Multi-core Processor Architecture

 Multi-core Processor Architecture

• What are Muti-cores- Multiple processor on one IC having two or more separate processing units. The cores fit on one processor socket where cores run in parallel

Why Multi-core-

* Multi-threeding

* Deeply pipelined circuits which helps lowers heat, improve power

* Parallel programming techniques

 Multi-core solution-

* Creating two or more cores on an equivalent Die will increase processing power as keeping clock speeds at an appropriate level.

* By splitting the processing task between two or more cores on an equivalent chip, designer can lower the clock speed.

* As systems become powerful, more cores are often integrated at same efficiency.

 Implementation-

* There are two main ways to possess multiple cores interact

* 1. The shared memory model

* 2. Distributed memory model

* within the shared memory model, all cores share an equivalent cache memory.

* within the distributed memory model, each core has its own cache memory.









• The Intel core duo design features a separate L1 cache memory for every core, but both cores share an L2 cache.











• The AMD Athlon 64 X2 implementation has separate L1 and L2 cache memory for every core.



 








Functions-

* Hyper threading

* Turbo boost technology

* Improved cache latency

* Intelligent power technology

* Better virtualisation

* Quick Path Interconnect

* Greater parallelism

* Efficient algorithms

 Advantages-

* Lowers heating

* Better power consumption

* Process larger data

* Shorter connections

* Occupies less space

* While working with many threads, a Multi Core processor with n cores can execute n threads simultaneously by assigning a core to every thread. If it must process quite n threads , say x, it can apply multithreading procedures with each core working with a mean of x/n threads.

* one core processor must multithread with every single thread.

Other Incentives-

* Improving an existing single core design or creating a far better one is problematic therein it

* Is time consuming to style

* Is risky to switch existing designs creating multi core processors is convenient therein

* The name “core dual” and similar names are good for marketing.

* it’s lower manufacturing costs.

* Uses proven processor designs.

Limitations-

* Expensive

* Faster battery drainage

* Processors feature shared caches which further split in cores

* doesn’t speed up w.r.t increase in cores

* Reduce production yiel

Problems- 

* A number of the present problems found with multi core processors include:

* Memory/Cache coherence. As I wrote some of the implementations have distributed L1 caches but there must share an L2 cache. This poses the matter of creating sure each core keeps the opposite updated with changes within the data in its own cache.

* Multi threading is additionally a drag when the software being run isn’t designed to require advantage of the multi core processor. This might mean that one core does most of the work which suggests that the processor is running no more efficiently than one core.

Applications-

* Machine vision

* CAD systems

* CNC machines

* Automated test systems

* Motion control

Future scope

* Current debates is on orgue over if future multi core processors should be heterogenous or homogeneous.

* That is, should all the cores be an equivalent or should there be a mixture of various types?

* Having all cores be an equivalent makes production easier and keeps its complexity to a minimum.

* Having different cores that are specialized in specific tasks increases complexity but has the potential to be far more efficient in speed and power consumption.

Current trends in market-

* Intel is functioning on 16 multi-core chip projects

* IBM introduced dual-core Power 5 which runs fourfold faster than its predessor

* Sun is manufacturing chips which are faster by increasing speed of every core

* a really general demand in Mark is to supply chips with more cores and fewer space(cellphone processor required more functions and fewer space)

* Snapdragon is functioning on making chip which doubles transistor per annum

Authors- Ayush Patidar Bhagwat Borole Nikita Borole


Comments