1 Microprocessor Evolution & Classification v2
1 Microprocessor Evolution & Classification v2
What is a microprocessor?
Definition 1: A microprocessor is a computer processor for which the data processing logic and
control is included on a single integrated circuit, or a small number of ICs. The microprocessor
contains the arithmetic, logic, and control circuitry required to perform the functions of a
computer's central processing unit.
Definition 2: A microprocessor is an electronic component that is used by a computer to do its
work. It is a central processing unit on a single integrated circuit chip containing millions of very
small components including transistors, resistors, and diodes that work together.
Microprocessors can be classified based on the following:
1. Generation
2. Architectures
3. Number of cores
Microprocessor Evolution
1965: Gordon Moore at Fairchild Semiconductor observes in an article for Electronics Magazine
that the number of transistors on a semiconductor chip doubles every year. For
microprocessors, it will double about every two years for more than three decades.
1968: Moore, Robert Noyce and Andy Grove found Intel Corp. to pursue the business of
“INTegrated Electronics”.
1969: Intel announces its first product, the world’s first metal oxide (MOS) static RAM, the 1101.
This signaled the end of magnetic core memory.
1971: Intel launches the world’s first microprocessor, the 4-bit 4004, designed by Federico Faggin.
The 2,000-transistor chip is made for a Japanese calculator, but a farsighted Intel ad calls it
“a micro progammable computer chip”. It has 4KB memory, 45 instructions and based on
PMOS technology.
1972: Intel announces the 8-bit 8008 processor. Teenagers Bill Gates and Paul Allen try to develop
a programming language for the chip, but it is not powerful enough.
1974: Intel introduces 8-bit 8080 processor, with 4,500 transistors and 10 times the performance
of its predecessor.
1975: The 8080 chip finds its first PC application in the Altair 8800, launching the PC revolution.
Gates and Allen succeed in developing the Altair Basic Language, which became Microsoft
Basic, for the 8080.
1976: The x86 architecture suffered a setback when Steve Jobs and Wozniak introduced the Apple
II computer using the 8-bit 6502 processor from MOS Technology. PC maker Commodore
also used the Intel competitor’s chip.
1
1978: Intel introduces the 16-bit 8086 microprocessor. It then became the industry standard.
1979: Intel introduces a lower-cost version of the 8086, the 8088, with 8-bit bus.
1980: Intel introduces the 8087 math co-processor.
1981: IBM picks the Intel 8088 to power is PC. An Intel executive would later call it “the biggest
win ever for Intel”.
1982: IBM signs Advanced Micro Devices as second source to Intel for 8086 and 8088
microprocessors. Intel introduces the 16-bit 80286 processor with 134,000 transistors.
1984: IBM develops the second-generation PC, the 80286-based PC-AT. The PC-AT running MS-
DOS became the de facto PC standard for almost 10 years.
1985: Intel exits the dynamic RAM business to focus on microprocessors, and it brings out the
80386 processor, a 32-bit chip with 275,000 transistors and the ability to run multiple
programs at one.
1986: Compaq Computer leapfrogs IBM with the introduction of an 80386-based PC.
1987: VIA technologies is founded in Fremont, Calif, to sell x86 core logic chip sets.
1989: The 80486 is launched, with 1.2 million transistors and a built-in math co-processor. Intel
predicts the development of multicore processor chips sometime after the year 2000.
Late1980s: The complex instruction set computing (CISC) architecture of the x86 comes under
fire from the rival reduced instruction set computing (RISC) architectures of the Sun Sparc,
the IBM/Apple/Motorolla PowerPC and the MIPS processors. Intel responds with its own
RISC processor, the i860.
1990: Compaq introduces the industry’s first PC servers, running the 80486.
1993: The 3.1 million transistor, 66 MHz Pentium processor with superscalar technology is
introduced.
1994: AMD and Compaq form an alliance to power Compaq computers with Am486
microprocessors.
1995: The Pentium Pro, a RISC slayer, debuts with radical new features that all instructions to be
anticipated and executed out of order. That, plus an extremely fast on-chip cache and dual
independent buses, enable big performance gains in some applications.
1997: Intel launches its 64-bit Epic processor technology. It also introduces the MMX Pentium for
digital signal processor applications, including graphics, audio and voice processing.
1998: Intel introduces the low-end Celeron processor.
1999: VIA acquires Cyrix Corp. and Centaur Technology, makers of x86 processors and x87 co-
processors.
2000: The Pentium 4 debuts with 42 million transistors.
2003: AMD introduces the x86-64, a 64-bit superset of the x86 instruction set.
2
2004: AMD demonstrates an x86 dual-core processor chip.
2005: Intel ships its first dual-core processor chip. Apple announces it will transition its Macintosh
computers from PowerPCs made by Free scale (formerly Motorolla) and to Intel’s x86
family of processors. AMD files antitrust litigation charging that Intel abuses “monopoly”
to exclude and limit competitions.
2006: Dell Inc. announces it will offer AMD processor-based systems.
Generation-based Classification of Microprocessors
1. First Generation – 4-bit Microprocessors (1971)
The Intel corporation came out with the first generation of microprocessors in 1971. They
were 4-bit processors namely Intel 4004. The speed of the processor was 740 kHz taking
60k instructions per second. It had 2300 transistors and 16 pins inside.
Built on a single chip, it was useful for simple arithmetic and logical operations. A control
unit was there to understand the instructions from memory and execute the tasks.
During this period, other microprocessors in the market including Rockwell international
PPS-4, INTEL-8008, and National semiconductors IMP-16 were in use. But, all these were
not TTL compatible processors.
The instructions of these microprocessors were processed serially, they fetched the
instruction, decoded and then executed it. When an instruction of the microprocessor was
finished, then the microprocessor updates the instruction pointer & fetched the following
instruction, performing this consecutive operation for each instruction in turn.
2. Second Generation – 8-bit Microprocessor (1973-1978)
It was useful for arithmetic and logic operations on 8-bit words. The 8008 had a clock speed
of 500kHz and 50k instructions per second.
Followed by an Intel 8080 microprocessor in 1974 with a speed of 2 MHz and 60k instruction
per second. Lastly came the 8085 microprocessor in 1976 having an ability of 769230
instructions per second with 3 MHz speed.
This was the period from 1973 to 1978 in which very efficient 8-bit microprocessors were
implemented like Motorola 6800 and 6801, Intel 8080, Intel-8085, and Zilog’s-Z80, which
were among the most popular ones. Owing to their super-fast speed, they were costly as they
were based on NMOS technology fabrication.
The second generation of the microprocessor is defined by overlapped fetch, decode and
execute the steps. When the first instruction is being processed in the execution unit, the
second instruction is decoded and the third instruction is fetched.
3. Third Generation – 16-bit Microprocessor (1978-1982)
The third generation began with 8086-88 microprocessors in 1978 with 4.77, 8 & 10 MHz
speed and 2.5 million instructions per second.
Other important inventions were Zilog Z800, and Intel 80286, which came out in 1982 and
could read 4 million instructions per second with 68 pins inside.
These processors had a performance like mini computers. During this period 16-bit
processors were created and designed using HMOS technology. Intel 8086/80186/80286 and
3
Motorola 68000 and 68010 with speeds of four times better than the 2nd generation
processors.
4. Fourth Generation – 32-bit Microprocessors (1981-1995)
As many industries converted from commercial microprocessors to in house designs, the
fourth generation microprocessors are entered with outstanding design with a million
transistors.
Intel was still the leader as many companies came out with 32-bit microprocessors around
1986. Their clock speed was between 16 MHz to 33 MHz with 275k transistors inside.
One of the first ones was the Intel 80486 microprocessor of 1986 with 16-100MHz clock
speed and 1.2 Million transistors with 8 KB of cache memory. Followed by the PENTIUM
microprocessor in 1993 which had 66 MHz clock speed and 8-bit of cache memory.
From 1981 to 1995, this generation developed 32-bit microprocessors by using HCMOS
fabrication. INTEL-80386 and Motorola’s 68020/68030 were the popular processors.
Leading edge microprocessors like Intel-80386, Motorola’s 68020/68030, issue & retire
more than one instruction per clock cycle.
5. Fifth Generation – 64-bit Microprocessors (1995-2010)
From 1995 to until now this generation has been bringing out high-performance and high-
speed processors that make use of 64-bit processors. Such processors include Pentium,
Celeron, and Dual and Quad-core processors.
Began in 1995, the Pentium processor was one of the first 64-bit processors with 1.2 GHz to
3 GHz clock speed. There were 291 Million transistors and 64kb instruction per second.
Fifth generation microprocessors employed decoupled super scalar processing, and their
design soon exceeded 10 million transistors. In fifth generation, PCs are a low-margin, high
volume business conquered by a single microprocessor.
Followed by i3, i5, i7 microprocessors in 2007, 2009, 2010 respectively. These were some
of the key points of this generation.
6. Sixth Generation – 64-bit Microprocessors (2015)
The 6th generation Intel® Core™ processor family, formerly Skylake H-Series (Mobile) is
manufactured on Intel’s latest 14 nm technology. Paired with an Intel® CM230 or 100 series
chipset, these processors offer dramatically higher CPU and graphics performance compared to the
previous generation.
They provide a broad range of power options and new advanced features that boost edge-to-cloud
performance for Internet of Things (IoT) designs.
The 6th generation Intel® Core™ processor family maintains a standardized thermal envelope for
45W (cTDP 35W), 35W, and 25W remaining consistent with the previous processor generation. The
new generation is ideal for a wide range of IoT applications, including retail transaction
terminals, digital signage, military and aerospace systems, casino gaming, and industrial automation.
Difference between 5th and 6th generations of microprocessors
The 5th Generation processors focused on Power saving whereas the 6th Generation, called
focused on Gaming & Performance.
4
7. Seventh Generation – 64-bit Microprocessors
Kaby Lake is Intel's codename for its seventh generation Core microprocessor family
announced on August 30, 2016. Like the preceding Skylake, Kaby Lake is produced using a
14 nm manufacturing process technology.
The 7th Intel® Core™ processor family and Intel® Celeron® processor families feature ultra-
low power, 64-bit multi-core processors built on Intel’s optimized 14 nm technology. These
processors offer even greater CPU and graphics performances as compared to the previous
generation.
The U-series of processors offers thermal design power (TDP) options of 15W with the
flexibility of 25W cTDP up and 8.5W cTDP down for exceptional balance of power and
performance.
Designed for small form-factor applications, this multi-chip package (MCP) integrates a
low-power CPU and platform controller hub (PCH) onto a common package substrate.
These processors are ideal for small, energy-efficient designs including digital signage,
point-of-sale terminals and medical tablets.
Some Intel Processors and their features
16MB real
80286 1982 16-bit 68 6-12.5 MHz
4 GB virtual
4GB real
Pentium 1993 32-bit 32-bit address 237 PGA 60-200
64-bit data bus
Pentium Pro 1995 32-bit 64GB real 387 PGA 150-200 MHz
5
36-bit address
bus
733 MHz-1.3
Itanium 2001 64-bit 64 address lines 423 PGA
GHz
RISC Processor
RISC stands for Reduced Instruction Set Computer. It is designed to reduce the execution time
by simplifying the instruction set of the computer. Using RISC processors, each instruction
requires only one clock cycle to execute results in uniform execution time. This reduces the
efficiency as there are more lines of code, hence more RAM is needed to store the instructions.
The compiler also has to work more to convert high-level language instructions into machine code.
Characteristics of RISC
The major characteristics of a RISC processor are as follows −
It consists of simple instructions.
It supports various data-type formats.
It utilizes simple addressing modes and fixed length instructions for pipelining.
One cycle execution time.
“LOAD” and “STORE” instructions are used to access the memory location.
It consists of larger number of registers.
It consists of less number of transistors.
6
CISC Processor
CISC stands for Complex Instruction Set Computer. It is designed to minimize the number of
instructions per program, ignoring the number of cycles per instruction. The emphasis is on
building complex instructions directly into the hardware.
The compiler has to do very little work to translate a high-level language into assembly level
language/machine code because the length of the code is relatively short, so very little RAM is
required to store the instructions.
Characteristics of CISC
Variety of addressing modes.
Larger number of instructions.
Variable length of instruction formats.
Several cycles may be required to execute one instruction.
Instruction-decoding logic is complex.
One instruction is required to support multiple addressing modes.
Special Processors
These are the processors which are designed for some special purposes. Here are some examples
of these processors:
Coprocessors
A coprocessor is a specially designed microprocessor, which can handle its particular function
many times faster than the ordinary microprocessor.
Example − Math Coprocessor.
Some Intel math-coprocessors are −
8087-used with 8086
80287-used with 80286
80387-used with 80386
Input/Output Processor
It is a specially designed microprocessor having a local memory of its own, which is used to control
I/O devices with minimum CPU involvement.
Example −
DMA (direct Memory Access) controller
Keyboard/mouse controller
Graphic display controller
SCSI port controller
Transputer (Transistor Computer)
A transputer is a specially designed microprocessor with its own local memory and having links
to connect one transputer to another transputer for inter-processor communications. It was first
designed in 1980 by Inmos and is targeted to the utilization of VLSI technology.
7
A transputer can be used as a single processor system or can be connected to external links, which
reduces the construction cost and increases the performance.
DSP (Digital Signal Processor)
This processor is specially designed to process the analog signals into a digital form. This is done
by sampling the voltage level at regular time intervals and converting the voltage at that instant
into a digital form. This process is performed by a circuit called an analogue to digital converter,
A to D converter or ADC.
A DSP contains the following components −
Program Memory − It stores the programs that DSP will use to process data.
Data Memory − It stores the information to be processed.
Compute Engine − It performs the mathematical processing, accessing the program from
the program memory and the data from the data memory.
Input/Output − It connects to the outside world.
Its applications are −
Sound and music synthesis
Audio and video compression
Video signal processing
2D and 3d graphics acceleration.
Some other types of processors include microprocessors, embedded processors, and application-
specific integrated circuit (ASIC) processors.
Core-based microprocessor Classification
Processors can be categorized depending on the number of its cores.
A processor that has two independent cores and can process two
Dual-core processor
threads simultaneously.
Has eight cores and can process multiple threads with even greater
Octa-core processor
efficiency.
Each type of these processors serves a specific kind of task and delivers a different level of
performance. For example, an octa-core processor is suitable for creative work like video editing,
whereas a dual processor might be more than enough for everyday computer activities.
8
Examples of Various Types of Processor
Real-world applications:
Consider a smartphone — they typically utilise an SOC (System on Chip) type of processor, often
with multiple cores (Octa-core or Hexa-core). On the other hand, a desktop or laptop usually utilises
a Dual-core or Quad-core processor. Game consoles like PlayStation or Xbox have custom
processors tailored for game performance. In contrast, Intel's Xeon processors, frequently used in
servers, are designed for heavy workloads, multitasking, and data management.
Note: the type of the processor used depends greatly on the device attributes and the tasks to be
performed. Knowledge of these types of processors helps you choose the right device for your
requirements and understand the technology around you better.
How Does Processor Architecture Impact Performance?
Processor architecture significantly affects the performance of a processor, and consequently the
device it's supporting. In fact, the choice between CISC and RISC architectures, for example, can
imply a trade-off between complexity and speed in a device, as RISC's simpler instructions can
usually be executed more quickly. Environment also matters. Devices intended for low power
consumption might benefit from a RISC architecture due to its straightforward execution cycle.
Conversely, a high-end gaming rig or server might demand the versatility and power offered by CISC.
Parallel processing capabilities, inherent in designs like superscalar or vector processors, can
dramatically enhance the processing of large data sets or complex tasks by distributing workload
across multiple processing units.
When it comes to performance metrics, it's interesting to note that just having more cores doesn't
necessarily mean faster speed or better performance. This concept is known as Amdahl's law, a
principle that identifies the maximum improvement gained by improving a particular part of a system.
The impact of Amdahl's Law on multicore processors can be represented by this equation:
1
𝑆𝑝𝑒𝑒𝑑𝑢𝑝 =
𝑃
(1 − 𝑃) +
𝑁
where 'P' represents the proportion of the program that can be parallelized and 'N' depicts the number
of processors.
Real-World Examples of Multicore Processor Use
To underline how invaluable multicore processors are in today's digital world, here are some examples
from everyday life:
Consider a content creator who uses video editing software like Adobe Premiere Pro. This rigorous
software can significantly benefit from a multicore processor, such as an Intel i7 or i9, which can
have six or more cores. Having more cores allows such software to process tasks faster. Conversely,
for general use like web browsing or document editing, a dual-core processor is perfectly sufficient.
For instance, Intel's Core i3 processor, which is a dual-core, is frequently used in entry-level laptops
meant for casual use. Lastly, let's look at video games. Modern gaming frequently depends on multi-
9
threaded operations, where different threads can run on different cores. Hence, quad-core processors
are often the baseline for gaming PCs to ensure smooth gameplay.
Through these examples, it should be clear how indispensable multicore processors have become in
powering various levels of computing needs in the present-day world. The choice of the type of
multicore processor, however, should be guided by the specific demands of the tasks at hand.
The processor performance can be calculated using this equation:
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑖𝑛𝑠𝑡𝑟𝑢𝑐𝑡𝑖𝑜𝑛𝑠
𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 =
𝑅𝑢𝑛𝑡𝑖𝑚𝑒
Practical Examples: How Various Processors Work
Example 1: Apple's new M1 chip is a blend of different types of processors. Based on ARM's RISC
architecture, it is designed for performance with power efficiency. The M1 chip can execute a high
number of instructions per cycle and uses advanced power management to trade off between energy
use and performance.
Example 2: Intel's i7 Processor, a product of CISC architecture, is adept at handling complex
instructions. Each instruction can accomplish several low-level operations like a memory load, an
arithmetic operation and a memory store, offering high performance, especially in single-threaded
processes.
Example 3: NVIDIA's Graphics Processing Unit (GPU), particularly good at concurrently handling
hundreds of threads, is an excellent example of a processor designed for parallel processing. Unlike
conventional CPUs, which are designed for single-threaded performance, GPUs are aimed at handling
real-time graphics rendering, a task that is highly parallelizable, making this processor type ideal for
performing complex algorithms for rendering images.
Types of Processor and Their Characteristics
Central Processing Unit (CPU): These processors are versatile and found universally across
computing systems. CPUs are designed for general-purpose tasks, such as running operating
systems and application software. They are typically characterized by their clock speed,
measured in gigahertz (GHz), and the number of cores they integrate.
Graphics Processing Unit (GPU): Unlike a CPU, a GPU is specifically optimized for rendering
images and video onto your screen. Because they deal with highly parallel tasks, GPUs are
designed with hundreds, if not thousands, of cores, and they excel at executing many light-weight
threads simultaneously.
Digital Signal Processors (DSP): These specialized microprocessors are highly optimized for
executing digital signal processing functions in real time. A DSP is characterized by its ability
to perform certain arithmetic operations very quickly.
Physics Processors: These are specialized hardware designed to handle physics calculations in
games and other simulation software. They offload these compute-intensive tasks from the CPU,
thus improving performance in simulations involving real-world physics.
10
Processor Characteristics and Their Effects on Performance
The performance of these processors is greatly determined by several key characteristics, each having
a direct impact on its efficacy in executing specific tasks. Here are some prominent characteristics
that distinguish processors and their effects on performance:
Number of The more cores a processor has, the more tasks it can process simultaneously,
cores which improves its overall performance
This is the amount of built-in memory a processor has. The cache stores
Cache size frequently used data and instructions closer to the processor to reduce access
times, boosting the speed of data lookup and retrieval.
11
Commonly Used Processors for AI Applications: Their Characteristics
Artificial Intelligence (AI) applications require extraordinary computing power due to the vast
amounts of data they need to process and the complex calculations they need to perform. Thus,
specialized processors and processing technologies have been developed specifically for these tasks.
Tensor Processing Units (TPU): Developed by Google, TPUs are application-specific
integrated circuits (ASICs) used for machine learning workloads. They are specifically
designed to accelerate matrix operations, which make up the bulk of the work in training
machine learning models. TPUs are known for their ability to process data more quickly and
with higher power efficiency than general-purpose processors.
Graphcore's Intelligence Processing Unit (IPU): An IPU is designed specifically to
accelerate AI computations. It has a massive number of cores and its architecture is made to
minimize the distance data has to travel for processing, speeding up model training and
inferencing.
NVIDIA's CUDA Cores: NVIDIA's GPUs, used for deep learning, feature CUDA cores that
are highly parallel processors. For AI applications, this massively parallel architecture enables
the simultaneous execution of thousands of threads, significantly accelerating the training of
machine learning models.
The choice of processor for AI applications should consider several factors: cost, power, cooling
requirements, and the type of workload. As AI processes are usually highly parallelizable tasks,
processors with more cores are generally more beneficial.
Nonetheless, the ideal processor for one AI application might not be the same for another; such trade-
offs should be scrutinized carefully. With rapid progress in AI technologies, the landscape of suitable
processors and their characteristics is continuously evolving.
12