From Chessprogramming wiki
Jump to: navigation, search

Home * Hardware * Nova

NOVA Rack [1]

Nova, (Data General Nova)
a 16-bit minicomputer series built by the American company Data General starting in 1968. The Nova was designed by DEC PDP-8 chief engineer and Data General co-founder Edson de Castro. It was packaged into a single rack mount case.

Unlike the PDP-8, Nova was a load/store architecture. It had four 16-bit accumulators, where two could be used as index registers, and a 15-bit address space and PC. The Nova is a big-endian architecture. Since there is no byte addressing, bytes need to be parsed out of words using swaps and masks high-order byte first [2]. Nova consists of a nibble-serial 4-bit ALU - its RISC-like instructions perform arithmetical and logical operations with the options to rotate, test and branch on the (skip next instructon on zero, carry) the 17-bit result, and also to discard the result otherwise written into the target accu. Basic models of the Nova came without built-in hardware multiply and divide. The first models were available with 1 to 8K words of magnetic core memory.

System software provided include the real time operating system RDOS, assembler, Basic interpreter, and Fortran and Algol compiler, expanded with Forth, Lisp, and C through third party vendors [3].

Nova Line


The SuperNOVA subsequently replaced initial magnetic core memory with faster ROM for library routines, and semiconductor (SuperNOVA SC) memory. The 4-bit ALU was extended to 16-bit using four instead of one bit slice 74181 ALU with speedup correspondingly [4].

Nova 1200

Nova 1200 Front panel [5]

The Nova 1200 contained the entire CPU one one board, first shipped in 1971 [6]. It still had the Nibble-serial ALU, and up to 32K words magnetic core memory.

Nova 8x0

The faster Nova 800 was released in 1971. The Nova 840 introduced memory mapping in 1973, allowing two discrete sessions running concurrently, each with its own controlling terminal [7].

Nova 2

The Nova 2 was essentially a simplified version of the earlier machines as increasing chip densities allowed the CPU to be reduced in size, with CPU and memory on a single board, introduced in 1973.


Eclipse S/130 front panel [8]

The Eclipse line, started in 1974, had an advanced, Nova upward-compatible instruction set, and included support for virtual memory and multitasking. The line was succeeded by the 32-bit Eclipse MV minicomputers in the early 80s, whose development was subject of Tracy Kidder's book The Soul of A New Machine.

Nova 3

In 1975, the Nova 3 combined features from all previous Nova designs, and added a hardware stack and appropriate stack instructions [9]. The Nova 3 was reduced to a chip set in 1976, called the microNOVA with hardware Multiply/Divide, optionally before, becoming a standard [10].

Nova 4

The Nova 4 was the last of the Nova line, released in 1987, the CPU a derivation of the Eclipse S/140. The Nova 4 is implemented around four AMD 2901 bit-slice chips and, unlike all earlier Novas, is microcoded.

Chess Programs

See also

External Links