NVIDIA cuda software and gpu parallel computing architecture

By David B. Kirk. 2007

486 citations recorded by the source corpus.

In the past, graphics processors were special purpose hardwired application accelerators, suitable only for conventional rasterization-style graphics applications. Modern GPUs are now fully programmable, massively parallel floating point processors. This talk will describe NVIDIA's massively multithreaded computing architecture and CUDA software for GPU computing. The architecture is a scalable, highly parallel architecture that delivers high throughput for data-intensive processing. Although not truly general-purpose processors, GPUs can now be used for a wide variety of compute-intensive applications beyond graphic.

Subfields: Computer Vision and Pattern Recognition, Hardware and Architecture, ML Systems & Hardware

How researchers see this paper

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

In Computer Vision and Pattern Recognition

In Hardware and Architecture

In ML Systems & Hardware

Citations in the corpus

Pathways: Asynchronous Distributed Dataflow for ML (2022, methodology): “Given the end of Dennard-scaling, accelerators implement hardware parallelism, often using SIMT (Kirk, 2007) or systolic array (Jouppi et al., 2020) designs.”
ProtTrans: Towards Cracking the Language of Life’s Code Through Self-Supervised Deep Learning and High Performance Computing (2020, background): “More powerful supercomputers [1], [2] and advanced libraries [3], [4], [5], [6], [7] enable the training of more complex models on bigger data sets using advanced processing units (incl.”
Use of SIMD Vector Operations to Accelerate Application Code Performance on Low-Powered ARM and Intel Platforms (2013, background): “These operations are closely related to the Single Instruction Multiple Thread (SIMT) instructions supported on NVIDIA CUDA GPU systems [4].”
AI Benchmark: Running Deep Neural Networks on Android Smartphones (2018, background): “Especially interesting in the context of AI and deep learning are Nvidia Tegra platforms that are supporting CUDA [68] and cuDNN [69] GPU-accelerated libraries of primitives for deep neural networks.”
Optimization principles and application performance evaluation of a multithreaded GPU using CUDA (2008, background): “The architecture allows ef(cid:2)cient data sharing and synchronization among threads in the same thread block [18].”
Program optimization space pruning for a multithreaded gpu (2008, background): “A more complete description can be found in [1, 20].”
ProtTrans: Towards Cracking the Language of Life’s Code Through Self-Supervised Learning (2021, background): “More powerful supercomputers [1], [2] and advanced libraries [3], [4], [5], [6], [7] enable the training of more complex models on bigger data sets using advanced processing units (incl.”
Accelerating Advanced MRI Reconstructions on GPUs (2008, unknown): “More complete descriptions are found in [23, 21, 26].”