The Central Processing Unit Trivia Quiz

Answer: Silicon

The basis of any computer chip, including the processor, is a semiconductor - made of one of a small family of chemical elements. This material is purified to an extreme degree and then the structures of the chip are created on it. Today, nearly all chips are made of silicon with a small number of specialized devices made of germanium.

However, processors are not among the latter category. The prevalence of silicon as the base material for electronics has led to the name "Silicon Valley" for the southern San Francisco Bay Area where a large number of computer and software companies have their headquarters.

Answer: Fetch a data word from a hard disk

While a computer has many components, the only ones a processor can directly access from machine language are itself and the main memory (RAM). In theory, a processor could be built to directly access other components, but this would mean the processor would then only work with specific makes and devices instead of many different ones available on the market.

A machine language like "return from subroutine" is an example of a complex instruction using implicit memory access. While the instruction itself does not specify any memory read or write, it in fact does read a value from a special memory area that was stored there when the program originally entered the subroutine and then uses that value as a jump target.

Processors also have several "Do nothing" instructions used to control timing and to aid in saving energy while the processor is idle.

Answer: The single processor chip contains two or four independent complete and equal processors

A core in the context of processor design is a complete CPU with all components. Multiple such complete processors can be put on a single chip. These processors share the same memory but are otherwise independent and thus cannot work the same program (the technically more correct term is thread instead of program - threads are independent sub-programs within a larger program). Thus, a machine with a dual core CPU will not be significantly faster while running a single CPU-intensive task like video encoding unless the program is specifically written to take the multiple CPUs into account and break down the work for them.

However, you can give a computer such a CPU-intensive task and still continue using it for a second task like playing on FunTrivia without it getting any slower on that.

Answer: One of a very limited set of internal data storage locations that can be manipulated extremely fast and conveniently

Registers are data storage locations that do not correspond to any main memory but temporarily hold important values within the processor. Early processors often had only 4 to 10 registers freely available to programmers. Current models have a much larger number, often 64 or 128. Data held in registers is the only data that can always freely be read and written within a single clock cycle.

In spite of their crucial role, the number of registers available to a processor has become less of a performance factor than it once was since modern CPUs also include cache memory that can often (but not always) deliver data at the same speed. Registers however do not only include those freely available to the programmer, but also special registers like the program counter, stack pointer and flags registers.

Answer: A cache miss

Cache misses are one of the most frequent performance drains on current computers. Since main memory is much slower than the registers and cache inside the CPU, waiting for an item to be read from there is usually several dozen clock cycles during which the CPU is unproductive, missing an operand or instruction to continue. Cache flushes occur when data is written to a cache but not yet to the main memory and are the process during which that data is saved back to the main RAM. Cache exceptions are just signals from the cache to the CPU logic that something other than a successful transfer has happened upon a request and cache failures are an unplanned reason to buy a new processor.

Answer: Dedicated circuits to make calculations with non-integer numbers

The electronics that make up the mathematical components of a processor are not well suited to deal with non-integer numbers, especially since such numbers can conceivably span a very large range from extremely large to extremely small. To solve this dilemma, processor architects have invented floating point numbers - the binary equivalent of expressing a decimal number in scientific notation (some number between 1 and 10, multiplied by ten to some power). Floating point operations are however much more complex than integer operations, so the earlier processors used a chip called a mathematical co-processor to provide this functionality that, at the time, did not fit the main chip.

While many processors of today do contain a memory controller and early processors did not, this was actually due to the fact that, at that time, memory chips did not need any control hardware. The processor just sent an address and the memory sent back or wrote the data. Loops in programs have always been supported in hardware and no processor as of 2008 includes any internet connectivity functions directly on the chip.

Answer: It stores a potential left-digit overflow resulting from an addition or subtraction

The carry flag is the electronic equivalent of the "carry the ten" when adding 27 to 48. It is extremely vital to ensure operations beyond the data length the processor can calculate at once will yield correct results. Beyond this purely mathematical function, it is also used to control program flow.

The latter use often involves a subtraction of numbers - when the second is larger than the first, you go into negatives and thus create a carry which the processor can then test for and execute different instructions depending on whether the second number was larger than the first or not.

Answer: Pipelining

To execute any instruction, the processor must go through many sequential steps, ranging from reading the instruction from memory (or the cache) over decoding it, fetching the operands, executing the actual calculation to writing back results and various housekeeping functions. To speed up this process, a processor uses a so-called pipeline in which many instructions are processed at once, each being in a different stage of execution.

When a pipelined processor now encounters a jump instruction with a condition (e.g. "jump when the zero flag is set"), it cannot know upon decoding the instruction which one it would need to fetch next. (It could be the jump target or simply the next one in sequence). In order to do any productive work until the jump target is known, the processor now "guesses" at which may be the right path to follow and begins executing instructions at that place. Should the guess prove right, all these instructions are valid and can continue executing, should it however be wrong, the instructions need to be discarded. In the first scenario, a significant performance gain is realized (the time spent waiting for the jump target was used to do valid work), in the second, the performance is no worse than if the processor had done nothing while waiting. Since jump instructions are very common, this situation and possible performance gain occurs very frequently.

It should be clear from this that guessing right often when predicting jump targets is very important in improving processor performance which is why modern processor designs dedicate a significant amount of chip space to making good jump predictions.

Answer: Wait for supply voltages to stabilize

To begin executing any instructions, the processor needs to be in a fully stable and predetermined electrical state. This includes all voltages being stable within their respective allowed ranges and all registers and other internal storage areas including the cache holding a defined value, which however does not need to be zero.

To the CPU, the BIOS of a personal computer is no different from any other software, consisting of machine code instructions to be executed. One of the first things the BIOS does is to determine the amount of memory installed.

Answer: None at all

The VSS pins do not carry a signal as such but provide the zero voltage to which all other voltages in the processor are compared. In the past, this was called GND for Ground and many other connections in the computer still use that designation for the zero voltage, but processors are using the standard terminology for integrated circuits built with field effect transistors: VSS is the zero voltage and VDD is the supply voltage (by convention usually positive). The letters S and D stand for Source and Drain respectively, two of the three contacts on a field effect transistor. (For more information on this type of electronic component, you can look up CMOS and MOSFET in Wikipedia).

Hope you enjoyed this quiz - it's my first, but I intend to bring you more information about the fascinating internals of the computer circuits!