1,000,000,000,000,000,000: This is the number of computer instructions that have been carried out in 1 second by the Frontier Supercomputer at the Oak Ridge National Laboratory, Tennessee US, making it the first computer to operate in the Exascale dimension.
As every year the Top 500 (supercomputer) ranking has been published and for the first time a supercomputer goes beyond the Exascale ceiling, with a reported speed of 1,102 Exaflops (floating point instructions per second).
It has been a hard race, and it did not meet the expectations circulating 10 years ago when exascale computing was foreseen by 2019. We are 3 years late. Still, consider that in 2012 the fastest supercomputer was the Fujitsu K, operating a 10.51 Petaflops (a hundred times slower than current best in class). That is still “faster” than what foreseen by Moore, a hundred fold gain versus a predicted 64 fold increase.
The most recent M2 Apple chip may be approaching 100 Tflops (although it makes little sense comparing the two…), that is over 10,000 times slower than the best in class supercomputer. Notice that the Frontier Supercomputer is using 8, 730,112 cores versus the 8+10 cores of the M2, hence you could say that the M2 cores are almost 50 times faster than the cores in the Frontier Supercomputer (but again, it is an apple and orange comparison…).
The Frontier Supercomputer will start normal operation by the end of this year and among other things it will be used to help research towards fusion energy.