Glossary of Chip Collecting Terms and Definitions
COmplementary Silicon MetAl-oxide Conductor. RCA terms for its first CMOS (COS/MOS) microprocessor architecture.
A chip technology where transistors are implemented as NPN or PNP junction type transistors.
Metal Oxide on Silicon Field Effect Transistor. A chip technology where the transistors are implemented as Field Effect Transistors.
A MOS or CMOS technology where FET gates are implemented with silicon
A MOS technology where FET gates are implemented with silicon
N-Channel, Depletion Load, Silicon Gate MOS
Same as CMOS, early RCA term for CMOS technology
Complementary Metal Oxide Semiconductor / Silicon on Sapphire - A radiation hardened chip technology developed by the department of defense and first put into production by RCA.
Intel's high speed CMOS process
MCS originally stood for Micro Computer Set, Set referring to the "set" of chips in the family. Intel continued to use this naming convention to describe its other families processors and eventually the "Set" became "System". The Sets included the processor chip and other chips specifically designed for used with processor, primarily memory and interface chips.
Here are few of Intel's MCS families:
Microprocessors are classified as one of three types of devices: single-chip central processing units, single-chip microcomputers (or microcontrollers), and bit-slice devices.
CPU, or Central Processing Unit, is a term borrowed from the microprocessor's bigger cousins. When computers filled rooms, the part of the computer that did the actual math and logic operations was referred to as the Central Processing Unit or the Main Frame. Using the same analogy the printed circuit board on which the microprocessor is located is referred to as the Mother or Main Board and microprocessor the Central Processing Unit. A better term for the microprocessor, which is in some use, is Micro Processor Unit MPU.
MPU's devices are often implemented on one piece of silicon (monolithic) and are comprised on two major logical sets of circuitry the control unit (CU) and the arithmetic and logic unit (ALU). The CU controls the actions of the CPU and the ALU does the math and logic functions. The CU is designed to accept instructions, decode them, and send the appropriate signals to the ALU to execute the instruction. The Intel 4004 was the 1st MPU. Other examples: RCA 1802 and MOS 6502.
Vintage ALU's often could only do integer math. For data that was floating point, any number with a decimal point, the calculation was handled by writing programs to break down the math operation into a series of integer operations. As this was a slow calculation, special purpose arithmetic chips were developed. These circuits were developed specifically to do mathematic functions on floating point data. These were designed to be co-processors with the MPU. Generally these math co-processors were referred to as a Floating Point Processor, or FPU. Early versions of these devices would provide a single function, such multiply, and not surprisingly were referred as hardware multipliers (examples: the Intel 8321, TRW TDC1010). Later true FPU's provided many different functions (examples: the Intel 8087, the Motorola 68881).
Micro Computer Units are MPU's that include RAM and ROM on a single chip. The are often referred to as being a computer-on-a-chip. This on-board memory allows a very inexpensive computing solution. Where a computer with an MPU may have several chips on a printed circuit board, the MCU includes all of the functions in a single chip. The MCU was developed to provide computing power to automate machines. MCU's are found in industrial robots/controllers, automobiles, vending machines, etc. Because MCU's are often found in industrial control equipment, they are also referred to a Microcontrollers. Essentially any application that required relatively simple logic and minimal storage was a candidate for an MCU. MCU production has far out numbered MPU production almost since day one. Examples: Intel 8048, TI TMS1000
Bit-slice microprocessors are the Legos of microprocessors. In bit-slice microprocessors the MPU is split apart into CU and ALU chips. At first blush, this may seem to go against the rationale for using a MPU. Splitting these functions apart into multiple increases complexity and expense. After all the MCU was built to shrink a computer and bit-slice seems to be expanding it. Actually it is and does, but there are some interesting advantages to the bit-slice approach.
The bit-slice microprocessor design has three very significant advantages. The first is that the ALU's can be attached together in horizontal configurations to create computers that can handle very large chunks of data at a time. An example of a bit-slice processor is the AMD 2901. With the AMD 2901 bit-slice, the 2901 is the ALU, the AMD 2910 is the CU. The 2901 was a peer of the Intel 8080, but the 8080 could only handle 8-bits of data at time. The 2901 was a 4-bit ALU, but 4 2901's could be linked together to create a computer that could handle 16-bits at a time, 8 could be put together to create a 32-bit computer and so on. Whereas the 8080 would have to use multiple cycles to process 16 or 32 bits, the appropriate 2901 configuration could handle it in a single cycle giving such a computer significantly more power than the 8080. The second, advantage of the Bit-slice design is the fact that the two chip design allowed the chips to use bipolar chip technology (example: the Intel 3002). Bipolar is very fast, but consumes lots of power and dissipates lots of heat. Because of the heat dissipation problem, bipolar chips could not be as dense (in number of transistors per area) as the PMOS or NMOS chips. It was not possible to build single-chip CPU's using bipolar technology. So, in addition to the wider data paths the bit-slice devices could achieve, they were inherently faster due the bipolar technology that was employed to build the chips. The third advantage the bit-slices had was the ability to allow users to create their own instruction sets for their applications. Instructions sets could be created to emulate, or enhance, existing processors such as the 6502 or 8080, or to create a unique instruction set specially adapted to maximize performance of a specific application. The combination of the speed and flexibility of bit-slice devices made the very popular and created a cult-like following, especially for the AMD 2901.
Arithmetic and Logic Units provide mathematical and logic functions. Without getting too technical ALU's really only perform logic functions. Math is performed by logic operations that have the effect of adding and subtracting. Boolean logic, AND, OR, NOT, XOR are performed. ALU's can also shift bits left and right. And, also, through logic functions ALU's can compare numbers. Expanded versions of ALU's, most often found in FPU's, have additional circuits to provide hardware based multiply, divide, and other transcendental math functions. ALU's are fixed in the size numbers that they can work with due the number of bits they are designed to handle. Currently ALU's range from 4-bits (handing numbers as big as 15) to 64-bits (handling numbers up to 1,844,674,406).
ALU's are an evolution in logical circuits. The earliest math circuits were called Half-Adders. Half-Adders could take two bits and add them together, giving a result and, if necessary, a carry. Full-Adders were the next step and provided the ability to add two-bit together, along with a carry from a previous addition. Full-adders gave designers the capability to put multiple Full-adders in parallel to add multi-bit numbers in a single operation. The ALU built on this concept and implemented multiple Full-Adders on a single chip. More advanced ALU's began to add other logic functions, shift, and compare capabilities. Half-adders, Full-adders, and ALU's all existed as Logic Chips prior to microprocessors. Microprocessors designers drew on these advances to create their new device.
Control Units are the traffic cops of a microprocessor. The Control Unit implements the microprocessor's instruction set. The Control Unit handles the order of execution of programs and fetches and decodes the instructions to be executed Based on the bit combinations of the instructions the Control Unit moves data around the microprocessor and sends the necessary signals to the ALU to perform the operation needed. After instructions are executed the Control Unit sets various signals, called flags, that indicate the status of the microprocessor and execution of instructions. More advanced Control Units can respond to unplanned events inside and outside of the microprocessor through Interrupts, which cause the Control Unit to invoke programming to deal with these events, such as requests for service or errors. More advanced Control Units use Pipelining to handle the execution of multiple instructions at once, pre-execute instructions, and predict jump sequences.
There are two types of Control Units: Hardwired and Microprogrammed. The difference between them has to do with how the microprocessor's instruction set is implemented, otherwise the function is pretty much the same. In a hardwired microprocessor, the instruction register, is hardwired to rest of the microprocessor. As described above, each bit, in each position, of an instruction causes the Control Unit to do something. They are instructions within an instruction to control how the microprocessor does what it does. This is a very efficient and cost-effective way of implementing a microprocessor, but it has one very major drawback. In the next generation of the microprocessor, if any significant changes are made to increase performance or even reliability. There is a very good chance that the instruction set will need to change because of the very tight coupling of the bits of the instruction and operation of the microprocessor.
To solve this problem, designers decided to put a buffer between the instructions and the Control Unit. Designers decided on a fixed macro-instruction set that would not change from one generation to the next, which programmers would use. The designers would also create a set of micro-instructions that that hardwired and optimized for each generation of microprocessor. The macro-instructions are then mapped to their corresponding micro-instruction(s). This mapping of micro-instructions are stored in a memory accessible by the Control Unit. During program execution, the Control Unit, when presented with these macro-instructions, fetches the corresponding micro-instruction(s) in place of the macro-instruction for execution. This concept allows the investment in programming to be preserved when upgrading to new generations of microprocessors. Famous examples of micropgramming include the IBM System/360 architecture and the Intel x86 architecture.