The world's most powerful supercomputers

But, can it show me "Hot single moms" in my area?
 
Last edited:
What isn't immediately obvious is that eight or so years ago, South Africa launched a 1PF system at 121 in the world. The top two currently with Rmax 1000x faster than that.
 
This list also excludes a lot of private supercomputers, which many readers may not be aware of. The industry is really huge right now.
 
Top500 is voluntary, but for most govt. organisations it's a bit of muscle flex and private don't even care about it. For instance, FinTech companies (e.g. XTX) don't bother about the Top500 because it's (a) giving away competitive secrets and (b) irrelevant to them.

In the community, it's still really useful, and grant funded systems reward the funders with some prestige. South Africa is launching a system that will be back on the Top500 and has (for right or wrong reasons) target an Rmax goal.

Green500 shouldn't be ignored either, since the top Green500 is also near the top of the Top500 which is a very significant achievement. Efficiency and performance.

Back in the years of Ferrari's dominance in F1, they were recruiting a ton of Physicists and CFD experts because they had a Top500 system (even though it wasn't on the Top500, it was way up there in performance if they'd ever bothered to benchmark it).
 
“Oak Ridge National Library”. Uhm who wrote this article? ChatGPT? Its the Oak Ridge National Laboratory, and its used for nuclear research ffs.
Terrible writing. There's also this:
Supercomputer Fugaku was installed in 2020 with 7,630,848 cores and requires 29,899.23 kW of power.

It runs on Red Hat Enterprise Linux and has theoretical peak performance of 537.21 petaflops.

It consumes 29,899kW of power.
Did anyone proof read this?
 
The GPUs are used for general computing. Mostly number crunching.
To add: in particular, they're ideally suited for SIMD workloads (single instruction on multiple data sets) so things like AI, ML, bioinformatics, etc.

And in HPC parlance, they're called GPGPU (general purpose GPU).
 
Will Quantum computers be able to better these machines?
Totally different application area right now.

QC is very specific at the moment, with limited, but arguably useful, results. That being said, it is going to take a lot of money, a lot of patience, and a lot of years before we reap more mainstream benefits from something like QC.

We have a split opinion on the long-term utility of QC because it is still impractical at its current infancy and (in)accuracy. I'm not convinced quantum supremacy will ever meaningfully occur, but QC is not my focus area, but many of my colleagues share my skepticism.

To answer the question - As of right now, QC isn't doing it.
 
Top
Sign up to the MyBroadband newsletter
X