As stated previously, the goal of the designers of the 7030 Stretch was to make the computer 100 times faster than the IBM 704. The 7030 used memory that was 6 times as fast as that in the 704 (2 usec vs 12 usec) and circuits that were about 10 times as fast as the 704. [CAM62, p 202]
They did not meet their design objective completely. With small programs that fit in the 704's word size, the performance increase on the 7030 was a factor of 70. On larger programs that strained the 704, the performance ratio was greater than 100.
Therefore, with about a factor-of-10 increase in CPU and memory speed, the 7030 achieved about a factor-of-100 increase in overall performance. While pipelining was not the only architectural advance, it was the most significant one. I conclude that Pipelining yielded a factor-of-10 performance improvement in the Stretch computer.
The original Stretch computer was delivered in 1961. Fewer than ten more were built. But the ideas were not lost. The Stretch was to be highly influential in the design of IBM's next family of computers, the System 360.
Tony Wesley
Comments to author: tony@tonywesley.com
Last Updated: November 26, 1995