Which specifications are important when comparing GPUs

CPU and GPU: make optimal use of both

Today's systems are required much more than ever before - whether for deep learning applications, massively parallel data processing, intensive 3D gaming or other demanding applications. CPU (Central Processing Unit, "main processor") and GPU (Graphics Processing Unit, "graphics processor") have very different tasks. What are CPUs used for? What are GPUs used for? When buying a new computer and comparing its technical characteristics, it is important to know what role each of these building blocks plays.

What is a CPU?
The CPU, made up of millions or billions of transistors, can have several processor cores ("computing cores") and is also commonly referred to as the computer's brain. It is the basic building block of all modern computer systems because it executes the program instructions and the processes required by the computer and the operating system. In addition, the CPU largely determines how quickly programs can be executed, from surfing the Internet to spreadsheets.

What does the abbreviation GPU stand for?
The GPU is a processor that consists of many smaller cores designed for more specialized tasks. Because of their interaction, the cores offer enormous performance when data processing processes can be distributed across many cores.

What is the difference between a CPU and a GPU?
CPUs and GPUs have a lot in common. Both are critically important computer elements. Both are microprocessors based on semiconductor technology. And both process data. However, CPUs and GPUs have different architectures and are used for different purposes.

The CPU lends itself to a wide variety of tasks, especially those where latency or per-core performance matters. As a powerful module for executing program commands, the CPU uses the smaller number of cores for individual tasks and their fast completion. This makes the CPU particularly suitable for jobs ranging from serial data processing to the operation of databases.

GPUs were initially specialized ASICs designed to speed up certain 3D rendering processes. Over time, these graphics engines with originally defined functions have become more programmable and flexible. While the graphics display and the increasingly lifelike reproduction of today's top computer games is still the main function of the GPUs, they have meanwhile also developed into multi-purpose parallel processors for a growing range of applications.

What is integrated graphics?
With integrated graphics or graphics with shared memory, the graphics unit is on the same chip as the CPU. Certain CPUs can have an integrated GPU instead of relying on a separate graphics controller. The GPUs, sometimes referred to as IGP (integrated graphics processor), share the main memory with the CPU.

Integrated graphics processors offer several advantages. Their integration into the CPU proves to be advantageous compared to independent graphics processors in terms of space requirements, costs and energy efficiency. They provide sufficient power to process graphics-related data and commands for common tasks such as surfing the web, streaming 4K movies, and playing casual games.

This concept is most commonly used in devices where compactness and energy efficiency are important, such as laptops, tablets, smartphones and some desktop PCs.

Accelerating deep learning and AI
Today, GPUs are being used for more and more tasks such as deep learning and artificial intelligence (AI). GPUs or other accelerators are ideal for deep learning training with multi-layer neural networks or with very large amounts of connected data, such as 2D images.

Deep learning algorithms were adapted for accelerated processing with a GPU, whereby a considerable increase in performance was achieved and training for various problems from practice was possible for the first time with a feasible and practicable effort.

In the meantime, CPUs and the software libraries created for them have evolved into much more powerful tools for deep learning tasks. For example, CPU-based systems have improved deep learning performance through extensive software optimizations and additional special AI hardware, such as Intel® Deep Learning Boost (Intel® DL Boost) in the latest scalable Intel® Xeon® processors .

The strengths of CPUs come into their own in many applications, such as high-definition, 3D and non-image-based deep learning with voice, text and time series data. CPUs can support much larger memory than even today's best GPUs for complex models or deep learning applications (e.g. 2D image recognition).

The combination of CPU and GPU as well as sufficiently large RAM offers excellent test options for deep learning and AI.

Decades of leadership position in CPU development
Intel has a long history of CPU innovation that began in 1971 with the introduction of the 4004 as the first commercial microprocessor to be fully integrated on a semiconductor chip.

Today, with Intel® CPUs based on the familiar x86 architecture, the desired AI solution can be implemented where it makes sense. From the high-performance, scalable Intel® Xeon® processor series in the data center and in the cloud to energy-efficient Intel® Core ™ processors at the edge of the network, Intel offers CPUs for every need.

Intelligent performance characteristics of the 11th generation of Intel® Core ™ processors
The 11th generation Intel® Core ™ processors are manufactured using Intel's optimized process technology, are based on a newly designed core architecture and a completely new graphics architecture, and have integrated AI instructions to enable intelligently optimized performance and experiences.

Systems with Intel® Core ™ processors of the 11th generation are equipped with the latest, integrated Intel® Iris® Xe Graphic equipped. Certain form factor units such as ultra-slim laptops will also contain the first discrete graphics processing unit (GPU) built on the Intel Xe Architecture based. The Intel® Iris® Xe MAX dedicated graphics ensures a big leap forward in the field of flat and light notebooks. You also get more power and new features for improved content creation and gaming experience.

The Intel® Iris® Xe Graphics are powered by Intel® Deep Learning Boost AI to improve content creation, photo and video editing, and have a low-power architecture for longer battery life so you can design and multitask.

Separate GPUs from Intel
Intel offers two options for separate GPUs, both based on the Intel Xe Architecture based.

Intel® Iris® Xe MAX Grafik is the first separate graphics processor for thin and light laptops that is based on the Intel Xe Architecture based. Optimized for coupling with 11th generation Intel® Core ™ processors, you get even more performance and new functions for an improved experience when creating content and gaming.

The Intel® Server GPU is a separate graphics processor for data centers that is based on the new Intel Xe Architecture based. The Intel® Server GPU is designed to scale exponentially, bringing Android gaming, media transcoding / encoding, and over-the-top (OTT) video streaming experiences to the next level.

Today it is no longer a matter of weighing CPU and GPU against each other. More than ever, both are needed together to meet the diverse requirements of data processing. Without a doubt, the best results are achieved when the right tool is used for each task.