What does that mean?
Edit: I found the full quote in the website of the Jülich Development Center and I guess it makes sense why it was editorialized for the eu-wide website.
“This is a historic milestone. With JUPITER, Europe is reaching the highest level of high-performance computing. JUPITER is also a testimony for Germany's long leadership in HPC. Today, it became the home of the most powerful computer in Europe and the fourth most powerful in the world. From European perspective, JUPITER is a pioneer. It shows that when we combine national vision with European cooperation, we can achieve global excellence.”
Source:
https://www.fz-juelich.de/en/news/archive/press-release/2025...
"The 65th edition of the TOP500 showed that the El Capitan system retains the No. 1 position. With El Capitan, Frontier, and Aurora, there are now 3 Exascale systems leading the TOP500. All three are installed at Department of Energy (DOE) laboratories in the United States."
HPC6 is listed below those, and far less powerful.
I think by some measures XAi already has the largest (200k H100 equivalent) and already constructing the 1 million gpu cluster.
"It is currently being commissioned and has achieved a preliminary HPL value of 793.4 Petaflop/s on a partial system."
Top 500 Rank: 4 (behind El Capitan, Frontier, and Aurora at the DOE)
System Config: JUPITER Booster - BullSequana XH3000, GH Superchip 72C 3GHz, NVIDIA GH200 Superchip, Quad-Rail NVIDIA InfiniBand NDR200, RedHat Enterprise Linux
Manufacturer: EVIDEN https://eviden.com/solutions/high-performance-computing/hpc-...
Location: EuroHPC/FZJ, Germany https://www.fz-juelich.de/en/ias/jsc/jupiter
Cores: 4,801,344
Rmax (PFlop/s) / Rpeak (PFlop/s): 793.40 / 930.00
Power (kW): 13,088
https://www.fz-juelich.de/en/ias/jsc/jupiter/tech
looks like its just a big Nvidia installation. IB, Ceph, K8s, H100s
Nice. How much does this cost?
> The scratch storage is based on 20 IBM Storage Scale 6000 systems utilizing NVMe disk technology, based on the IBM Storage Scale solution. With 29 PB of raw and 21 PB of useable capacity
That's also a lot of waste, isn't it?
RAID gonna RAID.
Sure, the HQ of the core component(Nvidia) is incorporated in US but that too is a multinational effort.
Edit: that might be speculative or a future addition cause they taped out less than 2 months ago https://www.eetimes.eu/sipearl-tapes-out-rhea1-processor-clo...
Which was delayed 2 years. I’m speculating this was supposed to be mostly or exclusively this but they needed a computer now. Or needed to spend the budget now.
Since when are supercomputers anything other than just the ability to afford a lot of hardware?
And so on, but that's just my 2 cents.
