TX-0

The TX-0, for Transistorized Experimental computer zero but affectionately referred to as the tixo, was the first fully transistorized computer to enter service and contained a then-huge 64K of 18-bit words of core memory. TX-0 went online in 1956 and was used continually into the 1960s.

Designed at the MIT Lincoln Laboratory largely as an experiment in transistorized design and the construction of very large core memory systems, the TX-0 was essentially a transistorized version of the equally famous Whirlwind, also built at Lincoln Labs. While the Whirlwind filled an entire floor of a large building, TX-0 fit in a single reasonably sized room and yet was somewhat faster. Like the Whirlwind, the TX-0 was equipped with a display system, in this case a 12" oscilloscope hooked to output pins of the processor allowing it to display 512×512 points in a 7" by 7" array.

TX-0 was never really intended to be used as a complete system on its own. The machine included 64K words of memory, requiring 16 bits of address space, but in order to keep costs down the instructions were only 18 bits long. This allowed for only 2 bits of instructions, a total of four. The TX-0 included only store, add and branch instructions as a basic set. However the fourth instruction, "operate", was further decoded and allowed access to a number of "micro-orders" which could be used separately or together to provide many other useful instructions.

With the successful completion of the TX-0, work turned immediately to the much larger and far more complex TX-1. However this project soon ran into difficulties due to its complexity, and was redesigned into a smaller form that would eventually be delivered as the TX-2 in 1958. Since core memory was very expensive at the time, several parts of the TX-0 memory were canibalized for the TX-2 project. After a time the TX-0 was no longer considered worth keeping, and was "loaned" (semi-permanently) to the MIT Research Laboratory of Electronics in July 1958, where it became a centerpiece of what would eventually evolve into the MIT Artificial Intelligence Lab.

Delivered from Lincoln Labs with only 4K of core, the machine no longer had to use 16 of its 18 bit instructions to store a location, so after about a year and a half the number of instruction bits were doubled to 4, for a total of 16 instructions, and an index register was added. This dramatically improved programmability of the machine, but still left room for a later upgrade to 8K. This newly-expanded TX-0 was used to develop a huge number of advances in computing, including speech and handwriting recognition, as well as the tools needed to work on such projects, including text editors and debuggers.

Meanwhile the TX-2 project was running into difficulties of its own, and several team members decided to leave the project and start their own company. After a short time selling "lab modules" in the form of single modules from the TX-2 design, the newly-formed Digital Equipment Corporation (DEC) decided to produce a "cleaned up" TX-0, and delivered it in 1961 as the PDP-1. The first PDP-1 would eventually be installed in the room next to TX-0, and would run side-by-side for some time.

External links

Tixo.Org (http://tixo.org)

RLE Technical Report 627 TX-0 Computer History (Oct 1974) PDF (https://dspace.mit.edu/bitstream/1721.1/4132/1/RLE-TR-627-42827671.pdf)

The TX-0: Its Past and Present (http://ed-thelen.org/comp-hist/TheCompMusRep/TCMR-V08.html)

TX-0 documentation (http://www.bitsavers.org/pdf/mit/tx-0)

TX-0 programs (http://www.bitsavers.org/bits/MIT/tx-0/listings/)