The late 1960’s and early 1970’s saw the rise of the mini-computer. These computers were mini because they no longer took up an entire room. While not something you would stick on your desk at home, they did fit under the desk of many offices. Typically there were built with multiple large circuit boards and their processor was implemented with many MSI (medium scale integration) IC’s and/or straight TTL. TTL versions of the 1970’s often were designed around the 74181 4-bit ALU, from which 12, 16 or even 32-bit processor architectures could be built from. DEC, Wang, Data General, Honeywell, HP and many others made such systems.
By the mid-1970’s the semiconductor industry had advanced enough that many of these designs could now be implemented on a few chips, instead of a few boards, so the new race to make IC versions of previous mini-computers began. DEC implemented their PDP-11 architecture into a set of ICs known as the LSI-11. Other companies (such as GI) also made PDP-11 type IC’s. HP made custom ICs (such as the nano-processor) for their new computers, Wang did similar as well.
Data General was not to be left out. Data General was formed in 1968 by ex DEC employees whom tried to convince DEC of the merits of a 16-bit minicomputer. DEC at the time made the 12-bit PDP-8, but Edson de Castro, Henry Burkhardt III, and Richard Sogge thought 16-bits was better, and attainable. They were joined by Herbert Richman of Fairchild Semiconductor (which will become important later on.) The first minicomputer they made was the NOVA, which was, of course, a 16-bit design and used many MSI’s from Fairchild. As semiconductor technology improved so did the NOVA line, getting faster, simpler and cheaper, eventually moving to mainly TTL.
Data General released the microNOVA mN601 in late 1976 based on their NOVA 3 minicomputer. The mN601 was a NMOS design running at 8.3MHz. It was packaged in a 40-pin DIP and dissipated 1W. It required a 2 phase (4.16MHz non-overlapped) clock, and four different power supplies. Not a particularly easy design to work with. The microNOVA is a 16-bit accumulator based design, with a 16-bit address/data bus and 15-bit PC. I/O is handled serially, 2 I/O ports each send 8-bits of data (synchronized by an I/O clock) to complete a 16-bit I/O operation. This serial I/O handling allowed the mN601 to actually have 3 unused pins, nearly unheard of in 1976.
Being based on a minicomputer made for a rather powerful instruction set. The mN601 has 72 instructions but most of them are highly serial, meaning that one instruction actually completes multiple tasks, common in a minicomputer but certainly not for a general purpose processor of the time. The processor even supported hardware multiply and divide. Memory addressing used 15-bits (one bit was reserved for telling the processor it was using indirect addressing) which allowed for access of up to 64K of memory (32kx16 bits).
In 1979 DGC released the mN602, based on the NOVA 4. The power supplies were reduced from 4 to 3, and memory support was now a full 128K using bank switching (this being the main difference between the NOVA 3 and NOVA 4). A new high speed data channel was added (making use of the previously unused pins) allowing I/O speeds of up to 2 MBytes/sec.
One of the more interesting aspects of the microNOVA was how it was sold. Unlike most minicomputers manufacturers of the time, Data General would not restrict you to just buying a system from them. If you wanted to deign your own Data General offered microNOVA boards, chipsets, and single processors. A development board would cost around $600 while a mN602 processor could be had for $75. The benefits of this were by increasing sales, even of just the IC’s, they could benefit from slightly higher economies of scale, as well as a wider development base. DEC later adopted a similar philosophy but Data General was the first.
There was one of NOVA compatible IC, which was made by Fairchild. The Fairchild F9440 was compatible with the original NOVA (it lacked a stack pointer which was added later in the NOVA line) and was made on an isoplanar I2L process and ran somewhat faster than the microNOVA, clocked at 10MHz. The only problem was Fairchild did not get permission to make a NOVA compatible processor so was subject to a lawsuit about the 9440. That and its $395 price did not help its sales.
The microNOVA continued to be made and sold into the mid-1980’s until it was eclipsed by 32-bit designs, specifically the Eclipse MV/8000. In the late 1980’s Data General released the 32-bit AViiON (NOVA ii), based on the Motorola 88000 RISC processor. Unfortunately for Data General Motorola ended development of the 88K when they decided to transition to PowerPC. Data General was unable to recover and eventually sold to EMC (owners also of VMWare and RSA Security) in 1999.
Perhaps then, NOVA was a fitting name, quickly appearing, bright and powerful, dominating the industry in many ways, before slowly fading to obscurity.