Cray-2

At 1.9 GFLOPS peak performance, it was the fastest machine in the world when it was released, replacing the Cray X-MP in that spot.

This had been attempted in the CDC 8600 in the early 1970s, but the emitter-coupled logic (ECL) transistors of the era were too difficult to package into a working machine.

This was solved in a unique fashion by forcing the electrically inert Fluorinert liquid through the circuitry under pressure and then cooling it outside the processor box.

The unique "waterfall" cooler system came to represent high-performance computing in the public eye and was found in many informational films and as a movie prop for some time.

Cray had previously attacked the problem of increased speed with three simultaneous advances: more functional units to give the system higher parallelism, tighter packaging to decrease signal delays, and faster components to allow for a higher clock speed.

So once again he turned to an 8600-like solution, doubling the clock speed through increased density, adding more of these smaller processors into the basic system, and then attempting to deal with the problem of getting heat out of the machine.

In the era of the CDC 6600 memory ran at the same speed as the processor, and the main problem was feeding data into it.

Cray solved this by adding ten smaller computers to the system, allowing them to deal with the slower external storage (disks and tapes) and "squirt" data into memory when the main processor was busy.

It was the foreground processor's task to "run" the computer, handling storage and making efficient use of the multiple channels into main memory.

Stride conflicts corresponding to the number of memory banks suffered a performance penalty (latency) as occasionally happened in power-of-2 FFT-based algorithms.

In order to address this internal threat, as well as a series of newer Japanese Cray-1-like machines, the Cray-2 memory system was dramatically improved, both in size as well as the number of "pipes" into the processors.

In 2012, Piotr Luszczek (a former doctoral student of Jack Dongarra), presented results showing that an iPad 2 matched the historical performance of the Cray-2 on an embedded LINPACK benchmark.

[1] Due to the use of liquid cooling, the Cray-2 was given the nickname "Bubbles", and common jokes around the computer made reference to this unique system.

Research conducted at the Lawrence Livermore National Laboratory in the early 1990s indicated that to a limited extent the perfluorinated polyether used to cool Cray-2 circuits would break down to form the extremely toxic gas perfluoroisobutylene.

[2] At the time, Cray had created a poster showing the transparent "bubble chamber" that the cooling fluid was run through for visual effect, with a spill of the same material glistening on the floor—the joke was that if this actually occurred, the facility would have to be evacuated.

[3] The manufacturer of the liquid developed a scrubber that could be placed in line with the pump that would catalytically degrade this toxic breakdown product.

A Cray-2 and its Fluorinert -cooling "waterfall", formerly serial number 2101, the only 8-processor system ever made, for NERSC
A Cray-2 operated by NASA
Detail of the upper part of the Cray-2
Inside of the Cray-2
Typical logic module, showing the tight packing. The pogo pins connecting the cards together are the gold-colored rods seen between the ICs.