To appreciate the scale of a supercomputer, you must understand how its performance is measured. We measured them in FLOPS, which stands for floating – point operations per second. Floating-point encoding allows for very long numbers to be handed with ease, and is useful in a number of scientific fields.
1. (CDC 6600)
Processing power: 3 million FLOPS
The CDC 6600 was designed by Seymour cray, the father of supercomputing, and is widely considered the world’s first supercomputer. It would be the fastest computer in the world for five years until it was replaced by the CDC 7600.
2. CRAY – 1
processing power: 80 million FLOPS
The Cray – 1 was designed and manufactured by Cray Research, and came to be one of the most popular and successful supercomputers in history. It was the first supercomputer to fully exploit vector processing.
3. CRAY X – MP
Processing Power: 400 million FLOPS
The CRAY X – MP was the successor of the Cray – 1, and was also built by Cray Research. It was the world’s fastest supercomputer from 1983 until 1985. The Cray X – MP is famous for helping render lighting effects for 1984’s The Last Starfighter.
4. CRAY – 2
Processing Power: 1.9 billion FLOPS
The Cray – 2 replace the Cray X – MP as the world’s fastest computer when it was introduced in 1985, a title it held until 1990. It was also the last Cray – branded computer to hold the title. The Cray – 3 was a famous flop, and drove Cray to bankruptcy in 1995.
5. Numerical Wind Tunnel
Processing Power: 124 billion FLOPS
The Numerical wind tunnel was developed by fujitsu in partnership with the national Aerospace Laboratory of Japan. It achieved the rank of world’s fastest supercomputer multiple times throughout 1993, 1994, and 1995.
Processing Power: 600 billion FLOPS
Japan continued their supercomputer dynasty throughout the mid-90s. After the Numerical wind tunnel came Hitachi SR2201, which was the world’s fastest for a portion of 1996. It held the title briefly until the ASCI Red exploded onto the scene.
7. ASCI RED
Processing Power: 1.33 trillion FLOPS
The A5CI red was built by Intel and installed at Sandia National Laboratories in 1996. It was the first supercomputer to cross the teraflop barrier, and would eventually reach speeds of two teraflops after upgrade. It was the world’s fastest until the turn of the millennium.
Processing Power: 1.02 quadrillion FLOPS
Though there were a number of notable supercomputers released throughout the 2000s, none come close to matching IBM’s Roadrunner. It was the first computers to break the petaflop barrier, achieving a peak performance of 1.7 petaflops/s.