It’s already 2022, and I hope there are no people who still don’t know what a CPU or GPU is! If you don’t know, don’t worry, today I will quickly introduce you to various PUs from A to Z, so you can join in the conversation next time during technical exchanges!
- APU (Accelerated Processing Unit)
Processor launched by AMD, which integrates CPU and discrete GPU cores on a single chip, combining scalar operations that CPUs excel at with parallel vector operations traditionally only available to GPUs.
- BPU (Brain Processing Unit)
A specialized processor designed by Horizon Robotics for artificial intelligence, used to support deep neural networks.
- CPU (Central Processing Unit)
The most common processor, mainly consisting of an arithmetic logic unit and a control unit, along with several registers, caches, and the buses for data, control, and status communication between them.
- DPU (Deep-Learning Processing Unit)
A processor designed specifically to optimize deep learning algorithms, typically featuring computing components, on-chip memory, and control logic for managing data communication and computational flow.
- DPU (Data Processing Unit)
This processor offloads workloads that are inefficient for CPUs and unmanageable by GPUs to a dedicated DPU, enhancing system efficiency and reducing total cost of ownership (TCO).
- EPU (Emotion Processing Unit)
An emotion synthesis engine developed by Emoshape that allows machines to process and synthesize emotions.
- FPU (Floating Processing Unit)
A part of the computer system dedicated to floating-point computation.
- GPU (Graphics Processing Unit)
Also known as display chips or visual processors, these are processors specifically designed for image and graphics-related calculations; some models are referred to as mining cards (those who know, know).
- HPU (Holographic Processing Unit)
A co-processor that can work alongside CPU and GPU, used to run various deep learning tasks locally.
- IPU (Intelligence Processing Unit)
A processor launched by Graphcore specifically for AI computation, featuring large-scale parallel processing with complete memory access rights, providing a new technical architecture for AI calculations.
- JPU
Currently, there is no PU starting with J; let’s wait and see.
- KPU (Kernel Processing Unit)
- LPU
No PU starting with L at the moment; if you have an idea, feel free to register one.
- MPU (Micro Processor Unit)
A miniaturized and integrated central processing unit with certain processing and computing capabilities.
- NPU (Neural-Network Processing Unit)
A dedicated embedded neural network chip using ASIC technology, which improves the computational speed of deep learning chips by simulating neural networks.
- OPU (Optical Processing Unit)
The first photon AI accelerator chip used for large-scale non-Von Neumann computing, achieving 1500 TeraOPS.
- PPU (Physical Processing Unit)
A processor dedicated to physical computations in computers, especially applied in video games.
- QPU (Quantum Processing Unit)
A physical or simulated processor that contains many interconnected quantum bits; operating these quantum bits can compute quantum algorithms. It is a core component of quantum computers or quantum simulators.
- RPU (Resistive Processing Unit)
A resistive processing unit that reportedly can increase the training speed of deep neural networks by up to 3000 times; we don’t know much about it, and we dare not ask, as it is still in the research stage.
- SPU (Streaming Processing Unit)
Directly maps multimedia graphic data streams to stream processors for processing, available in both programmable and non-programmable types.
- TPU (Tensor Processing Unit)
A dedicated integrated circuit developed by Google to accelerate deep neural network computations and machine learning.
- UPU (Universe Processing Unit)
- VPU (Video Processing Unit)
A new core engine for video processing platforms, capable of hardware decoding and reducing CPU load.
- WPU (Wearable Processing Unit)
A system-on-chip designed for wearables.
- XPU
The XPU architecture independently developed by Baidu Kunlun is designed for AI services in the cloud and at the edge.
- YPU
Currently, there is none; keep pushing!
- ZPU (Zylin Processing Unit)
A 32-bit open-source processor launched by the Norwegian company Zylin.
E课网 (www.eecourse.com) is a professional integrated circuit education platform under Moore Elite, dedicated to cultivating high-quality integrated circuit professionals in the semiconductor industry. The platform is oriented towards the job demands of integrated circuit companies, providing a practical training platform that aligns with corporate environments. Through online and offline training methods, we quickly train students to meet corporate needs.
E课网 has a mature training platform, a complete curriculum system, and strong teaching resources, planning 168 high-quality semiconductor courses covering the entire integrated circuit industry chain, and has four offline training bases. To date, we have deeply trained a total of 15,367 people and directly supplied 4,476 professionals to the industry. We have established deep cooperative relationships with 143 universities and have held 240 corporate-specific IC training sessions.