A museum curator suggests Russia's BESM supercomputer may have been superior to ours during the Cold War.
By: Doron Swade
Russia's BESM class of supercomputers, designed more than 40 years ago, may give the lie to U.S. declared technological supremacy during the Cold War years
November 18, 1992. Wednesday. Deep in the heart of Siberia lives a winking monster. It is rumored to occupy an entire floor and to be outfitted with thousands of flashing console lights. Once part of the U.S.S.R.'s military, space, engineering, meteorological, and computer-science programs, it is the last working version of the legendary BESM-6 supercomputer. It stands amid the debris of its three compatriots, which have been smashed and melted down for the salvage value of their precious-metal content. I have pushed forward the timing of our visit to the former Soviet Union in an effort to save the last machine from the same violent fate.
We have come to this place of snow and bitter winds to secure the working BESM-6 supercomputer for the National Museum of Science and Industry, in London. In preparation for the trip, I've read whatever I could find on Russian computers. The search of the literature has been both baffling and revealing. I've learned that Russian computer culture has its own tribal icons: Ural, MESM, Riad, Nairi, Strela, BESM, Elbrus -- acronymic tags as rich in shared history and personal associations to the Soviet computer community as our acronymic mantras are to us. Yet because of security controls during the Cold War years, those machines are practically unknown to Western historians of computing and barely mentioned in the historical canon.
The years of silence were broken when the Iron Curtain lifted and Westerners were given access to centers of excellence previously off-limits. They've returned with tales of Russian-designed machines built into DEC consoles that outperform their Western equivalents by several orders of magnitude -- like Volkswagens with V-8 engines -- and of ternary-logic computers (known as MIRs) for which there is no Western counterpart. Eager to corroborate those stories for ourselves, we set off for the Institute of Informatic Systems, in Novosibirsk, Siberia.
I am curious to see the legendary machines I've read about -- the URAL, the MIR, and especially the Elbrus, the Burroughs-based supercomputer that succeeded the BESM. Shortly after our arrival, I turn to one of our hosts, Dimitri, a young computer scientist at the institute who will be our primary translator, and ask about those historic computers. My questions are met by blank stares and polite evasions, so I let the subject drop.
We then embark on several days of intense negotiations over price and shipping arrangements for the historic equipment we've come to buy -- the BESM, a Kronos workstation, and an AGAT personal computer (the Russian Apple II), among other machines. Agreement on each clause is accompanied by elaborate speechifying, a mandatory one-gulp swig of vodka, and renewed good fellowship.
On the third day our unremitting schedule of back-to-back meetings is suddenly canceled. Out of the blue Dimitri announces, "At 3:30 you see Elbrus."
And so I learn the central tenet of doing business Russian-style: what matters is not what you do or your level of authority; what matters is with whom you've forged personal ties. The three days of negotiations appear to have established the requisite trust. Now our hosts cannot do enough for us.
* * *
November 21, 1992. Saturday. We need a break. The hours of nonstop talk and concomitant drink have left us bleary-eyed and hoarse. Dimitri and three of his friends from the institute take us to a vast flea market that operates year-round in the frozen wastes outside Novosibirsk. The market is called barakholka, which means, literally, "rubbish place." We're told to conceal our foray from the directors of the institute: they are nervous about the physical risk to foreigners from hostile natives. Dimitri warns us to carry no money or cameras and in no circumstances to speak English. If we want to purchase something, we are to signal and withdraw out of earshot. Our renegade young hosts from the institute will transact business for us.
The temperature is well below zero, and light snow is falling. Next to livestock, car parts, furs, frozen meat, and household goods, we see stalls with integrated circuits (ICs), electronic components, peripherals, radio parts, and partial chassis and subassemblies -- an open-air Siberian Lyle Street (London's stretch of shops selling used and cannibalized computer parts). Prized among the loot are reverse-engineered Sinclair ZX-Spectrum clones (Sinclairs were tiny computers built in the United Kingdom in the early 1980s that sold for about $150) with Russian documentation and games stored on audiocassette tapes. The clones come in a variety of shapes, colors, and designs, and bear little resemblance to their Western counterparts. Their motherboards were made unofficially in state electronics plants by underemployed workers, who then assembled the computers at home and sold them in ones or twos either privately or at flea markets. We end up purchasing two Sinclair clones; one of them comes with a guarantee -- a handwritten note with the telephone number of the teenager who assembled the device. Cost: the equivalent of $19 U.S.
We return to the institute with our treasures. Once inside I'm struck by a contradiction: the building's abundance of personal computers flies in the face of the regulations imposed by the Coordinating Committee for Multilateral Export Controls (CoCom) during the Cold War years -- regulations that restricted Eastern-bloc countries from purchasing advanced technology from the West. The idea was to prevent their military, in particular, from pirating Western advances. Long after 16-bit machines were off-the-shelf consumer items in the United States and Western Europe, CoCom regulations confined advanced technology for West-to-East trade to 8-bit machines with restrictions on clock speed and memory size. I mention this to Dimitri.
"Yellow PCs," he laughs, waving an arm at the secretaries' colored screens. He explains that the computers are branded and unbranded machines acquired through back-door deals with factories in East Asia under contract to Western companies.
"So," I say, "Russians have the same passion for personal computers as Westerners?"
Dimitri responds by pointing to the institute's barred windows. "How far apart do you think they are?" he asks.
I look at him quizzically.
"Just less than the width of a PC," he says. He assures me that he is serious and explains that the bars were installed to stop people from lowering PCs out the windows to others waiting below.
But something still baffles me. How, I wonder, does this jibe with the seemingly antitechnology bias I've noticed outside the institute's walls? Next to the cash register in most shops and hotels in the country is a Russian abacus, a schyotti. Shopkeepers and desk clerks do their calculations on the schyotti, and then enter the total and ring it up on the till, even though most of the tills can do the addition automatically.
When I ask Dimitri about this odd practice, he explains that the general public distrusts new technology, that the schyotti is a symbol of a traditional, trusted procedure. Paradoxically, the schyotti is now threatened by rampant inflation: the traditional wooden frames and wire crosspieces cannot hold enough beads to handle the smaller denominations of an increasingly devalued currency.
* * *
November 23, 1992. Monday. It's time to wrap up our negotiations for the BESM, arguably the most influential computer in the history of Soviet computer science. The giant machines -- from the prototype, the BESM-1 (1953), to the last of the line, the BESM-6 (1966) -- were the workhorses of scientific and military computation, and the four-BESM system at the institute had at one time supported more than 300 independent users. The BESM-6 is of particular interest: it is reputedly the last indigenous Russian computer that performed on a par with its Western counterpart, the Control Data supercomputer of the mid-1960s. More than 350 BESM-6s were built. The last of them were taken out of service in the early 1990s.
Our negotiations for the supercomputer have been tortuous but ultimately successful. The system we will be shipping home comprises a full BESM processor (a hybrid germanium/silicon discrete component transistor system with IC memory supporting standard disk and tape storage), power-supply cabinets, multiple peripherals, cabling, documentation, and spare parts. With a more detailed understanding of this seminal Soviet supercomputer, we can perhaps revisit Cold War claims about the purported Russian technology lag and dispel or confirm some of the myths about the technological prowess of our new and tentative allies.
* * *
Doron Swade (d.swade@ic.ac.uk) is senior curator of computing and information technology at the National Museum of Science and Industry, in London.