Mastodawn

美能源部攜手超微建10億美元AI超級電腦，應用於核能、核融合、治癌與國安等實驗

中央通訊社 2025-10-28 10:30:00 CST美國能源部與超微（AMD）合作，將建置兩座 AI 超級電腦，以加速核能、癌症治療及國家安全等領域的科學研究。第一座系統「Lux」預計於六個月內上線。
https://www.thenewslens.com/article/260364
#Stephen Streiffer #AI晶片 #藥物研發 #Discovery #超微 #MI355X #橡樹嶺國家實驗室 #能源部 #Chris Wright #AI超級電腦 #AMD #核融合 #國防 #蘇姿丰 #核武庫 #科技 #核能 #MI430 #甲骨文 #慧與科技 #Lux #國家安全 #美國

美能源部攜手超微建10億美元AI超級電腦，應用於核能、核融合、治癌與國安等實驗 - TNL The News Lens 關鍵評論網

美國能源部與超微（AMD）合作，將建置兩座 AI 超級電腦，以加速核能、癌症治療及國家安全等領域的科學研究。第一座系統「Lux」預計於六個月內上線。

TNL The News Lens 關鍵評論網

Benjamin Carr, Ph.D. 👨🏻‍💻🧬Oct 27

#US #DepartmentOfEnergy forms $1 billion #supercomputer and #AI partnership with #AMD
US is building two machines to ensure country has enough #supercomputers to run increasingly complex experiments require harnessing enormous amounts of data-crunching capability.
#Lux to be constructed and come online within next six months will be based around AMD's #MI355X #artificialintelligence chips
#Discovery will be based around AMD's #MI430 series of chips that are tuned for #hpc
https://www.reuters.com/business/energy/us-department-energy-forms-1-billion-supercomputer-ai-partnership-with-amd-2025-10-27/

Dr. Moritz Lehmann Oct 17

8x AMD Instinct #MI355X (288GB @8TB/s) take back the lead over 8x Nvidia #B200 (180GB @8TB/s) in #FluidX3D #CFD, achieving 362k MLUPs/s (vs. 219k MLUPs/s). Thanks to Jon Stevens from Hot Aisle to run the benchmarks! 🖖😊

In single-GPU, both perform about the same, but in 8x #GPU config, MI355X is 65% faster. The difference comes from PCIe bandwidth - MI355X does 55GB/s, B200 only 14GB/s. #Nvidia leaves a lot of perf on the table by not exposing #NVLink P2P to #OpenCL.

https://github.com/ProjectPhysX/FluidX3D

Benjamin Carr, Ph.D. 👨🏻‍💻🧬Sep 18

#AMD #ROCm 7.0 Officially Released With Many Significant Improvements
Key highlights of #ROCm7.0 include:
- AMD Instinct #MI350X and Instinct #MI355X officially supported
- #Ubuntu 24.04.3 LTS and #RockyLinux 9 with Linux 5.14 officially supported
- ROCm 7.0 supports #KVM Passthrough for MI350X and MI355X GPUs
- ROCm 7.0 supports #PyTorch 2.7, #TensorFlow 2.19.1
- Official support for Llama.cpp.
- AMD #GPU kernel driver code is now distributed separately from ROCm stack.
https://www.phoronix.com/news/AMD-ROCm-7.0-Released

AMD ROCm 7.0 Officially Released With Many Significant Improvements

Overnight the AMD ROCm 7.0 release tags began appearing within the public Git repositories

Benjamin Carr, Ph.D. 👨🏻‍💻🧬Jun 11, 2025

#AMD's Instinct #MI355X accelerator will consume 1,400 watts. #CDNA4 challenges #BlackwellUltra.
AMD's #MI350X-series #GPU are based on #CDNA 4 architecture that introduces support for FP4 and FP6 precision formats alongside FP8 and FP16.
Both SKUs will come with 288GB HBM3E memory that will offer up to 8 TB/s of bandwidth, but MI350X will offer maximum FP4/FP6 performance of 18.45 PFLOPS, whereas the MI355X is said to push the maximum FP4/FP6 performance to 20.1 PFLOPS.
https://www.tomshardware.com/pc-components/gpus/amds-instinct-mi355x-accelerator-will-consume-1-400-watts

AMD's Instinct MI355X accelerator will consume 1,400 watts

CDNA 4 challenges Blackwell Ultra.

Tom's Hardware