IBM’s Scientific 700 series

Here are some notes on the physical aspects of IBM’s 700 series of scientific computers. This note covers the IBM 701, 704 and 709. All of the equipment I describe here was built and delivered by IBM. There was also the 702 and 705 which were business machines which had different accouterments and with which I was not familiar and do not discuss.

There was a separate power control console with meters and power sequencing logic. It was perhaps 5 feet high and 4 feet wide and perhaps two feet thick.

These machines all used 36 bit sign and magnitude words. The 701 had limited support of 18 bit signed arithmetic.

The central processing unit (CPU) in each case was a box about 6 feet long and five feet high and nearly three feet wide. At the long end was an operator console with indicator lamps and switches. The other end of the CPU was a hinge to open the box into two leaves, perhaps 18 inches thick, for ease of maintenance. I never saw that hinge closed except just before of after shipment. It normally stood open at 90° likely with an oscilloscope at the ready. In each leaf there were vertically stacked four horizontal ranks of units. Each unit held four vacuum tubes together with assorted capacitors, resistors and a few diodes(?). A unit was about 7 by 7 by 1 inches.

The 701 had one or two boxes of Williams tube memory. Each box provided 2048 words of storage. Access time was 12 μs but usually 24 μs counting the refresh cycles. These boxes were about 6 feet tall, 6 feet wide and 3 feet deep. They were cabled directly to the CPU.

The early 704’s had one or two boxes of core memory—each box with 4094 words of memory. These boxes were about 10 feet long, 5 feet high and 3 feet deep. Later memory upgrades (738 core memory) delivered 32K words in one large box. Most or all 709’s had the latter. The 738 was about 5 feet high and T shaped. Each of the 3 T branches was about 5 feet long and 2 feet thick. The core memories had access times of 12 μs with no refresh cycles required.

This page tells how IO devices appeared to programs.

The “unit record equipment” consisted of a printer, card reader and a card punch, each with its own plug-board. This equipment was used also in the 7090 and 7094 series. The printer was a somewhat stripped down business accounting machine that printed with one print wheel per column. The printer was about 6 feet wide, 4 feet high and 4 feet deep. The normal accounting machine would have a card reader but this printer had none. The programming interface was rather like punching a card with Hollerith information. It printed 72 columns of data at 150 lines per minute, or half that speed for wider text.

The card reader read 150 cards per minute, 9-row first. It was about 3 by 3 by 3 feet. The card punch did about 100 cards per minute and was about 4 by 4 by 3 feet.

The 701 had magnetic tape drives with two drives per unit. The unit was about 5 feet wide 3 feet deep and 5 feet high. The unit could write data forward and read in either direction. I think that the density was 100 characters per inch. There were 7 tracks including a parity track. Livermore had two units providing four drives.

The 704 and 709 had taller and faster tape drives, one drive per unit. These drives were components of other IBM computers. I think that there was a tape controller between the tape drives and CPU. I think that there had been no such controller for the 701.

The 701 and 704 had optional drum memory. Most early 704’s had drums because Fortran required them. I think there were two drums per box and some machines had two boxes. The boxes were about 5 feet long 4 feet deep and 6 feet high. I think there were 2K words per drum. With the 738 core memory there was little need for drums and few if any 709’s had drums.

The 709 had channels which would be called DMA today. There were small control boxes providing operator access to the channel. Such access was seldom needed. the control boxes were about 3 feet high and 2 by 2 feet. A channel was in a separate box and I do not remember the dimensions. Most machines had two channels but 6 could be installed on a 709. Like today’s DMA each channel could be performing an IO operation independent of the CPU and other channels.

Some 704’s had a display which looked like a TV set designed by an IBM engineer. There was also the accompanying film recorder several feet square and 5 feet high.

The tape drive included vacuum columns