Author: Jikesong, Deputy Director of Heterogeneous Computing R&D at Tencent CSIG 〇. Background of This Article About two years ago, I discussed the current state and issues of GPU virtualization with many colleagues on Tencent’s internal network. Since then, some new research directions have emerged, and some industry changes may completely overturn some of the … Read more
In the field of modern high-performance computing and parallel processing, two concepts frequently arise:SIMD (Single Instruction Multiple Data) and SIMT (Single Instruction Multiple Threads). For those who are encountering these terms for the first time, they may find their meanings and differences confusing. This article will help you understand the basic concepts, application scenarios, and … Read more
In modern parallel computing architectures, GPUs (Graphics Processing Units) are no longer just “graphics chips” used for rendering, but rather powerful “multi-core processors” with exceptional general-purpose parallel computing capabilities. Among them, NVIDIA’s CUDA programming framework employs the SIMT (Single Instruction Multiple Threads) model allows us to easily write parallel programs with a “thread” mindset. However, … Read more
According to Lei Feng Network, this article is based on Tan Xu’s answer to the question “How to Understand Nvidia’s Multi-GPU Communication Framework NCCL?” on Zhihu, and Lei Feng Network has obtained authorization for reprint. Question Details: In deep learning, multi-GPU parallel training is often required, and Nvidia’s NCCL library NVIDIA/nccl (https://github.com/NVIDIA/nccl) is frequently used … Read more
Source: Xingneng Assets Editor | Eva In 1987, Tsugio Makimoto, the former chief engineer of Hitachi, proposed that semiconductor products might advance along a path of alternating “standardization” and “customization,” oscillating every ten years. He published this idea in Electronics Weekly in 1991, referring to it as “Makimoto’s Wave.” In recent years, the rapid development … Read more
This time, the Chinese AI industry is truly different. Recently, there has been a piece of news in the tech circle that, although seemingly ordinary, is incredibly exciting upon reflection—DeepSeek has released version 3.1, introducing the UE8M0 FP8 precision format, specifically optimized for domestic AI chips. Following the announcement, A-share chip stocks surged across the … Read more
By Ian Ferguson, Vice President of Business Development at SiFive When I joined SiFive in early 2024, my initial focus (which later expanded) was to explore how to drive broader adoption of SiFive’s RISC-V technology in data center system-on-chip components. Introducing new technologies and replacing long-standing dominant vendors is not a task that can be … Read more
Recently, there have been reports in the market that Huawei’s AI chip is shifting from a dedicated ASIC architecture to a general-purpose GPGPU architecture in order to challenge NVIDIA’s market position. It is said that the new architecture is compatible with the CUDA ecosystem, which can break through existing software bottlenecks and expand more application … Read more
The development resources provided by NVIDIA mostly belong to the library or API level, including CUDA, CUDNN, CuFFT, CuBLAS, TensorRT, etc. Developers with a solid foundation in C++/Python programming languages are required to leverage the parallel computing advantages of GPU/CUDA, which somewhat limits the popularity of applications related to parallel computing. DeepStream is a suite … Read more
First of all, I would like to thank everyone for their support of “Discussion on CUDA Shared Virtualization Support“!Here, I would like to add adisadvantage of the driver forwarding scheme.In the division of CUDA computing power and video memory, there is a significant approach that involves creating a fake libcuda library. This fake libcuda library … Read more