article · Zenodo (CERN European Organization for Nuclear Research) · 2020
The Rise of GPU Computing - Things You Must Know About GPU
dereckshayes
Cited by 0Open access
Abstract
There are loads of speaking happening within the pc graphics industry like 3D games and the best plant identification app. NVIDIA woke up the world to the power of graphics when it invented the GPU long back in October 1999.\n\nBefore that nobody knew what is GPU and how does it work. However, since then there was persistently new standards set in visual computing with breathtaking interactive graphics accessible on gadgets starting from potable media participant to notebooks to workstations.\n\nThe graphics card manufacturer’s expertise in programmable GPUs has led to breakthroughs in parallel processing which makes supercomputing inexpensive and widely accessible. The GPU was conceived primarily for the task of rendering graphics. For the avid teen gamer.\n\nHowever, in the early 2000's, the engineers at NVIDIA decided that the GPU rendering shaders and high throughout ALU/arithmetic units meant that the GPU made for an exceptional parallel computing platform. Before parallel algorithms were implemented using awkward OpenGL/DirectX techniques that were terribly cumbersome for the non-graphics programmer.\n\nThus the Compute Unified architecture was born and paved the way for the CUDA programming language. It’s a language that can be accessible to anyone who knows C and C++ and they could well know how CUDA works.\n\nOne of the main reasons for the move towards parallel processing is that we have reached the point where increasing clock speed on the processor is no longer available. The only substantial way of increasing overall processor speed is to increase the numbers of processors. Even in the CPU world, once the Pentium 4's reached, Intel had no choice but to move to Dual Core and Quad Core architectures.\n\nIn the world of GPU vs CPU computing, there is a number of games currently played by all parties. Intel recently made claims that the GPU is only about 10-14 x faster than their CPU cores, while NVIDIA claimed that performance is really 60x-100x. Unsurprisingly both of these companies employed professional engineers to implement extremely low-level optimizations to swing the odds in their favor.\n\nThe latest about mobile graphics throw light on the latest going on mobile graphics technology. The latest trend in power management is delivered by energy-efficient technology like Switchable Graphics. Imagine combining the low power advantages of an integrated graphics processor with the performances of a discrete graphics processor.\n\nWith Switchable Graphics now you can have the best of both worlds with maximum performance when you need it and long battery life when on the go. Thanks to the booming video game industry and the fierce competition between ATI and NVIDIA, the unending thirst for better graphics cards is likely for the next decade. In all its significant boon for all the computational scientists.
Cite this paper
dereckshayes. (2020). The Rise of GPU Computing - Things You Must Know About GPU. Zenodo (CERN European Organization for Nuclear Research). https://doi.org/10.5281/zenodo.3714828
dereckshayes. “The Rise of GPU Computing - Things You Must Know About GPU.” Zenodo (CERN European Organization for Nuclear Research), 2020, https://doi.org/10.5281/zenodo.3714828.
dereckshayes. 2020. “The Rise of GPU Computing - Things You Must Know About GPU.” Zenodo (CERN European Organization for Nuclear Research). https://doi.org/10.5281/zenodo.3714828.
dereckshayes (2020) ‘The Rise of GPU Computing - Things You Must Know About GPU’, Zenodo (CERN European Organization for Nuclear Research). Available at: https://doi.org/10.5281/zenodo.3714828.
@article{dereckshayes2020rise,
title = {{The Rise of GPU Computing - Things You Must Know About GPU}},
author = {dereckshayes},
journal = {Zenodo (CERN European Organization for Nuclear Research)},
year = {2020},
doi = {10.5281/zenodo.3714828}
}Made from the index record. Check names, capitalisation and page numbers against the paper. Need another style? AI for Research has over 10,000.