|

Collector Nico Baaijens
Free Office Suite Highly Recommended

Webbrowser - eMail
Fast - Free - Complete
Free Antivirus
|
|
| |
Antiquities (4)
Charles Babbage's Difference Engine (1812). The principle on which this famous machine was constructed is called the Method of Differences. If any table is taken and each of the tabular numbers is successively subtracted from the following one, another column of figures will be obtained which is called the first order of differences. If again this column is subjected to a similar operation, a second order of differences is arrived at, and by repeating the process, a third order of differences may be obtained, and so on. Babbage intended to give his machine six orders of differences, but it was never completed. |
|
|
|
A.J. Thompson's integrating and differencing machine (1920). Thompson connected four Triumphator business machines so that the number on the register of any machine is transferred to the setting levers of the machine immediately below.
Once commercial machines could be used as difference engines, this type of lash-up was no longer necessary. |
|

|
Herman Hollerith developed an electric adding mechanism and incorporated it into his tabulating equipment. This made it feasible for e.g. railroads to use punch card machines for their waybill statistics: what was shipped, who shipped it, who received it, how much it weighed, the shipping charges and routes taken. Insurance companies were quick to see the advantages of punch cards and machine tabulation. |
|
|
|
Howard Aiken's Automatic Sequence Controlled Calculator (1937). An electro-mechanical relais-based digital computer. Aiken's credo: 'There exist problems beyond our ability to solve, not because of theoretical difficulties, but because of insufficient means of mechanical computation.' IBM's Thomas Watson jr. ordered the building of this huge machine because he saw a future beyond punch card equipment. |
|

|
Captain Hermann Goldstine (left) and J. Presper Eckert Jr. with a decade plug-in unit. The ENIAC (Electronic Numerical Integrator And Computor) had a total of two hundred such units, giving it a memory capacity of twenty numbers of ten digits. |
|
|
The IBM Stretch computer from the 1950s contains many high-performance design features that we usually think of as being associated only with current day superscalar microprocessors. Gene Amdahl and John Backus were influential in proposing an instruction "lookahead" approach to start memory fetches early and queue up operations for a fast arithmetic unit. John Cocke and Harwood Kolsky later helped refine the lookahead design by developing a timing simulator used in tradeoff studies. The register set and function unit partitioning as well as the resulting pre-execution of certain instructions in the instruction stream are ideas from Stretch that have influenced high-end processor design within IBM for decades. |
|
|
|
The IBM 7080 was a variable word length BCD transistor computer in the IBM 700/7000 series commercial architecture line, introduced in August 1961, that provided an upgrade path from the vacuum tubeIBM 705 computer. After the introduction of the IBM 7070, in June 1960, as an upgrade path for both the IBM 650 and IBM 705 computers, IBM realized that it was so incompatible with the 705 that few users of that system wanted to upgrade to the 7070. That prompted the development of the 7080, which was fully compatible with all models of the 705 and added many improvements. |
|

|
IBM System/360 Model 91
The largest model of IBM's first mainframe series. Announced in 1966. This system was installed and fully operational as one of the very first supercomputers at NASA's Goddard Space Flight Center. Within the central processing unit (CPU) there are five highly autonomous execution units which allow the machine to overlap operations and process many instructions simultaneously.
The Model 91 CPU cycle time is 60 nanoseconds. This is the time it takes to perform a basic processing operation. Its memory cycle time - - the time it takes to fetch and store eight bytes of data in parallel -- is 780 nanoseconds.
The main memory, however, is organized into 16 interleaved elements, so that the CPU may start a memory cycle with a different element every 60 nanoseconds instead of waiting the full 780 nanoseconds. The NASA Model 91 operated with 2,097,152 bytes of main memory interleaved 16 ways. Model 91s could accommodate up to 6,291,496 bytes of main storage. |
|

|
|
| |
|