The GPU Computing Era

By John R. Nickolls, William (Bill) J. Dally. 2010

1,073 citations recorded by the source corpus.

Subfields: Computer Networks and Communications, Hardware and Architecture, Information Systems

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Networks and Communications

In Hardware and Architecture

In Information Systems

Citations in the corpus

LambdaNetworks: Modeling Long-Range Interactions Without Attention (2021, background): “See Table 4 for an example. usable with additional functionalities such as dilation and striding and enjoys optimized implementations on specialized hardware accelerators (Nickolls & Dally, 2010; Jouppi et al., 2017).”
A fully-synthesizable single-cycle interconnection network for Shared-L1 processor clusters (2011, methodology): “The computing tile in Fermi is the streaming multiprocessor (SM) a tightly coupled processor cluster where 64-Kbyte configurable scratchpad memory/L1 cache is shared among 32 CUDA processor cores [3].”
Benchmarking TPU, GPU, and CPU Platforms for Deep Learning (2019, background): “But GPU is a better choice for small batch sizes, because it executes computation in warps, so it packs small batches and schedules them on stream multiprocessors more easily [39].”
GPUs and the Future of Parallel Computing (2011, methodology): “Nvidia’s current Fermi GPU architecture supports both high-performance graphics and highperformance computing.(3) In its highestperforming computing configuration, Fermi offers a peak throughput of more than 1 teraflop of single-precision floating-point performance (665 Gflops of double-precision) a”
Summarizing CPU and GPU Design Trends with Product Data (2019, background): “The latter takes on compute-intensive tasks for graphics rendering, big-data analytics, signal processing, artiﬁcial intelligence, and physics simulation [7].”
Advances in Hyperspectral Image and Signal Processing: A Comprehensive Overview of the State of the Art (2017, background): “Another high-performance computing architecture that has provided excellent performance when accelerating hyperspectral imaging computations is the graphical processing unit (GPU) [323].”
A Comprehensive Performance Comparison of CUDA and OpenCL (2011, background): “Particularly, today's GPUs (Graphic Processing Units), greatly outperforming CPUs in arithmetic throughput and memory bandwidth, can use hundreds of parallel processor cores to execute tens of thousands of parallel threads [2].”
Towards a generalised GPU/CPU shallow-flow modelling tool (2013, methodology): “A more promising technique involves graphics processing units (GPUs), which are commonplace for gaming systems but were only recently exposed for exploitation in scientific computing Owens et al. (2007); Nickolls and Dally (2010).”
A survey on platforms for big data analytics (2014, background): “However, due to their massively parallel architecture, recent developments in GPU hardware and related programming frameworks have given rise to GPGPU (general-purpose computing on graphics processing units) [31].”
Accelerating CUDA graph algorithms at maximum warp (2011, result): “We also refer interested readers to other studies on the impact of architectural traits on various applications [19, 21], other than graph algorithms.”
Efficient Parallel Graph Exploration on Multi-Core CPU and GPU (2011, background): “The idea of using the graphics processor for general purpose computation has also become popular, since this approach has yielded tremendous performance when applied to suitable problems [1].”
Recent Developments in High Performance Computing for Remote Sensing: A Review (2011, background): “In recent years GPUs have evolved into a highly parallel, multithreaded, many-core processors with tremendous computational speed and very high memory bandwidth [24].”
Improving GPU performance via large warps and two-level warp scheduling (2011, methodology): “After a warp is selected by Our term “GPU core” corresponds to a single Streaming Multiprocessor (SM) in NVIDIA’s terminology [20].”