MICROPROCESSOR AND INTERFACING

ASSIGNMENT-02

NAME: V. Dharaneswar Reddy

REG.NO: 20BCE2351
SLOT: D1+TD1
TOPIC: Case Study of Multicore Processors
SUBMITTED TO: Rajakumar K

Case Study of Multicore Processors

Multicore processor:
A Multi-processor is an integrated circuit that has two or more
processors linked to it for higher efficiency, consume less power, and
faster simultaneous processing of multiple tasks. It typically has two or
more processors that read and carry out programming instructions.
A unique variety of multiprocessor is the multi-core processor. A
computing machine with two or more separate cores is referred to as a
multicore processor. It can be compared to an integrated circuit that
has two or more separate processors (referred to as cores) attached to
them. The cores may be put into many dies in a single chip package, or
they may be placed onto a single integrated circuit die (a chip
multiprocessor, or CMP). In homogeneous multi-core systems, every
core is the same, however in heterogeneous multi-core systems, they
are not.
Multicore processors are MMID: each core runs a different thread that
accesses a different location of memory (Multiple Data).
All the cores in a multi-core CPU share the same memory. Different
multi-core processor implementations are feasible, depending on the
requirements of the application. It could be implemented using a set of
 homogeneous cores, a set of heterogeneous cores, or a mix of both. In
a homogeneous core architecture, each CPU core is identical and
employs a divide-and-conquer method to increase total processor
performance by dividing applications with high computational needs
into those with lower demands, which are then performed in parallel.
Other significant benefits of using a homogenous multi-core CPU
include reduced design complexity, reusability, reduced testing work,
and hence an easier time to market requirement.
Why multi-core processors are needed?
▪ It is challenging to increase single-core clock frequencies.
▪ Modern software is multithreaded aware.
▪ The current state of computer architecture (shift towards more
parallelism).
Parallel Programming:
Parallel programmes are used when a major work can be broken down
into smaller tasks that can be completed at the same time. The following
things are involved in parallel programming:
• Breaking down a data set into smaller jobs or components.
• Interfering with these smaller jobs, which are then being carried out
concurrently by several processors.

Parallel programming techniques:

The most typical are data parallel and message passing
• Message Passing: To programme using message passing, libraries
that control the data transfer between processors must be linked
with and called. In essence, the user calls liberty to actively transfer
data amongst processors. the MPI, or message passing interface.
• Data Parallel: Writing a programme and compiling it using a data
parallel compiler is data parallel programming. To disseminate the
data to all processes, the compiler translates the programme into
standard code and makes calls to a message-passing library.
Architecture of multicore processor
The architecture of a multi-core processor provides communication
between each variable core, and it divides and distributes all processing
responsibilities effectively. Two or more cores and a number of caches
make up every multi-core processor. Once all processing activities have
been completed, the processed data from each gateway. In terms of
overall performance, this technique outperforms a single core CPU.
Advantages:
1. Performance
2. Reliability
3. Software Interactions
4. Multitasking
5. Power consumption
6. Obsolescence Avoidance
7. Isolation
Disadvantages:
1. Application speed
2. Jitter
3. Analysis
4. Resource sharing
5. Software Interference
Multi-core processor is used for:
1. Virtualization
2. Databases
3. Analytics and HPC
4. Cloud
5. Visualization
Homogeneous vs Heterogeneous Multicore processors:
A multicore processor's cores could be homogenous or heterogeneous.
For x86 computer architectures, mainstream multicore processors from
Intel and AMD are homogeneous and offer identical cores. However, it
is frequently inefficient to use a complicated equipment for a simple
task or to achieve maximum efficiency. The market for heterogeneous
multicore processors employs computers with various cores for various
tasks. In embedded or arm processors that may combine
microprocessor and microcontroller cores in the same device,
heterogeneous cores are typically encountered.
There are three general goals for heterogeneous multicore processors:
1. Optimized performance
2. Optimized power
3. Optimized security
Examples of multicore processors are
intel i9,i7,i5 or AMD zen 3 family, zen 2 family, zen+ family.

Evolution of Multi-core processor:

Performance and cost have always been taken into consideration while
designing microprocessors. In order to fulfil this constantly increasing
demand, Gordon Moore, the founder of Intel Corporation, predicted
that the number of transistors on a chip would double once every 18
months. This prediction is known as Moore's Law in the semiconductor
industry. One billion transistors have been integrated on a chip to
increase integration density, which has been made possible by cutting-
edge chip fabrication and integrated circuit processing technology.
Pollack's rule, which specifies that performance gains are roughly equal
to the square root of complexity increases, regulates microarchitecture.
Therefore, a CPU core's performance would only grow by 40% when
its logic was doubled. Modern chip fabrication methods have major
limitations, one of which is power dissipation. According to studies,
transistor leakage current increases as chips get smaller and smaller,
significantly increasing static power dissipation. Increasing the
frequency of operation, which allows for quicker programme execution,
is another alternative strategy for enhancing performance.
The frequency is currently limited to 4GHz, though, as any increase
over this range results in an increase in power dissipation. Limits on
system cost and battery life force the design team to consider power
over performance in such a scenario. For typical air-cooled
microprocessor server boxes, power consumption has gotten so high
that purchasing refrigeration or liquid cooling equipment may be
 necessary. Designers were finally met with the power wall, or maximum
amount of power a CPU could use.
Major Challenges faced by multi-core processors:
Although multi-core CPUs provide many advantages, the technology still
has some major challenges to overcome. One of the biggest issues is that
computer programs run slower on multi-core processors than single-core
processors. As has been correctly stated, multi-core systems do not
automatically become faster as the number of cores grows. Programmers
must write software that makes use of the increasing number of
processors in a multi-core machine without increasing the amount of time
needed to develop a software. Many apps currently in use were designed
to work on a single processor, failing to make use of the potential of multi-
core systems.
Similarly, heterogeneous cores use specialised, application-specific
processor cores to solve the issue of running a variety of apps
simultaneously on a computer. Examples include a remedial core that
handles less computationally challenging applications, a complex core that
handles computationally intensive applications, and a DSP core that
handles multimedia applications requiring intensive mathematical
computations. As an alternative, multi-core CPUs could be constructed
with a mixture of homogeneous and heterogeneous cores to increase
performance while utilising both techniques. The IBM CELL multi-core
processor, which has eight comparably sized and power-efficient
accelerators for a single general-purpose CPU and has demonstrated
performance effectiveness, employs this method.
The single-core implementations' power and frequency restrictions
opened the door for multi-core technology, which will remain popular in
the market. However, achieving the highest performance throughput
requires resolving all the issues that multi-core processors are now facing.
Numerous technological advancements are planned in this area of
technology, including a new multi-core programming language and tools
for converting current software to "multi-core aware" software. Despite
being one of the most difficult technologies to adopt, there is a lot of
research being done to maximise the potential of multi-core computers.