Cray Research Inc.

Cray-1A

$8,000,000 in 1977
160M FLOPS
5.5 Tons
18MiB

The First True Supercomputer

Cray-1A

The Cray-1A was a later version of the Cray-1, the first Cray supercomputer created by Cray Research in 1975. As the first vector processor supercomputer, the Cray-1A performed operations on multiple data values simultaneously and was faster than earlier supercomputers.

Over its lifetime, Cray sold 80 Cray-1s, making it one of the most successful supercomputers ever made and proving that there was government demand for supercomputing at any price.

To learn more:

The First True Supercomputer

The Cray-1A included several unique features that set it apart from other supercomputers of the time. It featured a unique C-shape design that enabled faster speeds due to shorter wires. It was also the first supercomputer to use integrated circuits.  However, the Cray-1A predates microprocessors that power most modern devices, using high-speed transistors instead.

The hundreds of circuit boards had transistors soldered in grids, which were wired together manually by workers trained in textiles.

To cool the system enough to obtain peak speeds of 160 megaflops, the Cray-1A design used freon gas circulated through metal tubing that ran behind the circuits. Company founder Seymour Cray focused on maximizing speed over reliability, so the Cray-1A operated just 96 hours between hardware faults.

 

To learn more:

IMAGES

The first Cray supercomputer created by Cray Research in 1975. As the first vector processor supercomputer, the Cray-1A performed operations on multiple data.

Cray Research explored different form factors for successive models. Each one exploited advances in integrated circuits. Cray's overriding goal was speed, tempered by keeping system temperature low.

The circuit boards in the Cray-1A were connected by miles of copper wire. To optimize speed, the length of each wire was carefully measured so that data passed from input to processor to memory at exactly the right time. Seymour Cray remarked "anyone can build a fast processor. The trick is to build a fast system." The white squares visible in the wiring are labels, which helped repair technicians replace faulty processors and wires on customer premises.

The Cray-1 was later named the Cray-1A to differentiate it from the Cray-1S and Cray-1M models.

The Cray-1, the first Cray supercomputer created by Cray Research in 1975.

To operate Cray supercomputers, programmers entered instructions and monitored output on separate terminals. Cray-1s were supplied with custom operating system, UniCOS, a specialized version of Unix for use on Cray supercomputers. Peripherals visible in the background are disk storage.

Beneath the ”love seat” in each Cray is a sophisticated cooling system. The vertical aluminum spandrels (between each stack of processor boards) carried liquid refrigerant that removed the tremendous heat generated by processing. The refrigerant was circulated to the base where the heat was transferred by refrigerated air, typically from there the heated air was ventilated outside the building.

Microprocessors were still in their infancy when the Cray-1 was developed, so its processing power came from many boards of integrated circuits wired together.

A Cray-1 and peripherals

CLOSER LOOK

Interior view of a section of the Cray-1s wiring and modules

VIDEOS

Exploring CRAY Supercomputers
With Clive England, Founder of Cray-history.net
2:02 mins
Part 1 - Cray 1_8053
With Clive England, Founder of Cray-history.net
1:10 mins
Part 2 - Cray 1-M_8054
With Clive England, Founder of Cray-history.net
1:10 mins

3D VIEWER