Who Invented the First Supercomputer?
Who Invented the First Supercomputer?
Who Invented the Supercomputer? begins with Charles Babbage. The inventor had disliked the untidiness of logarithmic tables and saw the errors that human calculation had. Babbage drew inspiration from the ideas of Blaise Pascal and Gottfried Leibniz. In 1823, he received government backing for his Difference Engine design. This machine is considered to be the first mechanically operated computer.
Cray was a pioneer in the development of the first supercomputer. The Cray-1 used X-MP processors which ran at 9.5 nanoseconds per cycle and the Cray-1A processors at 12.5 nanoseconds per cycle. The supercomputer was capable of 200 megaflops per processor and had a theoretical peak performance of 400 megaflops. It was the first supercomputer to run Unix and became an instant celebrity.
In addition to having four cores, the CRAY-1 employed twelve pipelined functional units. Each unit contained an add and a multiply module. Using refrigeration, the liquid carried heat away from the modules in the supercomputer. Because the system used liquid to cool the modules, troubleshooting required draining the cooling system and running the machine at a stable temperature. However, the system was so complex that Cray’s engineers were forced to make significant changes to the system.
In addition to its innovative architecture, the Cray-1
also boasted a unique physical design. Its “cray” shape is reminiscent of the shape of a letter C. The concentric bench allows circuits to be closer to each other, and a twisted-pair cable between modules minimizes electrical reflection. As a result, the Cray-1 is capable of running more than one billion simultaneous tasks.
The Cray-1’s I/O was designed for large data sets. In contrast to earlier Cray designs, it used four six-channel controllers, each with sixteen bits of data and three control bits. The speed of data transfers was up to 500 kbps. Depending on the configuration, the memory could be reduced to 0.25 or 0.5 megawords. If this was not enough, it could be upgraded to 32 megawords.
While Cray’s CRAY-1 supercomputing system was a commercial success
, the company continued to push for improvements. The CDC 7600 featured walnut trim and blue glass doors. The Cray-2 supercomputer broke the 1.7 gigaflop ceiling, becoming the first machine to do so. The CDC 8600 followed it in 1986. In 1993, Cray Research, Inc. continued to push the boundaries of supercomputing with the Cray-2. The Cray-2’s 512MW model was the fastest supercomputer ever built and was removed from service at the Minnesota Super Computing Center.
The Cray-1 supercomputer was built by Cray Research in 1976. The first Cray-1 system cost eight million dollars. It had 160 million floating-point operations per second and 8 megabytes of main memory. In 1976, Seymour Cray, CEO of Cray Research, gave a lecture describing the computer’s performance. He explained the scalar properties of Gallium Arsenide and pulse technology. Afterwards, the company sold more than 100 Cray-1 computers.Who Invented the First Supercomputer?
After the CRAY-1’s debut, the company became a major company, with sales of supercomputers surpassing one billion dollars. The CRAY-1’s rapid growth made it possible to handle many tasks simultaneously. However, Cray’s work has been challenged ever since. As the company grew, its products are becoming increasingly more sophisticated. However, the CRAY-1 has been one of the most influential supercomputers ever created.
IBM’s Mark I
The IBM-NCAR partnership follows a banner week for the supercomputer industry, in which both Intel and SGI announced new supercomputers. The next milestone, the petaflop, represents one quadrillion calculations per second. IBM and NCAR are each racing to build supercomputers that can process that many calculations. A petaflop machine could appear on the list as early as later this year. But what will the supercomputers be like in a few years?
The new supercomputer will be used by researchers at the NCAR Climate Research Center to study the effects of climate change, changing weather patterns, and drought on crop yields. Scientists at NCAR will use it to produce the next report for the IPCC and UN. Researchers will also use it to analyze the Bluefire system’s data.
These scientists were a founding member of the IPCC, which gained fame after Al Gore’s 2007 book, “The Inconvenient Truth.”
The first project aimed to build a supercomputer that would be compatible with a variety of other supercomputers, including ACS-1. Amdahl, who had become an IBM Fellow in 1965, argued that the ACS effort should focus on developing one single machine instead of multiple machines for national labs. The ACS team regrouped and merged the hardware and software architecture groups under Herb Schorr. The combined group began with twelve people at IBM Research Yorktown. Another dozen came from Poughkeepsie and San Jose.
Despite its size and weight, the Mark I was a popular machine.
In 1944, Harvard’s Mark I computer was a big success. But the Harvard News Office claimed that Aiken was the sole inventor of the machine, despite the fact that it was designed by IBM. The press release ignored the fact that IBM was a major partner in the project, and it angered Thomas Watson, who had approved the project. However, he later reluctantly attended the dedication ceremony of the Mark I.
The CDC 6600 was the first high-performance computer built by IBM. Later, IBM announced the System/360 mainframe line, and its CDC 7600 minicomputer. As the computer industry matured, IBM sought to keep up with business demands and physics. By the end of the 1970s, it had stopped growing and began to shrink. The company had to find another way to grow its supercomputers. In 1961, Harwood Kolsky recommended the California location for its new supercomputer lab, which would allow scientists from the Livermore plant and San Jose to consult with the IBM Mark I.
The ACS-1 and ACS-360 web pages detail the history of the systems and their development. The Turing Award lecture by John Cocke and Emerson Pugh also referenced IBM’s efforts to build a supercomputer. The book IBM’s 360 and Early 370 Systems also contains detailed information on the history of the IBM supercomputer. The IBM Bluefire system is the first to utilize a water cooling system, while most supercomputers use air cooling.
Fujitsu’s Numerical Wind Tunnel
During the late 1990s, the National Aerospace Laboratory of Japan and Fujitsu worked together to develop a supercomputer called the NUMERICAL WIND TUNNEL. This supercomputer, built with vector parallel architecture, was the first supercomputer to reach sustained performance of close to 100 Gflop/s. During its lifetime, the NUMERICAL WIND TUNNEL topped the TOP500 with a Rmax of 124.0 GFlop/s and a peak of 235.8 Gflop/s in November 1993.
The Numerical Wind Tunnel was developed in 1993 by Fujitsu and the National Aerospace Laboratory. Used to simulate wind turbulence and forecast weather. It was capable of 124.5 gigaflops when it first debuted, making it the first computer to surpass the 100-gigaflop barrier. It also featured 256 megabytes of memory per processor. Later, the Numerical Wind Tunnel was upgraded to 167 processors and reached 170 gigaflops.Who Invented the First Supercomputer?
In the 1990s, Japanese systems had begun dominating
the supercomputer market, taking the lead from US designs. The Numerical Wind Tunnel, designed by Fujitsu, was the fastest supercomputer from 1993 to 1996. It could compute a million times faster than the previous record holder, the IBM PDP-11. Gigaflops is one billion floating point operations per second. It was also one of the fastest supercomputers of all time.
The numerical wind tunnel was developed based on the Navier-Stokes equations of incompressible and viscous fluid. It has been validated using experimental results, such as the simulation of a low-rise building with a green roof. The green roof reduces the drag force on the building and the variations near the roof, but has little or no effect on the time-averaged velocity field.
The VP-200 and VP-100 systems were released in 1983 and 1986, with the VP-400 being even faster. The E-series was introduced mid-1987, and added a multi-add pipelined floating point unit to increase performance potential. Fujitsu became the leading domestic supplier for VF-200 and VP-400 systems, and is the number one choice for wind tunnel simulations. The company has over 80 systems installed and many of them have been named in the TOP500.
The numerical wind tunnel is a tool for reproducing the test results obtained from a full-scale Jules Verne climatic wind tunnel. In wind turbine applications, the numerical wind tunnel simulates the flow inside and around stator vanes. It is useful in developing wind turbine designs, and the data gained from this research is invaluable for predicting the future performance of the turbine. But in other fields, it is more important than ever to have these data in hand.Who Invented the First Supercomputer?