year. This #partnership is the latest in a series of #infrastructure deals for Anthropic as it scales its computing capacity to meet growing demand. https://www.cnbc.com/2026/07/22/amd-anthropic-ai-chip-investment.html?eicker.news #tech #media #news

#AMD Says “#Helios” Racks And #MI400 Series #GPU On Track For 2H 2026
AMD has been working on the Open Rack Wide v3 specification, a double-wide rack that AMD calls “Helios” and that is the platform for delivering #rackscale compute based on the “Altair” MI400 series GPU accelerators. This series includes the #MI450, #MI430X, and the #MI455X for the Helios rackscale systems, which will have 64, 72, or 128 GPUs per system, and possibly MI440X for eight-way system nodes.
https://www.nextplatform.com/2026/02/23/amd-says-helios-racks-and-mi400-series-gpus-on-track-for-2h-2026/

#Oracle Cloud will deploy 50,000 #AMD Instinct #MI450 #AIchips starting in 2026, signalling increased #competition for #Nvidia’s market-leading #GPUs. This move follows #OpenAI’s recent deals with both AMD and Oracle, highlighting the growing demand for diverse AI chip suppliers. https://www.cnbc.com/2025/10/14/oracle-cloud-to-deploy-50000-amd-ai-chips-as-alternative-to-nvidia.html?eicker.news #tech #media #news

#Intel Announces "#CrescentIsland" #Inference-Optimized #Xe3P #GPU With 160GB vRAM, but it's not shipping until at least H2'2026
Given the timing this will be going up against the AMD Instinct #MI450 series and NVIDIA #VeraRubin.
This new graphics card is built on next-gen Xe3P Celestial micro-architecture. Xe3P will be optimized around performance-per-Watt and Crescent Island will feature 160GB of LPDDR5x memory to allow plenty of space for #largelanguagemodels (#LLM).
https://www.phoronix.com/review/intel-crescent-island

#AMD’s #software efforts, particularly its #collaboration with #OpenAI, have been instrumental in securing a multibillion-dollar deal to provide chips for OpenAI’s AI products. This partnership, which includes OpenAI’s influence on AMD’s #MI450 #chipdesign and software compatibility, positions AMD as a stronger competitor to #Nvidia in the #AIhardware market. https://www.forbes.com/sites/richardnieva/2025/10/09/amd-openai-chip-software/?eicker.news #tech #media #news

KI-Beschleuniger MI450: AMD geht mit 2 nm „All-In“, Nvidia Rubin nutzt noch 3 nm (Update) https://www.computerbase.de/news/grafikkarten/ki-beschleuniger-mi450-amd-geht-mit-2-nm-all-in-nvidia-rubin-nutzt-noch-3-nm.94602/#update-2025-10-09T11:18 #AMD #MI450 #2nm

KI-Beschleuniger MI450: AMD geht mit 2 nm „All-In“, Nvidia Rubin nutzt noch 3 nm https://www.computerbase.de/news/grafikkarten/ki-beschleuniger-mi450-amd-geht-mit-2-nm-all-in-nvidia-rubin-nutzt-noch-3-nm.94602/ #AMD #MI450 #2nm

AMD and OpenAI team up for massive 6 gigawatt GPU partnership

https://web.brid.gy/r/https://nerds.xyz/2025/10/amd-openai/

#AMD and #OpenAI have signed a multi-year chip supply deal worth tens of billions. AMD will supply #6gigawatts of #computecapacity to OpenAI, starting with the Instinct #MI450 #GPU in 2026. OpenAI will also have the option to purchase up to 10% of #AMDstock, with vesting tied to AMD’s stock price milestones. https://techcrunch.com/2025/10/06/amd-to-supply-6gw-of-compute-capacity-to-openai-in-chip-deal-worth-tens-of-billions/?AIagents.at #AIagent #AI #ML #NLP #LLM #GenAI

NextGen #MI450 #AI Lineup, #AMD Says There Will Be ‘No Excuses, No Hesitation’ in Choosing Team Red Over NVIDIA In AI Workloads
In terms of what to expect with #MI400 lineup, you are looking at integration of #HBM4, up to 432 GB, this will bring massive bandwidth improvements. AMD plans to aggressively expand #rackscale options with MI400 series as well, introducing highly anticipated #Helios rack as well, which is claimed to rival Vera Rubin's top configuration on-paper.
https://wccftech.com/with-the-mi450-amd-says-there-will-be-no-excuses-no-hesitation-in-choosing-team-red-over-nvidia/

AMD wittert den Milan-Moment: MI450 soll „bei jeder Art von KI-Workload“ führen https://www.computerbase.de/news/grafikkarten/amd-wittert-den-milan-moment-mi450-soll-bei-jeder-art-von-ki-workload-fuehren.94276/ #AMD #AI #MI450