As game developers, part of our job is to keep current with the latest technologies. But I can’t help myself, I love old computers, and I am going to tell you the story of the ill-fated BINAC. Although it was great for its time, history remembers the BINAC as a failure, and the little information to be found about it is scattered all around. So I collected everything I could, and here is the story I came up with.
Many great companies start from the association of a technical genius and a visionary. John Mauchly and Presper Eckert were such people, they were behind the creation of the famous ENIAC computer and they designed its successor, the EDVAC. Unwilling to concede the ownership of all their ideas to the Moore School of Electrical Engineering they decided to leave and founded the Eckert and Mauchly Computer Company in 1946. After securing a contract to build the UNIVAC, a successor to the EDVAC, they happily started hiring and working. The UNIVAC was a huge project, it would take five years to complete, and EMCC needed another contract to keep the money coming…In 1947, the “let’s quickly make a small project on the side to finance our big project” was already a bad idea, and the BINAC was a brilliant demonstration of the principle. Of course, nothing would be possible without overconfidence and commitment to unrealistic schedules together with crazy specifications and a customer tangled in bureaucracy, but the BINAC got it all.
Just to understand how crazy the specifications were, let’s remind that the ENIAC cost $500,000, weighted 60,000 pounds, and was three years in the making. The yet to be build EDVAC would cost about the same for a weight of 17,300 pounds and was to be four years in the making. Now EMCC committed to the BINAC, a computer with roughly the same abilities as the former two, but for a price of $100,000, a weight under 700 pounds (the idea was to have it on a plane), and delivered in 7 months. Yay! All aboard the doom train!
Of course, they completely ran over those specs. But the BINAC came to life anyway, taking two years, costing $278,000 and weighing around 5,000 pounds. Given the constraints it was a remarkable achievement. It had an odd design, consisting basically of two identical computers meant to check on each other for errors, it had a whooping 512 words of memory with 30 bits each (plus one for the sign), and its instructions all had a 6 bits opcode with a 9 bits address (it only had 16 opcodes in total, but everything was octal, so there were two unused bits), so you had two instructions per word. It only had two registers, and had no immediate opcodes. There was no computer-controlled I/O either, all transfer from and to tape had to be triggered manually, and when the program was done running (or when the breakpoint switch was set), the operator could dump the memory on the printer and find the results back. The ALU could add, subtract, multiply, divide and bitshift. The only conditional instruction was “jump if the accumulator is negative”.
But did it work? Yes it did! When demonstrated to the customer, Northrop Aircraft, the BINAC found 36 solutions to the Poisson equations in a little more than three hours. I actually have no idea what that means, but it sounds impressive. But that was to be the last thing the BINAC did, because from there Northrop Aircraft decided to “take over” with little to no idea how to properly handle and use this amazing piece of technology. They dismantled it and threw it in the back of a truck, carrying it without much care from Pennsylvania to California, they then left it unattended for months on a hangar’s floor only to finally have a newly graduated electrical engineer try to put the pieces back together. When the BINAC finally did run again, it was too unreliable to be of any use, and the security measures in place would not allow anybody from EMCC to come and have a look. In the end, Northrop found out that a few gyroscopes would solve the problem they were intending to use the BINAC for…So here is the fate of the BINAC: a brilliant computer that was never used for any practical purpose, designed for a misunderstood problem, left to rust by people who did not care about it, blamed for its unreliability, then finally trashed. There is not much place in history for such a machine, but it proved to be a wonderful learning ground, and paved the road for many computers to come.
By the way, if you have some technical docs about the BINAC, I am highly interested. The only one I could find is a 9 page overview with a list of opcodes.