Supercomputers are used to run high time/space complexity or complicated simulations that involve a large number of variables.
IBM’s Summit, the former world’s most powerful supercomputer which was capable of 150 petaflops(one thousand million million (1015) floating-point operations per second) was taken over by Japanese supercomputer, Fugaku which was developed by RIKEN and Fujitsu is capable of 442 petaflops which is almost 3 time more then Summit.
specification of Fugaku
| Manufacturer: | Fujitsu |
|---|---|
| Cores: | 7,630,848 |
| Memory: | 5,087,232 GB |
| Processor: | A64FX 48C 2.2GHz |
| Interconnect: | Tofu interconnect D |
Performance | |
| Linpack Performance (Rmax) | 442,010 TFlop/s |
| Theoretical Peak (Rpeak) | 537,212 TFlop/s |
| Nmax | 21,288,960 |
| HPCG [TFlop/s] | 16,004.5 |
Power Consumption | |
| Power: | 29,899.23 kW (Optimized: 26248.36 kW) |
| Power Measurement Level: | 2 |
And like most supercomputers it is using Linux by Red Hat Enterprise. 
Reason is its reliability, stability and security.
More specification and statistics about Fugaku.
Current rankings
Source: Statista Reference with RTX 3090
And just to give some reference of Fugaku with NVIDIA GeForce RTX 3090, Fugaku is 12278X faster then rtx 3090.
Geforce RTX 3090 = 36 Teraflops
Fugaku = 442010 Teraflops
| Geforce RTX 3090 | Fujistu Fugaku |
| 0.00008% | 99.99992% |
Contribution in research
In the case of Fugaku, the machine’s power is also used to research the COVID-19, more specifically to run simulations of how respiratory droplets move through the air in different settings like offices or train carriages.
Cost of manufacturing
The estimated cost of this super computer is about 1.2bn$. And it took about 9 years to manufacture.
Fugaku world’s best supercomputer and karan s world’s best smart person blog man programmer pro gamer <3