Have you ever wondered how gaming laptops give you such realistic experiences? How the graphics seem almost real?
Most of us have GPUs (Graphics Processing Units) in our laptops. These GPUs are an example of parallel computing. Graphics as we all know are an array of pixels. Taking a normal processor each of these pixels are processed one after the other. But isn’t this redundant? Is one-pixel dependent on the other? This is what led to the concept of parallelism where you would have a separate processing unit assigned to a group of pixels. As the processing speed increases the quality increases and hence visual effects magnify. In today’s world we see the concept of deep learning being applied to most services that we use in our lives. Here too parallelism helps speed up the process
There can also be another alternative namely the ASICs but GPUs have their own advantages over them like flexibility in upgrading the software and floating-point operations. GPUs also have a fast memory. New direct memory access (DMA) techniques which allows high-volume sensor data to be streamed to the GPU without consuming GPU clock cycles. GPUs are now increasingly found in radar processing. Radar has numerous modes, some of which pilots want to run simultaneously. GPUs are perfect for this application, as they can run multiple processing pipelines at the same time. While FPGA manufacturers offer the ability to synthesize a small number of algorithm “images” on the same chip, the algorithms can’t be run simultaneously. It takes a second or so.
This concept of parallel computing in fact could also be an alternative means to answer the problem faced by Moore’s law. Instead of loading a chip with more transistors, creating multicores and performing the algorithm parallelly solves the issue on how to increase the speed. With progress of time we will see a rise in this concept of computing used in most places, while running algorithms the GPUs will be the forerunners alongside ASICs/FPGAs and CPUs will take a backseat in managing resource sharing.
– Rahul Reji