GPUs and Multicore CPUs and Parallel and Concurrent Processing
GPUS and Multicore CPUs process information in Parallel to improve speed and efficiency. The GPU does this by using thousands of smaller cores and Pipelining instructions. Whereas the CPU uses multiple cores and and Parallel Systems, Pipelining in a processor to improve efficiency. Watch a quick Craig and Dave Summary here 8. OCR A Level (H046-H446) SLR2–1.1 Multi-core & parallel systems (youtube.com)
Parallel Processing
Parallel processing refers to the method by which a system executes multiple operations simultaneously. CPUs achieve parallelism using multi-core processors where each core executes a different task at the same time.
GPUs take this further with hundreds or thousands of smaller cores, each running separate threads of a larger task, like rendering graphics, making them highly efficient for tasks that can be divided into many similar operations processed at the same time. They are able to Fetch, Decode and Execute thousands of instructions at the same time using pipelining as you can see below:
Concurrent Processing:
Concurrent processing is about dealing with multiple tasks by switching between them quickly, giving the impression that the tasks are being performed simultaneously. This can be managed by a single-core CPU by time-slicing where the CPU rapidly switches between tasks, or by multi-core CPUs where each core handles a different task, allowing different tasks to progress at the same time without necessarily advancing them all in the same instant.
Multi-core and parallel systems
Multi-core CPUs are able to distribute workload across multiple CPU cores, thus achieving significantly higher performance.
The IBM Blue Gene supercomputer has 4,098 processors, allowing 560 Teraflops of processing. Supercomputers are used on problems such as weather forecasting, running climate change models, processing Big Data or sequencing DNA.
Many personal computers and mobile devices are dual-core or quad-core, meaning they have two or four processing chips.
The improvement in performance gained by using a multi-core processor is dependent on the software being able to take advantage of the parallel processing capabilities. Maximizing the usage of the computing resources provided by multi-core processors requires adjustments both to the operating system and to existing application software.
Graphics processing unit (GPU)
A GPU is a specialised electronic circuit which is very efficient at manipulating computer graphics and image-processing. Whereas a CPU has a few cores optimised for sequential serial processing, a GPU has a massively parallel architecture consisting of thousands of smaller, more efficient cores designed for handling multiple tasks simultaneously.
Its highly parallel structure makes it suitable for tasks where the processing of large blocks of visual data is done simultaneously, i.e. in parallel. In a personal computer, a GPU may be present on a graphics card, or embedded on the motherboard. GPUs are now finding generalised uses in computers used for applications such as machine learning, oil exploration, image processing and financial transactions.
Part 1: Serial processing vs Parallel
This activity shows the difference between serial and parallel processing of instructions.
- Make 2 groups.
- One group has a student who writes down a sentence 10 times on their own.
- The OTHER group has a student who is an organiser CPU, who will tell the other students what to do. As a group they are required to write down a statement collectively 10 times.
- Time how long it takes them to write down the same statement 10 times, record these times on the board.
- Discuss the issues with this method, advantages/disadvantages, suitability of the parallel processing group and the serial processing group